We've moved way beyond the old days of building discovery, recommendation, decision support, and other AI tools using traditional ML and pattern recognition techniques. The future of universal personal assistance for discovery and learning is upon us. How will multimodality image, video, and audio understanding, and reasoning abilities of large foundation models change how we build these systems? I will shed some initial light on this topic by discussing 3 trends: First, the move to a single multimodal large model with reasoning abilities; Second, the fundamental research on personalization and user alignment; Third, the combination of System 1 and System 2 cognitive abilities into a single universal assistant.

Large Language Models (LLMs) pose a new paradigm of modeling and computation for information tasks. Recommendation systems are a critical application domain poised to benefit significantly from the sequence modeling capabilities and world knowledge inherent in these large models. In this paper, we introduce PLUM, a framework designed to adapt pre-trained LLMs for industry-scale recommendation tasks. PLUM consists of item tokenization using Semantic IDs, continued pre-training (CPT) on domain-specific data, and task-specific fine-tuning for recommendation objectives. For fine-tuning, we focus particularly on generative retrieval, where the model is directly trained to generate Semantic IDs of recommended items based on user context. We conduct comprehensive experiments on large-scale internal video recommendation datasets. Our results demonstrate that PLUM achieves substantial improvements for retrieval compared to a heavily-optimized production model built with large embedding tables. We also present a scaling study for the model's retrieval performance, our learnings about CPT, a few enhancements to Semantic IDs, along with an overview of the training and inference methods that enable launching this framework to billions of users in YouTube.

Generative recommendation (GR) is an emerging paradigm where user actions are tokenized into discrete token patterns and autoregressively generated as predictions. However, existing GR models tokenize each action independently, assigning the same fixed tokens to identical actions across all sequences without considering contextual relationships. This lack of context-awareness can lead to suboptimal performance, as the same action may hold different meanings depending on its surrounding context. To address this issue, we propose ActionPiece to explicitly incorporate context when tokenizing action sequences. In ActionPiece, each action is represented as a set of item features. Given the action sequence corpora, we construct the vocabulary by merging feature patterns as new tokens, based on their co-occurrence frequency both within individual sets and across adjacent sets. Considering the unordered nature of feature sets, we further introduce set permutation regularization, which produces multiple segmentations of action sequences with the same semantics. Our code is available at: https://github.com/google-deepmind/action_piece.

Despite their success in many domains, large language models (LLMs) remain under-studied in scenarios requiring optimal decision-making under uncertainty. This is crucial as many real-world applications, ranging from personalized recommendations to healthcare interventions, demand that LLMs not only predict but also actively learn to make optimal decisions through exploration. In this work, we measure LLMs’ (in)ability to make optimal decisions in bandits, a state-less reinforcement learning setting relevant to many applications. We develop a comprehensive suite of environments, including both context-free and contextual bandits with varying task difficulties, to benchmark LLMs’ performance. Motivated by the existence of optimal exploration algorithms, we propose efficient ways to integrate this algorithmic knowledge into LLMs: by providing explicit algorithm-guided support during inference; and through algorithm distillation via in-context demonstrations and fine-tuning, using synthetic data generated from these algorithms. Impressively, these techniques allow us to achieve superior exploration performance with smaller models, surpassing larger models on various tasks. We conducted an extensive ablation study to shed light on various factors, such as task difficulty and data representation, that influence the efficiency of LLM exploration. Additionally, we conduct a rigorous analysis of the LLM’s exploration efficiency using the concept of regret, linking its ability to explore to the model size and underlying algorithm.

It has become increasingly clear that recommender systems that overly focus on short-term engagement prevents users from exploring diverse interests, ultimately hurting long-term user experience. To tackle this challenge, numerous diversification algorithms have been proposed as the final stage of recommender systems. These algorithms typically rely on measures of item similarity, aiming to maximize the dissimilarity across items in the final set of recommendations. However, in this work, we demonstrate the benefits of going beyond item-level similarities by utilizing higher-level user understanding-specifically, user intents that persist across multiple interactions or recommendation sessions-in diversification. Our approach is motivated by the observation that user behaviors on online platforms are largely driven by their underlying intents. Therefore, final recommendations should ensure that a diverse set of user intents is accurately represented. While user intent has primarily been studied in the context of search, it is less clear how to incorporate real-time dynamic intent predictions in recommender systems.

Large Language Models (LLMs) empower recommendation systems through their advanced reasoning and planning capabilities. However, the dynamic nature of user interests and content poses a significant challenge: While initial fine-tuning aligns LLMs with domain knowledge and user preferences, it fails to capture such real-time changes, necessitating robust update mechanisms. This paper investigates strategies for updating LLM-powered recommenders, focusing on the trade-offs between ongoing fine-tuning and Retrieval-Augmented Generation (RAG). Using an LLM-powered user interest exploration system as a case study, we perform a comparative analysis of these methods across dimensions like cost, agility, and knowledge incorporation. We propose a hybrid update strategy that leverages the long-term knowledge adaptation of periodic fine-tuning with the agility of low-cost RAG. We demonstrate through live A/B experiments on a billion-user platform that this hybrid approach yields statistically significant improvements in user satisfaction, offering a practical and cost-effective framework for maintaining high-quality LLM-powered recommender systems.

Chain-of-thought (CoT) prompting for language models demonstrates impressive performance across reasoning tasks, but typically needs labeled exemplars of the reasoning process. In this work, we introduce a new prompting approach, analogical prompting, designed to automatically guide the reasoning process of large language models. Inspired by analogical reasoning, a cognitive process in which humans draw from relevant past experiences to tackle new problems, our approach prompts language models to self-generate relevant exemplars or knowledge in the context, before proceeding to solve the given problem. This method presents several advantages: it obviates the need for labeling or retrieving exemplars, offering generality and convenience; it can also tailor the generated exemplars and knowledge to each problem, offering adaptability. Experimental results show that our approach outperforms 0-shot CoT and manual few-shot CoT in a variety of reasoning tasks, including math problem solving in GSM8K and MATH, code generation in Codeforces, and other reasoning tasks in BIG-Bench.

We present STEP-BACK PROMPTING, a simple prompting technique that enables LLMs to do abstractions to derive high-level concepts and first principles from instances containing specific details. Using the concepts and principles to guide reasoning, LLMs significantly improve their abilities in following a correct reasoning path towards the solution. We conduct experiments of STEP-BACK PROMPTING with PaLM-2L, GPT-4 and Llama2-70B models, and observe substantial performance gains on various challenging reasoning-intensive tasks including STEM, Knowledge QA, and Multi-Hop Reasoning. For instance, STEP-BACK PROMPTING improves PaLM-2L performance on MMLU (Physics and Chemistry) by 7% and 11% respectively, TimeQA by 27%, and MuSiQue by 7%.

Fine-tuning is becoming widely used for leveraging the power of pre-trained foundation models in new downstream tasks. While there are many successes of fine-tuning on various tasks, recent studies have observed challenges in the generalization of fine-tuned models to unseen distributions (i.e., out-of-distribution; OOD). To improve OOD generalization, some previous studies identify the limitations of fine-tuning data and regulate fine-tuning to preserve the general representation learned from pre-training data. However, potential limitations in the pre-training data and models are often ignored. In this paper, we contend that overly relying on the pre-trained representation may hinder fine-tuning from learning essential representations for downstream tasks and thus hurt its OOD generalization. It can be especially catastrophic when new tasks are from different (sub)domains compared to pre-training data. To address the issues in both pre-training and fine-tuning data, we propose a novel generalizable fine-tuning method LEVI (Layer-wise Ensemble of different VIews), where the pre-trained model is adaptively ensembled layer-wise with a small task-specific model, while preserving its efficiencies. By combining two complementing models, LEVI effectively suppresses problematic features in both the fine-tuning data and pre-trained model and preserves useful features for new tasks. Broad experiments with large language and vision models show that LEVI greatly improves fine-tuning generalization via emphasizing different views from fine-tuning data and pre-trained features.

Exposure bias and its induced feedback loop effect are well-known problems in recommender systems. Exploration is believed to be the key to break such feedback loops. While classical contextual bandit algorithms such as Upper-Confidence-Bound and Thompson Sampling have been successful in addressing the exploration-exploitation trade-off in the single-task settings with one clear reward signal, modern recommender systems often leverage multiple rich sources of feedback such as clicks, likes, dislikes, shares, satisfaction survey responses, and employ multi-task learning in practice. It is unclear how one can incorporate exploration in the multi-task setup with different objectives. In this paper, we study an efficient bandit algorithm tailored to multi-task recommender systems, named Multi-task Neural Linear Bandit (mtNLB). In particular, we investigate efficient feature embeddings in the multi-task setups that could be used as contextual features in the Neural Linear Bandit, a contextual bandit algorithm that nicely combines the representation power from DNN and simplicity in uncertainty calculation from linear models. We further study cost-effective approximations of the uncertainty estimate and principled ways to incorporate uncertainty into the multi-task scoring of items. To showcase the efficacy of our proposed method, we conduct live experiments on a large-scale commercial recommendation platform that serves billions of users. We evaluate the quality of the uncertainty estimate and demonstrate its ability to improve exploration across the different dimensions of the reward signals in comparison to baseline approaches.

We introduce SELF-DISCOVER, a general framework for LLMs to self-discover the task-intrinsic reasoning structures to tackle complex reasoning problems that are challenging for typical prompting methods. Core to the framework is a self-discovery process where LLMs select multiple atomic reasoning modules such as critical thinking and step-by-step thinking, and compose them into an explicit reasoning structure for LLMs to follow during decoding. SELF-DISCOVER substantially improves GPT-4 and PaLM 2’s performance on challenging reasoning benchmarks such as BigBench-Hard, grounded agent reasoning, and MATH, by as much as 32% compared to Chain of Thought (CoT). Furthermore, SELF-DISCOVER outperforms inference-intensive methods such as CoT-Self-Consistency by more than 20%, while requiring 10-40x fewer inference compute. Finally, we show that the self-discovered reasoning structures are universally applicable across model families: from PaLM 2-L to GPT-4, and from GPT-4 to Llama2, and share commonalities with human reasoning patterns.

Industrial recommendation systems process billions of daily user feedback which are complex and noisy. Efficiently uncovering user preference from these signals becomes crucial for high-quality recommendation. We argue that those signals are not inherently equal in terms of their informative value and training ability, which is particularly salient in industrial applications with multi-stage processes (e.g., augmentation, retrieval, ranking). Considering that, in this work, we propose a novel self-auxiliary distillation framework that prioritizes training on high-quality labels, and improves the resolution of low-quality labels through distillation by adding a bilateral branch-based auxiliary task. This approach enables flexible learning from diverse labels without additional computational costs, making it highly scalable and effective for Google-scale recommenders. Our framework consistently improved both offline and online key business metrics across three Google major products. Notably, self-auxiliary distillation proves to be highly effective in addressing the severe signal loss challenge posed by changes such as Apple iOS policy. It further delivered significant improvements in both offline (+17% AUC) and online metrics for a Google Apps recommendation system. This highlights the opportunities of addressing real-world signal loss problems through self-auxiliary distillation techniques.

Knowledge Distillation (KD) is a powerful approach for compressing a large model into a smaller, more efficient model, particularly beneficial for latency-sensitive applications like recommender systems. However, current KD research predominantly focuses on Computer Vision (CV) and NLP tasks, overlooking unique data characteristics and challenges inherent to recommender systems. This paper addresses these overlooked challenges, specifically: (1) mitigating data distribution shifts between teacher and student models, (2) efficiently identifying optimal teacher configurations within time and budgetary constraints, and (3) enabling computationally efficient and rapid sharing of teacher labels to support multiple students. We present a robust KD system developed and rigorously evaluated on multiple large-scale personalized video recommendation systems within Google. Our live experiment results demonstrate significant improvements in student model performance while ensuring consistent and reliable generation of high-quality teacher labels from a continuous data stream of data.

Sequential models are invaluable for powering personalized recommendation systems. In the context of short-form video (SFV) feeds, where user behavior history is typically longer, systems must be able to understand users’ long-term interests. However, deploying large sequence models to extensive web-scale applications faces challenges due to high serving cost. To address this, we propose an industrial framework designed for efficiently serving large user sequence models. Specifically, the proposed infrastructure decouples serving of the user sequence model and the main recommendation model, with the user sequence model being served offline (asynchronously) with periodical refresh. The proposed infrastructure is also model-agnostic; thus, it can be used to support any type of user sequence models (even LLMs) with controllable costs. Empirical results show that large user models deployed with our framework significantly and consistently enhance the quality of the main recommendation model with minimal serving costs increase.

In recent years, massive transformer-based architectures have driven breakthrough performance in practical applications like autoregressive text-generation (LLMs) and click-prediction (recommenders). A common recipe for success is to train large models on massive web-scale datasets [3, 15], e.g., modern recommenders are trained on billions of user-item click events, and LLMs are trained on trillions of tokens extracted from the public internet. We are close to hitting the computational and economical limits of scaling up the size of these models, and we expect the next frontier of gains to come from improving the: (i) data quality of the training dataset, and (ii) data efficiency of the extremely expensive training procedure. Inspired by this shift, we present a set of “data-centric” techniques for recommendation and language models that summarizes a dataset into a terse data summary, which is both (i) high-quality, i.e., trains better quality models, and (ii) improves the data-efficiency of the overall training procedure. We propose techniques from two disparate data frameworks: (i) data selection (a.k.a., coreset construction) methods that sample portions of the dataset using grounded heuristics, and (ii) data distillation techniques that generate synthetic examples which are optimized to retain the signals needed for training high-quality models. Overall, this work sheds light on the challenges and opportunities offered by data optimization in web-scale systems, a particularly relevant focus as the recommendation community grapples with the grand challenge of leveraging LLMs.

Randomly-hashed item ids are used ubiquitously in recommendation models. However, the learned representations from random hashing prevents generalization across similar items, causing problems of learning unseen and long-tail items, especially when item corpus is large, power-law distributed, and evolving dynamically. In this paper, we propose using content-derived features as a replacement for random ids. We show that simply replacing ID features with content-based embeddings can cause a drop in quality due to reduced memorization capability. To strike a good balance of memorization and generalization, we propose to use Semantic IDs [15], a compact and discrete item representation, as a replacement for random item ids. Semantic IDs are learned from frozen content embeddings using RQ-VAE and thus can capture the hierarchy of concepts in items. Similar to content embeddings, the compactness of Semantic IDs poses a problem of adaption in recommendation models. We propose novel methods for adapting Semantic IDs in industry-scale ranking models, through hashing sub-pieces of of the Semantic-ID sequences. In particular, we find that the SentencePiece model [10] that is commonly used in LLM tokenization outperforms manually crafted pieces such as N-grams. To the end, we evaluate our approaches in a real-world ranking model for YouTube recommendations. Our experiments demonstrate that Semantic IDs can replace the direct use of video IDs by improving the generalization ability on new and long-tail item slices without sacrificing overall model quality.

Traditional recommendation systems are subject to a strong feedback loop by learning from and reinforcing past user-item interactions, which in turn limits the discovery of novel user interests. To address this, we introduce a hybrid hierarchical framework combining Large Language Models (LLMs) and classic recommendation models for user interest exploration. The framework controls the interfacing between the LLMs and the classic recommendation models through “interest clusters”, the granularity of which can be explicitly determined by algorithm designers. It recommends the next novel interests by first representing “interest clusters” using language, and employs a fine-tuned LLM to generate novel interest descriptions that are strictly within these predefined clusters. At the low level, it grounds these generated interests to an item-level policy by restricting classic recommendation models, in this case a transformer-based sequence recommender to return items that fall within the novel clusters generated at the high level. We showcase the efficacy of this approach on an industrial-scale commercial platform serving billions of users. Live experiments show a significant increase in both exploration of novel interests and overall user enjoyment of the platform.

Multi-task prediction models and value models are the de-facto standard ranking components in modern large-scale content recommendation systems. However, they are typically optimized to model users’ passive consumption behaviors, and rank content in a way to grow only consumption-centric values. In this talk, we discuss the key insight that it is possible to model sparse participatory content-generation actions as well and grow ecosystem value through a new ranking system. We made the following key technical contributions in this system: (1) introducing ranking for content generation based on a categorization of user participation actions of different sparsity, including proxy intent action or access point clicks. (2) improving sparse task prediction quality and stability by causal task relationship modeling, conditional loss modeling and ResNet based shared bottom network. (3) personalizing the value model to minimize conflicts between different values, through e.g. ranking inspiring content higher for users who actively generate content. (4) conducting systematic evaluation of proposed approach in a large short-form video UGC (User-Generated Content) platform.

Effective exploration is believed to positively influence the long-term user experience on recommendation platforms. Determining its exact benefits, however, has been challenging. Regular A/B tests on exploration often measure neutral or even negative engagement metrics while failing to capture its long-term benefits. We here introduce new experiment designs to formally quantify the long-term value of exploration by examining its effects on content corpus, and connecting content corpus growth to the long-term user experience from real-world experiments. Once established the values of exploration, we investigate the Neural Linear Bandit algorithm as a general framework to introduce exploration into any deep learning based ranking systems. We conduct live experiments on one of the largest short-form video recommendation platforms that serves billions of users to validate the new experiment designs, quantify the long-term values of exploration, and to verify the effectiveness of the adopted neural linear bandit algorithm for exploration.

Finetuning language models on a collection of datasets phrased as instructions has been shown to improve model performance and generalization to unseen tasks. In this paper we explore instruction finetuning with a particular focus on (1) scaling the number of tasks, (2) scaling the model size, and (3) finetuning on chain-of-thought data. We find that instruction finetuning with the above aspects dramatically improves performance on a variety of model classes (PaLM, T5, U-PaLM), prompting setups (zero-shot, few-shot, CoT), and evaluation benchmarks (MMLU, BBH, TyDiQA, MGSM, open-ended generation, RealToxicityPrompts). For instance, Flan-PaLM 540B instruction-finetuned on 1.8K tasks outperforms PaLM 540B by a large margin (+9.4% on average). Flan-PaLM 540B achieves state-of-the-art performance on several benchmarks (at time of release), such as 75.2% on five-shot MMLU. We also publicly release Flan-T5 checkpoints,1 which achieve strong few-shot performance even compared to much larger models, such as PaLM 62B. Overall, instruction finetuning is a general method for improving the performance and usability of pretrained language models.

In recent years, social media users spend significant amount of time on Short-Form Video (SFV) platforms. Its success in creating an immersive viewership experience is not only from the content, but also due to its unique UI innovation: instead of providing choices for users to click, SFV platforms actively recommend content to users to watch one at a time. In this paper, we highlight unique challenges rooted from such UI changes for SFV recommendation system design. Firstly, there is yet much unexplored for sources of system biases under the new UI, as there are no clicks nor the common click-based position biases. Additionally, when training multiple types of user activities, positive labels for activities like sharing and commenting can be much sparser and more skewed than traditional click-based recommendation systems, as the latter can filter non-click impressions when generating "post-click" activities.

Large language models have achieved impressive performance on various natural language processing tasks. However, so far they have been evaluated primarily on benchmarks where all information in the input context is relevant for solving the task. In this work, we investigate the *distractibility* of large language models, i.e., how the model prediction can be distracted by irrelevant context. In particular, we introduce Grade-School Math with Irrelevant Context (GSM-IC), an arithmetic reasoning dataset with irrelevant information in the problem description. We use this benchmark to measure the distractibility of different prompting techniques for large language models, and find that the model is easily distracted by irrelevant information. We also identify several approaches for mitigating this deficiency, such as decoding with self-consistency and adding to the prompt an instruction that tells the language model to ignore the irrelevant information.

Industry recommender systems usually suffer from highly-skewed long-tail item distributions where a small fraction of the items receives most of the user feedback. This skew hurts recommender quality especially for the item slices without much user feedback. While there have been many research advances made in academia, deploying these methods in production is very difficult and very few improvements have been made in industry. One challenge is that these methods often hurt overall performance; additionally, they could be complex and expensive to train and serve.

Deep learning techniques have advanced rapidly in recent years, leading to significant progress in pre-trained and fine-tuned large-scale AI models. For example, in the natural language processing domain, the traditional "pre-train, fine-tune" paradigm is shifting towards the "pre-train, prompt, and predict" paradigm, which has achieved great success on many tasks across different application domains such as ChatGPT/BARD for Conversational AI and P5 for a unified recommendation system. Moreover, there has been a growing interest in models that combine vision and language modalities (vision-language models) which are applied to tasks like Visual Captioning/Generation. Considering the recent technological revolution, it is essential to emphasize these paradigm shifts and highlight the paradigms with the potential to solve different tasks. We thus provide a platform for academic and industrial researchers to showcase their latest work, share research ideas, discuss various challenges, and identify areas where further research is needed in pre-training, fine-tuning, and prompt-learning methods for large-scale AI models. We foster the development of a strong research community focused on solving challenges related to large-scale AI models, providing superior and impactful strategies that can change people's lives in the future.

Recommender systems play an important role in many content platforms. While most recommendation research is dedicated to designing better models to improve user experience, we found that research on stabilizing the training for such models is severely under-explored. As recommendation models become larger and more sophisticated, they are more susceptible to training instability issues, i.e., loss divergence, which can make the model unusable, waste significant resources and block model developments. In this paper, we share our findings and best practices we learned for improving the training stability of a real-world multitask ranking model for YouTube recommendations. We show some properties of the model that lead to unstable training and conjecture on the causes. Furthermore, based on our observations of training dynamics near the point of training instability, we hypothesize why existing solutions would fail, and propose a new algorithm to mitigate the limitations of existing solutions. Our experiments on YouTube production dataset show the proposed algorithm can significantly improve training stability while not compromising convergence, comparing with several commonly used baseline methods.

Recommendation system serves as a conduit connecting users to an incredibly large, diverse and ever growing collection of contents. In practice, missing information on fresh (and tail) contents needs to be filled in order for them to be exposed and discovered by their audience. We here share our success stories in building a dedicated fresh content recommendation stack on a large commercial platform. To nominate fresh contents, we built a multi-funnel nomination system that combines (i) a two-tower model with strong generalization power for coverage, and (ii) a sequence model with near real-time update on user feedback for relevance. The multi-funnel setup effectively balances between coverage and relevance. An in-depth study uncovers the relationship between user activity level and their proximity toward fresh contents, which further motivates a contextual multi-funnel setup. Nominated fresh candidates are then scored and ranked by systems considering prediction uncertainty to further bootstrap content with less exposure. We evaluate the benefits of the dedicated fresh content recommendation stack, and the multi-funnel nomination system in particular, through user corpus co-diverted live experiments. We conduct multiple rounds of live experiments on a commercial platform serving billion of users demonstrating efficacy of our proposed methods.

Learning high-quality feature embeddings efficiently and effectively is critical for the performance of web-scale machine learning systems. A typical model ingests hundreds of features with vocabularies on the order of millions to billions of tokens. The standard approach is to represent each feature value as a $d$-dimensional embedding, which introduces hundreds of billions of parameters for extremely high-cardinality features. This bottleneck has led to substantial progress in alternative embedding algorithms. Many of these methods, however, make the assumption that each feature uses an independent embedding table. This work introduces a simple yet highly effective framework, Feature Multiplexing, where one single representation space is used for many different categorical features. Our theoretical and empirical analysis reveals that multiplexed embeddings can be decomposed into components from each constituent feature, allowing models to distinguish between features. We show that multiplexed representations give Pareto-optimal space-accuracy tradeoffs for three public benchmark datasets. Further, we propose a highly practical approach called Unified Embedding with three major benefits: simplified feature configuration, strong adaptation to dynamic data distributions, and compatibility with modern hardware. Unified embedding gives significant improvements in offline and online metrics compared to highly competitive baselines across five web-scale search, ads, and recommender systems, where it serves billions of users across the world in industry-leading products.

Modern recommender systems perform large-scale retrieval by embedding queries and item candidates in the same unified space, followed by approximate nearest neighbor search to select top candidates given a query embedding. In this paper, we propose a novel generative retrieval approach, where the retrieval model autoregressively decodes the identifiers of the target candidates. To that end, we create semantically meaningful tuple of codewords to serve as a Semantic ID for each item. Given Semantic IDs for items in a user session, a Transformer-based sequence-to-sequence model is trained to predict the Semantic ID of the next item that the user will interact with. We show that recommender systems trained with the proposed paradigm significantly outperform the current SOTA models on various datasets. In addition, we show that incorporating Semantic IDs into the sequence-to-sequence model enhances its ability to generalize, as evidenced by the improved retrieval performance observed for items with no prior interaction history.

"Garbage in, Garbage out" is a familiar maxim to ML practitioners and researchers, because the quality of a learned data representation is highly crucial to the quality of any ML model that consumes it as an input. To handle systems that serve billions of users at millions of queries per second (QPS), we need representation learning algorithms with significantly improved efficiency. At Google, we have dedicated thousands of iterations to develop a set of powerful techniques that efficiently learn high quality data representations. We have thoroughly validated these methods through offline evaluation, online A/B testing, and deployed these in over 50 models across major Google products. In this paper, we consider a generalized data representation learning problem that allows us to identify feature embeddings and crosses as common challenges. We propose two solutions, including: 1. Multi-size Unified Embedding to learn high-quality embeddings; and 2. Deep Cross Network V2 for learning effective feature crosses. We discuss the practical challenges we encountered and solutions we developed during deployment to production systems, compare with SOTA methods, and report offline and online experimental results. This work sheds light on the challenges and opportunities for developing next-gen algorithms for web-scale systems.

The last decade has witnessed many successes of deep learning-based models for industry-scale recommender systems. These models are typically trained offline in a batch manner. While being effective in capturing users’ past interactions with recommendation platforms, batch learning suffers from long model-update latency and is vulnerable to system biases, making it hard to adapt to distribution shift and explore new items or user interests. Although online learning-based approaches (e.g., multi-armed bandits) have demonstrated promising theoretical results in tackling these challenges, their practical real-time implementation in large-scale recommender systems remains limited. First, the scalability of online approaches in servicing a massive online traffic while ensuring timely updates of bandit parameters poses a significant challenge. Additionally, exploring uncertainty in recommender systems can easily result in unfavorable user experience, highlighting the need for devising intricate strategies that effectively balance the trade-off between exploitation and exploration. In this paper, we introduce Online Matching: a scalable closed-loop bandit system learning from users’ direct feedback on items in real time. We present a hybrid offline + online approach for constructing this system, accompanied by a comprehensive exposition of the end-to-end system architecture. We propose Diag-LinUCB – a novel extension of the LinUCB algorithm – to enable distributed updates of bandits parameter in a scalable and timely manner. We conduct live experiments in YouTube and show that Online Matching is able to enhance the capabilities of fresh content discovery and item exploration in the present platform.

Learning expressive representations for high-dimensional yet sparse features has been a longstanding problem in information retrieval. Though recent deep learning methods can partially solve the problem, they often fail to handle the numerous sparse features, particularly those tail feature values with infrequent occurrences in the training data. Worse still, existing methods cannot explicitly leverage the correlations among different instances to help further improve the representation learning on sparse features since such relational prior knowledge is not provided. To address these challenges, in this paper, we tackle the problem of representation learning on feature-sparse data from a graph learning perspective. Specifically, we propose to model the sparse features of different instances using hypergraphs where each node represents a data instance and each hyperedge denotes a distinct feature value. By passing messages on the constructed hypergraphs based on our Hypergraph Transformer (HyperFormer), the learned feature representations capture not only the correlations among different instances but also the correlations among features. Our experiments demonstrate that the proposed approach can effectively improve feature representation learning on sparse features.

Prompt-Tuning is a new paradigm for finetuning pre-trained language models in a parameter efficient way. Here, we explore the use of HyperNetworks to generate hyper-prompts: we propose HyperPrompt, a novel architecture for prompt-based task-conditioning of self-attention in Transformers. The hyper-prompts are end-to-end learnable via generation by a HyperNetwork. HyperPrompt allows the network to learn task-specific feature maps where the hyper-prompts serve as task global memories for the queries to attend to, at the same time enabling flexible information sharing among tasks. We show that HyperPrompt is competitive against strong multi-task learning baselines with as few as 0.14% of additional task-conditioning parameters, achieving great parameter and computational efficiency. Through extensive empirical experiments, we demonstrate that HyperPrompt can achieve superior performances over strong T5 multi-task learning baselines and parameter-efficient adapter variants including Prompt-Tuning and HyperFormer++ on Natural Language Understanding benchmarks of GLUE and SuperGLUE across many model sizes.

Over the years we have seen recommender systems shifting focus from optimizing short-term engagement toward improving long-term user experience on the platforms. While defining good long-term user experience is still an active research area, we focus on one specific aspect of improved long-term user experience here, which is user revisiting the platform. These long term outcomes however are much harder to optimize due to the sparsity in observing these events and low signal-to-noise ratio (weak connection) between these long-term outcomes and a single recommendation. To address these challenges, we propose to establish the association between these long-term outcomes and a set of more immediate term user behavior signals that can serve as surrogates for optimization.

Multi-Task Learning (MTL) is a powerful learning paradigm to improve generalization performance via knowledge sharing. However, existing studies find that MTL could sometimes hurt generalization, especially when two tasks are less correlated. One possible reason that hurts generalization is spurious correlation, i.e., some knowledge is spurious and not causally related to task labels, but the model could mistakenly utilize them and thus fail when such correlation changes. In MTL setup, there exist several unique challenges of spurious correlation. First, the risk of having non-causal knowledge is higher, as the shared MTL model needs to encode all knowledge from different tasks, and causal knowledge for one task could be potentially spurious to the other. Second, the confounder between task labels brings in a different type of spurious correlation to MTL. Given such label-label confounders, we theoretically and empirically show that MTL is prone to taking non-causal knowledge from other tasks. To solve this problem, we propose Multi-Task Causal Representation Learning (MT-CRL) framework. MT-CRL aims to represent multi-task knowledge via disentangled neural modules, and learn which module is causally related to each task via MTL-specific invariant regularization. Experiments show that MT-CRL could enhance MTL model's performance by 5.5% on average over Multi-MNIST, MovieLens, Taskonomy, CityScape, and NYUv2, and show it could indeed alleviate spurious correlation problem.

We explore how generating a chain of thought---a series of intermediate reasoning steps---significantly improves the ability of large language models to perform complex reasoning. In particular, we show how such reasoning abilities emerge naturally in sufficiently large language models via a simple method called chain of thought prompting, where a few chain of thought demonstrations are provided as exemplars in prompting. Experiments on three large language models show that chain of thought prompting improves performance on a range of arithmetic, commonsense, and symbolic reasoning tasks. The empirical gains can be striking. For instance, prompting a 540B-parameter language model with just eight chain of thought exemplars achieves state of the art accuracy on the GSM8K benchmark of math word problems, surpassing even finetuned GPT-3 with a verifier.

Industrial recommendation platforms are increasingly concerned with how to make recommendations that cause users to enjoy their long term experience on the platform. Reinforcement learning emerged naturally as an appealing approach for its promise in 1) combating feedback loop effect resulted from myopic system behaviors; and 2) sequential planning to optimize long term outcome. Scaling RL algorithms to production recommender systems serving billions of users and contents, however remain challenging. Sample inefficiency and instability of online RL hinder its widespread adoption in production. Offline RL enables usage of off-policy data and batch learning. It on the other hand faces significant challenges in learning due to the distribution shift.

Personalized item search has become an essential tool for online platforms---where users interact with a large corpus of items (e.g., click, purchase, like) via a search query---to provide their users with a more satisfactory search experience. The record (or history) of users' past interactions serves as a valuable asset to achieve personalization. While user history data can span over a long period of time, only a part of the history is relevant to a user's current search intent. Moreover, since historical interactions take place at aperiodic points in time, modeling their relevance to the current search query entangles complex temporal dependencies. We propose multi-resolution attention to address these challenges for personalized item search. Our approach captures higher-order temporal relations between user queries and their history across several temporal subspaces (i.e., resolutions), each corresponding to distinct temporal ranges with adaptive time boundaries that are also learned directly from data. We achieve this by coupling the conventional multi-head attention module with a differentiable soft-thresholding mechanism, which essentially operates as a masking function in the temporal domain. Comparisons with strong baselines on an open-source benchmark dataset confirm the efficacy of our approach.

Users who come to recommendation platforms are heterogeneous in activity levels. There usually exists a group of core users who visit the platform regularly and consume a large body of content upon each visit, while others are casual users who tend to visit the platform occasionally and consume less each time. As a result, consumption activities from core users often dominate the training data used for learning. As core users can exhibit different activity patterns from casual users, recommender systems trained on historical user activity data usually achieve much worse performance on casual users than core users. To bridge the gap, we propose a model-agnostic framework L2Aug to improve recommendations for casual users through data augmentation, without sacrificing core user experience. L2Aug is powered by a data augmentor that learns to generate augmented interaction sequences, in order to fine-tune and optimize the performance of the recommendation system for casual users. On four real-world public datasets, L2Aug outperforms other treatment methods and achieves the best sequential recommendation performance for both casual and core users. We also test L2Aug in an online simulation environment with real-time feedback to further validate its efficacy, and showcase its flexibility in supporting different augmentation actions.

Multi-task learning aims to solve multiple machine learning tasks at the same time, with good solutions being both generalizable and Pareto optimal. A multi-task deep learning model consists of a shared representation learned to capture task commonalities, and task-specific sub-networks capturing the specificities of each task. In this work, we offer insights on the under-explored trade-off between minimizing task training conflicts in multi-task learning and improving multi-task generalization, i.e. the generalization capability of the shared presentation across all tasks. The trade-off can be viewed as the tension between multi-objective optimization and shared representation learning: As a multi-objective optimization problem, sufficient parameterization is needed for mitigating task conflicts in a constrained solution space; However, from a representation learning perspective, over-parameterizing the task-specific sub-networks may give the model too many ”degrees of freedom” and impedes the generalizability of the shared representation.

Modern recommender systems have evolved rapidly along with deep learning models that are well-optimized for overall performance, especially those trained under Empirical Risk Minimization (ERM). However, a recommendation algorithm that focuses solely on the average performance may reinforce the exposure bias and exacerbate the “rich-get-richer” effect, leading to unfair user experience. In a simulation study, we demonstrate that such performance gap among various user groups is enlarged by an ERM-trained recommender in the long-term. To mitigate such amplification effects, we propose to optimize for the worst-case performance under the Distributionally Robust Optimization (DRO) framework, with the goal of improving long-term fairness for disadvantaged subgroups. In addition, we propose a simple-yet-effective streaming optimization improvement called Streaming-DRO (S-DRO), which effectively reduces loss variances for recommendation problems with sparse and long-tailed data distributions. Our results on two large-scale datasets suggest that (1) DRO is a flexible and effective technique for improving worst-case performance, and (2) Streaming-DRO outperforms vanilla DRO and other strong baselines by improving the worst-case and overall performance at the same time.

In this work we study the problem of measuring the fairness of a machine learning model under noisy information. Focusing on group fairness metrics, we investigate the particular but common situation when the evaluation requires controlling for the confounding effect of covariate variables. In a practical setting, we might not be able to jointly observe the covariate and group information, and a standard workaround is to then use proxies for one or more of these variables. Prior works have demonstrated the challenges with using a proxy for sensitive attributes, and strong independence assumptions are needed to provide guarantees on the accuracy of the noisy estimates. In contrast, in this work we study using a proxy for the covariate variable and present a theoretical analysis that aims to characterize weaker conditions under which accurate fairness evaluation is possible. Furthermore, our theory identifies potential sources of errors and decouples them into two interpretable parts y and E. The first part y depends solely on the performance of the proxy such as precision and recall, whereas the second part E captures correlations between all the variables of interest. We show that in many scenarios the error in the estimates is dominated by y via a linear dependence, whereas the dependence on the correlations E only constitutes a lower order term. As a result we expand the understanding of scenarios where measuring model fairness via proxies can be an effective approach. Finally, we compare, via simulations, the theoretical upper-bounds to the distribution of simulated estimation errors and show that assuming some structure on the data, even weak, is key to significantly improve both theoretical guarantees and empirical results.

We consider the problem of listwise learning-to-rank (LTR) on data with \textit{partitioned preference}, where a set of items are sliced into ordered and disjoint partitions, but the ranking of items within a partition is unknown. The Plackett-Luce (PL) model has been widely used in listwise LTR methods. However, given $N$ items with $M$ partitions, calculating the likelihood of data with partitioned preference under the PL model has a time complexity of $O(N+S!)$, where $S$ is the maximum size of the top $M-1$ partitions. This computational challenge restrains existing PL-based listwise LTR methods to only a special case of partitioned preference, \textit{top-$K$ ranking}, where the exact order of the top $K$ items is known. In this paper, we exploit a random utility model formulation of the PL model and propose an efficient approach through numerical integration for calculating the likelihood. This numerical approach reduces the aforementioned time complexity to $O(N+MS)$, which allows training deep-neural-network-based ranking models with a large output space. We demonstrate that the proposed method outperforms well-known LTR baselines and remains scalable through both simulation experiments and applications to real-world eXtreme Multi-Label (XML) classification tasks. The proposed method also achieves state-of-the-art performance on XML datasets with relatively large numbers of labels per sample.

Large scale recommender models find most relevant items from huge catalogs, and they play a critical role in modern search and recommendation systems. To model the input space with large-vocab categorical features, a typical recommender model learns a joint embedding space through neural networks for both queries and items from user feedback data. However, with millions to billions of items in the corpus, users tend to provide feedback for a very small set of them, causing a power-law distribution. This makes the feedback data for long-tail items extremely sparse.

Developing robust NLP models that perform well on many, even small, slices of data is a significant but important challenge, with implications from fairness to general reliability.To this end, recent research has explored how models rely on spurious correlations, and how counterfactual data augmentation (CDA) can mitigate such issues.In this paper we study how and why modeling counterfactuals over multiple attributes can go significantly further in improving model performance.We propose RDI, a context-aware methodology which takes into account the impact of secondary attributes on the model's predictions and increases sensitivity for secondary attributes over reweighted counterfactually augmented data.By implementing RDI in the context of toxicity detection, we find that accounting for secondary attributes can significantly improve robustness, with improvements in sliced accuracy on the original dataset up to 7% compared to existing robustness methods.We also demonstrate that RDI generalizes to the coreference resolution task and provide guidelines to extend this to other tasks.

Embedding learning of categorical features (e.g. user/item IDs) is at the core of various recommendation models. The standard approach creates an embedding table where each row represents a dedicated embedding vector for every unique feature value. However, this method fails to efficiently handle high-cardinality features and unseen feature values (e.g. new video ID) that are prevalent in real-world recommendation systems. In this paper, we propose an alternative embedding framework Deep Hash Embedding (DHE), replacing embedding tables by a deep embedding network to compute embeddings on the fly. DHE first encodes the feature value to a unique identifier vector with multiple hashing functions and transformations, and then applies a DNN to convert the identifier vector to an embedding. The encoding module is deterministic, non-learnable, and free of storage, while the embedding network is updated during the training time to learn embedding generation. Empirical results show that DHE achieves comparable AUC against the standard one-hot full embedding, with smaller model sizes. Our work sheds light on the design of DNN-based alternative embedding schemes for categorical features without using embedding table lookup.

As multi-task models gain popularity in a wider range of machine learning applications, it is becoming increasingly important for practitioners to understand the fairness implications associated with those models. Most existing fairness literature focuses on learning a single task more fairly, while how ML fairness interacts with multiple tasks in the joint learning setting is largely under-explored. In this paper, we are concerned with how group fairness (e.g., equal opportunity, equalized odds) as an ML fairness concept plays out in the multi-task scenario. In multi-task learning, several tasks are learned jointly to exploit task correlations for a more efficient inductive transfer. This presents a multi-dimensional Pareto frontier on (1) the trade-off between group fairness and accuracy with respect to each task, as well as (2) the trade-offs across multiple tasks. We aim to provide a deeper understanding on how group fairness interacts with accuracy in multi-task learning, and we show that traditional approaches that mainly focus on optimizing the Pareto frontier of multi-task accuracy might not perform well on fairness goals. We propose a new set of metrics to better capture the multi-dimensional Pareto frontier of fairness-accuracy trade-offs uniquely presented in a multi-task learning setting. We further propose a Multi-Task-Aware Fairness (MTA-F) approach to improve fairness in multi-task learning. Experiments on several real-world datasets demonstrate the effectiveness of our proposed approach.

The Mixture-of-Experts (MoE) architecture is showing promising results in improving parameter sharing in multi-task learning (MTL) and in scaling high-capacity neural networks. State-of-the-art MoE models use a trainable "sparse gate'" to select a subset of the experts for each input example. While conceptually appealing, existing sparse gates, such as Top-k, are not smooth. The lack of smoothness can lead to convergence and statistical performance issues when training with gradient-based methods. In this paper, we develop DSelect-k: a continuously differentiable and sparse gate for MoE, based on a novel binary encoding formulation. The gate can be trained using first-order methods, such as stochastic gradient descent, and offers explicit control over the number of experts to select. We demonstrate the effectiveness of DSelect-k on both synthetic and real MTL datasets with up to 128 tasks. Our experiments indicate that DSelect-k can achieve statistically significant improvements in prediction and expert selection over popular MoE gates. Notably, on a real-world, large-scale recommender system, DSelect-k achieves over 22% improvement in predictive performance compared to Top-k. We provide an open-source implementation of DSelect-k.

Neural networks lack adversarial robustness, i.e., they are vulnerable to adversarial examples that through small perturbations to inputs cause incorrect predictions. Further, trust is undermined when models give miscalibrated predictions, i.e., the predicted probability is not a good indicator of how much we should trust our model. In this paper, we study the connection between adversarial robustness and calibration and find that the inputs for which the model is sensitive to small perturbations (are easily attacked) are more likely to have poorly calibrated predictions. Based on this insight, we examine if calibration can be improved by addressing those adversarially unrobust inputs. To this end, we propose Adversarial Robustness based Adaptive Label Smoothing (AR-AdaLS) that integrates the correlations of adversarial robustness and calibration into training by adaptively softening labels for an example based on how easily it can be attacked by an adversary. We find that our method, taking the adversarial robustness of the in-distribution data into consideration, leads to better calibration over the model even under distributional shifts. In addition, AR-AdaLS can also be applied to an ensemble model to further improve model calibration.

Reinforcement Learning (RL) has been sought after to bring next-generation recommender systems to further improve user experience on recommendation platforms. While the exploration-exploitation tradeoff is the foundation of RL research, the value of exploration in (RL-based) recommender systems is less well understood. Exploration, commonly seen as a tool to reduce model uncertainty in regions of sparse user interaction/feedback, is believed to cost user experience in the short term, while the indirect benefit of better model quality arrives at a later time. We focus on another aspect of exploration, which we refer to as user exploration to help discover new user interests, and argue it can improve user experience even in the more imminent term.

How can we build recommender systems to take into account fairness? Real-world recommender systems are often composed of multiple models, built by multiple teams. However, most research on fairness focuses on improving fairness in a single model. Further, recent research on classification fairness has shown that combining multiple "fair" classifiers can still result in an "unfair" classification system. This presents a significant challenge: how do we understand and improve fairness in recommender systems composed of multiple components?

Reinforcement Learning (RL) techniques have been sought after as the next-generation tools to further advance the field of recommendation research. Different from classic applications of RL, recommender agents, especially those deployed on commercial recommendation platforms, have to operate in extremely large state and action spaces, serving a dynamic user base in the order of billions, and a long-tail item corpus in the order of millions or billions. The (positive) user feedback available to train such agents is extremely scarce in retrospect. Improving the sample efficiency of RL algorithms is thus of paramount importance when developing RL agents for recommender systems. In this work, we present a general framework to augment the training of model-free RL agents with auxiliary tasks for improved sample efficiency. More specifically, we opt to add additional tasks that predict users' immediate responses (positive or negative) toward recommendations, i.e., user response modeling, to enhance the learning of the state and action representations for the recommender agents. We also introduce a tool based on gradient correlation analysis to guide the model design. We showcase the efficacy of our method in offline experiments, learning and evaluating agent policies over hundreds of millions of user trajectories. We also conduct live experiments on an industrial recommendation platform serving billions of users and tens of millions of items to verify its benefit.

Despite deep neural network (DNN)'s impressive prediction performance in various domains, it is well known now that a set of DNN models trained with the same model specification and the exact same training data could produce very different prediction results. People have relied on the state-of-the-art ensemble method to estimate prediction uncertainty. However, ensembles are expensive to train and serve for web-scale traffic systems.

Learning effective feature crosses is the key behind building recommender systems. However, the sparse and large feature space requires exhaustive search to identify effective crosses. Deep & Cross Network (DCN) was proposed to automatically and efficiently learn bounded-degree predictive feature interactions. Unfortunately, in models that serve web-scale traffic with billions of training examples, DCN showed limited expressiveness in its cross network at learning more predictive feature interactions. Despite significant research progress made, many deep learning models in production still rely on traditional feed-forward neural networks to learn feature crosses inefficiently.

Most existing recommender systems focus primarily on matching users (content consumers) to content which maximizes user satisfaction on the platform. It is increasingly obvious, however, that content providers have a critical influence on user satisfaction through content creation, largely determining the content pool available for recommendation. A natural question thus arises: can we design recommenders taking into account the long-term utility of both users and content providers? By doing so, we hope to sustain more content providers and a more diverse content pool for long-term user satisfaction. Understanding the full impact of recommendations on both user and content provider groups is challenging. This paper aims to serve as a research investigation of one approach toward building a content provider aware recommender, and evaluating its impact in a simulated setup.

Highly skewed long-tail item distribution is very common in recommendation systems. It significantly hurts model performance on tail items. To improve tail-item recommendation, we conduct research to transfer knowledge from head items to tail items, leveraging the rich user feedback in head items and the semantic connections between head and tail items. Specifically, we propose a novel dual transfer learning framework that jointly learns the knowledge transfer from both model-level and item-level: 1. The model-level knowledge transfer builds a generic meta-mapping of model parameters from few-shot to many-shot model. It captures the implicit data augmentation on the model-level to improve the representation learning of tail items. 2. The item-level transfer connects head and tail items through item-level features, to ensure a smooth transfer of meta-mapping from head items to tail items. The two types of transfers are incorporated to ensure the learned knowledge from head items can be well applied for tail item representation learning in the long-tail distribution settings. Through extensive experiments on two benchmark datasets, results show that our proposed dual transfer learning framework significantly outperforms other state-of-the-art methods for tail item recommendation in hit ratio and NDCG. It is also very encouraging that our framework further improves head items and overall performance on top of the gains on tail items.

Many recent advances in neural information retrieval models, which predict top-K items given a query, learn directly from a large training set of (query, item) pairs. However, they are often insufficient when there are many previously unseen (query, item) combinations, often referred to as the cold start problem. Furthermore, the search system can be biased towards items that are frequently shown to a query previously, also known as the 'rich get richer' (a.k.a. feedback loop) problem. In light of these problems, we observed that most online content platforms have both a search and a recommender system that, while having heterogeneous input spaces, can be connected through their common output item space and a shared semantic representation. In this paper, we propose a new Zero-Shot Heterogeneous Transfer Learning framework that transfers learned knowledge from the recommender system component to improve the search component of a content platform. First, it learns representations of items and their natural-language features by predicting (item, item) correlation graphs derived from the recommender system as an auxiliary task. Then, the learned representations are transferred to solve the target search retrieval task, performing query-to-item prediction without having seen any (query, item) pairs in training. We conduct online and offline experiments on one of the world's largest search and recommender systems from Google, and present the results and lessons learned. We demonstrate that the proposed approach can achieve high performance on offline search retrieval tasks, and more importantly, achieved significant improvements on relevance and user interactions over the highly-optimized production system in online experiments.

NLP models are shown to suffer from robustness issues, i.e., a model's prediction can be easily changed under small perturbations to the input.In this work, we present a Controlled Adversarial Text Generation (CAT-Gen) model that, given an input text, generates adversarial texts through controllable attributes that are known to be irrelevant to task labels.For example, in order to attack a model for sentiment classification over product reviews, we can use the product categories as the controllable attribute which should not change the sentiment of the reviews.Experiments on real-world NLP datasets demonstrate that our method can generate more diverse and fluent adversarial texts, compared to many existing adversarial text generation approaches.We further use our generated adversarial examples to improve models through adversarial training, and we demonstrate that our generated attacks are more robust against model retraining and different model architectures.

In batch reinforcement learning (RL), one often constrains a learned policy to be close to the behavior (data-generating) policy, e.g., by constraining the learned action distribution to differ from the behavior policy by some maximum degree that is the same at each state. This can cause batch RL to be overly conservative, unable to exploit large policy changes at frequently-visited, high-confidence states without risking poor performance at sparsely-visited states. To remedy this, we propose residual policies, where the allowable deviation of the learned policy is state-action-dependent. We derive a new for RL method, BRPO, which learns both the policy and allowable deviation that jointly maximize a lower bound on policy performance. We show that BRPO achieves the state-of-the-art performance in a number of tasks.

Much of the previous machine learning (ML) fairness literature assumes that protected features such as race and sex are present in the dataset, and relies upon them to mitigate fairness concerns. However, in practice factors like privacy and regulation often preclude the collection of protected features, or their use for training or inference, severely limiting the applicability of traditional fairness research. Therefore, we ask: How can we train a ML model to improve fairness when we do not even know the protected group memberships? In this work we address this problem by proposing Adversarially Reweighted Learning (ARL). In particular, we hypothesize that non-protected features and task labels are valuable for identifying fairness issues, and can be used to co-train an adversarial reweighting approach for improving fairness. Our results show that ARL improves Rawlsian Max-Min fairness, with notable AUC improvements for worst-case protected groups in multiple datasets, outperforming state-of-the-art alternatives.

Improving user satisfaction is at the forefront of industrial recommender systems. While significant progress has been made by utilizing logged implicit data of user-item interactions (i.e., clicks, dwell/watch time, and other user engagement signals), there has been a recent surge of interest in measuring and modeling user satisfaction, as provided by orthogonal data sources. Such data sources typically originate from responses to user satisfaction surveys, which explicitly ask users to rate their experience with the system and/or specific items they have consumed in the recent past. This data can be valuable for measuring and modeling the degree to which a user has had a satisfactory experience on the recommendation platform, since what users do (engagement) does not always align with what users say they want (satisfaction as measured by surveys).

The challenge of building a good recommendation system is deeply connected to missing data---unknown features and labels to suggest the most "valuable" items to the user. The mysterious properties of the power law distributions that generally arises out of recommender (and social systems in general) create skewed and long-tailed consumption patterns that are often still puzzling to many of us. Missing data and skewed distributions create not just accuracy and recall problems, but also capacity allocation problems, which are at the roots of recent debate on inclusiveness and responsibility. So how do we move forward in the face of these immense conceptual and practical issues?

Modern online content-sharing platforms host billions of items like music, videos, and products uploaded by various providers for users to discover items of their interests. To satisfy the information needs, the task of effective item retrieval (or item search ranking) given user search queries has become one of the most fundamental problems to online content-sharing platforms. Moreover, the same query can represent different search intents for different users, so personalization is also essential for providing more satisfactory search results. Different from other similar research tasks, such as ad-hoc retrieval and product retrieval with copious words and reviews, items in content-sharing platforms usually lack sufficient descriptive information and related meta-data as features. In this paper, we propose the end-to-end deep attentive model (EDAM) to deal with personalized item retrieval for online content-sharing platforms using only discrete personal item history and queries. Each discrete item in the personal item history of a user and its content provider are first mapped to embedding vectors as continuous representations. A query-aware attention mechanism is then applied to identify the relevant contexts in the user history and construct the overall personal representation for a given query. Finally, an extreme multi-class softmax classifier aggregates the representations of both query and personal item history to provide personalized search results. We conduct extensive experiments on a large-scale real-world dataset with hundreds of million users from a large video media platform at Google. The experimental results demonstrate that our proposed approach significantly outperforms several competitive baseline methods. It is also worth mentioning that this work utilizes a massive dataset from a real-world commercial content-sharing platform for personalized item retrieval to provide more insightful analysis from the industrial aspects.

Many real-world recommender systems need to be highly scalable: matching millions of items with billions of users, with milliseconds latency. The scalability requirement has led to widely used two-stage recommender systems, consisting of efficient candidate generation model(s) in the first stage and a more powerful ranking model in the second stage.

Machine learning applications, such as object detection and content recommendation, often require training a single model to predict multiple targets at the same time. Multi-task learning through neural networks became popular recently, because it not only helps improve the accuracy of many prediction tasks when they are related, but also saves computation cost by sharing model architectures and low-level representations. The latter is critical for real-time large-scale machine learning systems. However, classic multi-task neural networks may degenerate significantly in accuracy when tasks are less related. Previous works (Misra et al. 2016; Yang and Hospedales 2016; Ma et al. 2018) showed that having more flexible architectures in multi-task models, either manually-tuned or softparameter-sharing structures like gating networks, helps improve the prediction accuracy. However, manual tuning is not scalable, and the previous soft-parameter sharing models are either not flexible enough or computationally expensive. In this work, we propose a novel framework called SubNetwork Routing (SNR) to achieve more flexible parameter sharing while maintaining the computational advantage of the classic multi-task neural-network model. SNR modularizes the shared low-level hidden layers into multiple layers of subnetworks, and controls the connection of sub-networks with learnable latent variables to achieve flexible parameter sharing. We demonstrate the effectiveness of our approach on a large-scale dataset YouTube8M. We show that the proposed method improves the accuracy of multi-task models while maintaining their computation efficiency.

As more researchers have become aware of and passionate about algorithmic fairness, there has been an explosion in papers laying out new metrics, suggesting algorithms to address issues, and calling attention to issues in existing applications of machine learning. This research has greatly expanded our understanding of the concerns and challenges in deploying machine learning, but there has been much less work in seeing how the rubber meets the road. In this paper we provide a case-study on the application of fairness in machine learning research to a production classification system, and offer new insights in how to measure and address algorithmic fairness issues. We discuss open questions in implementing equality of opportunity and describe our fairness metric, conditional equality, that takes into account distributional differences. Further, we provide a new approach to improve on the fairness metric during model training and demonstrate its efficacy in improving performance for a real-world product.

In this paper, we study counterfactual fairness in text classification, which asks the question: How would the prediction change if the sensitive attribute referenced in the example were different? Toxicity classifiers demonstrate a counterfactual fairness issue by predicting that "Some people are gay" is toxic while "Some people are straight" is nontoxic. We offer a metric, counterfactual token fairness (CTF), for measuring this particular form of fairness in text classifiers, and describe its relationship with group fairness. Further, we offer three approaches, blindness, counterfactual augmentation, and counterfactual logit pairing (CLP), for optimizing counterfactual token fairness during training, bridging the robustness and fairness literature. Empirically, we find that blindness and CLP address counterfactual token fairness. The methods do not harm classifier performance, and have varying tradeoffs with group fairness. These approaches, both for measurement and optimization, provide a new path forward for addressing fairness concerns in text classification.

Characterizing temporal dependence patterns is a critical step in understanding the statistical properties of sequential data. Long Range Dependence (LRD) --- referring to long-range correlations decaying as a power law rather than exponentially w.r.t. distance --- demands a different set of tools for modeling the underlying dynamics of the sequential data. While it has been widely conjectured that LRD is present in language modeling and sequential recommendation, the amount of LRD in the corresponding sequential datasets has not yet been quantified in a scalable and model-independent manner. We propose a principled estimation procedure of LRD in sequential datasets based on established LRD theory for real-valued time series and apply it to sequences of symbols with million-item-scale dictionaries. In our measurements, the procedure estimates reliably the LRD in the behavior of users as they write Wikipedia articles and as they interact with YouTube. We further show that measuring LRD better informs modeling decisions in particular for RNNs whose ability to capture LRD is still an active area of research. The quantitative measure informs new Evolutive Recurrent Neural Networks (EvolutiveRNNs) designs, leading to state-of-the-art results on language understanding and sequential recommendation tasks at a fraction of the computational cost.

Recommender systems are one of the most pervasive applications of machine learning in industry, with many services using them to match users to products or information. As such it is important to ask: what are the possible fairness risks, how can we quantify them, and how should we address them? In this paper we offer a set of novel metrics for evaluating algorithmic fairness concerns in recommender systems. In particular we show how measuring fairness based on pairwise comparisons from randomized experiments provides a tractable means to reason about fairness in rankings from recommender systems. Building on this metric, we offer a new regularizer to encourage improving this metric during model training and thus improve fairness in the resulting rankings. We apply this pairwise regularization to a large-scale, production recommender system and show that we are able to significantly improve the system's pairwise fairness.

Many recommendation systems retrieve and score items from a very large corpus. A common recipe to handle data sparsity and power-law item distribution is to learn item representations from its content features. Apart from many content-aware systems based on matrix factorization, we consider a modeling framework using two-tower neural net, with one of the towers (item tower) encoding a wide variety of item content features. A general recipe of training such two-tower models is to optimize loss functions calculated from in-batch negatives, which are items sampled from a random mini-batch. However, in-batch loss is subject to sampling biases, potentially hurting model performance, particularly in the case of highly skewed distribution. In this paper, we present a novel algorithm for estimating item frequency from streaming data. Through theoretical analysis and simulation, we show that the proposed algorithm can work without requiring fixed item vocabulary, and is capable of producing unbiased estimation and being adaptive to item distribution change. We then apply the sampling-bias-corrected modeling approach to build a large scale neural retrieval system for YouTube recommendations. The system is deployed to retrieve personalized suggestions from a corpus with tens of millions of videos. We demonstrate the effectiveness of sampling-bias correction through offline experiments on two real-world datasets. We also conduct live A/B testings to show that the neural retrieval system leads to improved recommendation quality for YouTube.

In this paper, we introduce a large scale multi-objective ranking system for recommending what video to watch next on an industrial video sharing platform. The system faces many real-world challenges, including the presence of multiple competing ranking objectives, as well as implicit selection biases in user feedback. To tackle these challenges, we explored a variety of soft-parameter sharing techniques such as Multi-gate Mixture-of-Experts so as to efficiently optimize for multiple ranking objectives. Additionally, we mitigated the selection biases by adopting a Wide & Deep framework. We demonstrated that our proposed techniques can lead to substantial improvements on recommendation quality on one of the world's largest video sharing platforms.

Industrial recommender systems deal with extremely large action spaces -- many millions of items to recommend. Moreover, they need to serve billions of users, who are unique at any point in time, making a complex user state space. Luckily, huge quantities of logged implicit feedback (e.g., user clicks, dwell time) are available for learning. Learning from the logged feedback is however subject to biases caused by only observing feedback on recommendations selected by the previous versions of the recommender. In this work, we present a general recipe of addressing such biases in a production top-K recommender system at Youtube, built with a policy-gradient-based algorithm, i.e. REINFORCE. The contributions of the paper are: (1) scaling REINFORCE to a production recommender system with an action space on the orders of millions; (2) applying off-policy correction to address data biases in learning from logged feedback collected from multiple behavior policies; (3) proposing a novel top-K off-policy correction to account for our policy recommending multiple items at a time; (4) showcasing the value of exploration. We demonstrate the efficacy of our approaches through a series of simulations and multiple live experiments on Youtube.

Understanding temporal dynamics has proved to be highly valuable for accurate recommendation. Sequential recommenders have been successful in modeling the dynamics of users and items over time. However, while different model architectures excel at capturing various temporal ranges or dynamics, distinct application contexts require adapting to diverse behaviors.

The ability to capture Long Range Dependence (LRD) in a stochastic process is of prime importance in the context of predictive models. A sequential model with a longer-term memory is better able contextualize recent observations. In this article, we apply the theory of LRD stochastic processes to modern recurrent architectures, such as LSTMs and GRUs, and prove they do not provide LRD under assumptions sufficient for gradients to vanish. Motivated by an information-theoretic analysis, we provide a modified recurrent neural architecture that mitigates the issue of faulty memory through redundancy while keeping the compute time constant. Experimental results on a synthetic copy task, the Youtube-8m video classification task and a recommender system show that we enable better memorization and longer-term memory.

Many recommendation systems produce result sets with large numbers of highly similar items. Diversifying these results is often accomplished with heuristics, which are impoverished models of users' desire for diversity. However, integrating more complex statistical models of diversity into large-scale, mature systems is challenging. Without a good match between the model's definition of diversity and users' perception of diversity, the model can easily degrade users' perception of the recommendations. In this work we present a statistical model of diversity based on determinantal point processes (DPPs). We train this model from examples of user preferences with a simple procedure that can be integrated into large and complex production systems relatively easily. We use an approximate inference algorithm to serve the model at scale, and empirical results on live YouTube homepage traffic show that this model, coupled with a re-ranking algorithm, yields substantial short- and long-term increases in user engagement.

Recommendation systems, prevalent in many applications, aim to surface to users the right content at the right time. Recently, researchers have aspired to develop conversational systems that offer seamless interactions with users, more effectively eliciting user preferences and offering better recommendations. Taking a step towards this goal, this paper explores the two stages of a single round of conversation with a user: which question to ask the user, and how to use their feedback to respond with a more accurate recommendation. Following these two stages, first, we detail an RNN-based model for generating topics a user might be interested in, and then extend a state-of-the-art RNN-based video recommender to incorporate the user's selected topic. We describe our proposed system Q&R, i.e., Question & Recommendation, and the surrogate tasks we utilize to bootstrap data for training our models. We evaluate different components of Q&R on live traffic in various applications within YouTube: User Onboarding, Homepage Recommendation, and Notifications. Our results demonstrate that our approach improves upon state-of-the-art recommendation models, including RNNs, and makes these applications more useful, such as a >1% increase in video notifications opened. Further, our design choices can be useful to practitioners wanting to transition to more conversational recommendation systems.

Neural-based multi-task learning has been successfully used in many real-world large-scale applications such as recommendation systems. For example, in movie recommendations, beyond providing users movies which they tend to purchase and watch, the system might also optimize for users liking the movies afterwards. With multi-task learning, we aim to build a single model that learns these multiple goals and tasks simultaneously. However, the prediction quality of commonly used multi-task models is often sensitive to the relationships between tasks. It is therefore important to study the modeling tradeoffs between task-specific objectives and inter-task relationships. In this work, we propose a novel multi-task learning approach, Multi-gate Mixture-of-Experts (MMoE), which explicitly learns to model task relationships from data. We adapt the Mixture-of-Experts (MoE) structure to multi-task learning by sharing the expert submodels across all tasks, while also having a gating network trained to optimize each task. To validate our approach on data with different levels of task relatedness, we first apply it to a synthetic dataset where we control the task relatedness. We show that the proposed approach performs better than baseline methods when the tasks are less related. We also show that the MMoE structure results in an additional trainability benefit, depending on different levels of randomness in the training data and model initialization. Furthermore, we demonstrate the performance improvements by MMoE on real tasks including a binary classification benchmark, and a large-scale content recommendation system at Google.

Many techniques to utilize side information of users and/or items as inputs to recommenders to improve recommendation, especially on cold-start items/users, have been developed over the years. In this work, we test the approach of utilizing item side information, specifically categorical attributes, in the output of recommendation models either through multi-task learning or hierarchical classification. We first demonstrate the efficacy of these approaches for both matrix factorization and neural networks with a medium-size real-word data set. We then show that they improve a neural-network based production model in an industrial-scale recommender system. We demonstrate the robustness of the hierarchical classification approach by introducing noise in building the hierarchy. Lastly, we investigate the generalizability of hierarchical classification on a simulated dataset by building two user models in which we can fully control the generative process of user-item interactions.

Indexes are models: a \btree-Index can be seen as a model to map a key to the position of a record within a sorted array, a Hash-Index as a model to map a key to a position of a record within an unsorted array, and a BitMap-Index as a model to indicate if a data record exists or not. In this exploratory research paper, we start from this premise and posit that all existing index structures can be replaced with other types of models, including deep-learning models, which we term \em learned indexes. We theoretically analyze under which conditions learned indexes outperform traditional index structures and describe the main challenges in designing learned index structures. Our initial results show that our learned indexes can have significant advantages over traditional indexes. More importantly, we believe that the idea of replacing core components of a data management system through learned models has far reaching implications for future systems designs and that this work provides just a glimpse of what might be possible.

The success of recommender systems often depends on their ability to understand and make use of the context of the recommendation request. Significant research has focused on how time, location, interfaces, and a plethora of other contextual features affect recommendations. However, in using deep neural networks for recommender systems, researchers often ignore these contexts or incorporate them as ordinary features in the model.

When searching on the web or in an app, results are often returned as lists of hundreds to thousands of items, making it difficult for users to understand or navigate the space of results. Research has demonstrated that using clustering to partition search results into coherent, topical clusters can aid in both exploration and discovery. Yet clusters generated by an algorithm for this purpose are often of poor quality and do not satisfy users. To achieve acceptable clustered search results, experts must manually evaluate and refine the clustered results for each search query, a process that does not scale to large numbers of search queries. In this article, we investigate using crowd-based human evaluation to inspect, evaluate, and improve clusters to create high-quality clustered search results at scale. We introduce a workflow that begins by using a collection of well-known clustering algorithms to produce a set of clustered search results for a given query. Then, we use crowd workers to holistically assess the quality of each clustered search result to find the best one. Finally, the workflow has the crowd spot and fix problems in the best result to produce a final output. We evaluate this workflow on 120 top search queries from the Google Play Store, some of whom have clustered search results as a result of evaluations and refinements by experts. Our evaluations demonstrate that the workflow is effective at reproducing the evaluation of expert judges and also improves clusters in a way that agrees with experts and crowds alike.

When building a recommender system, how can we ensure that all items are modeled well? Classically, recommender systems are built, optimized, and tuned to improve a global prediction objective, such as root mean squared error. However, as we demonstrate, these recommender systems often leave many items badly-modeled and thus under-served. Further, we give both empirical and theoretical evidence that no single matrix factorization, under current state-of-the-art methods, gives optimal results for each item.

Desktop users frequently open and switch between multiple windows. Here we present an experiment comparing 3 window switching interfaces: the Cards interface spreads windows out like a vertical stack of cards with the most recent window at the front; the Mosaic interface places each window in a grid ordered by recency; and, the Exposé interface provides an map-like overview based on the relative size and position of windows. Experimental results suggest that the Mosaic interface scales, enabling faster window selection than the Cards interface and less erroneous window selection than the Exposé interface.

A relatively new feature in Google Play Store presents mobile app search results grouped by topic, helping users to quickly navigate and explore. The underlying Search Results Clustering (SRC) system faces several challenges, including grouping search results in topical coherent clusters as well as finding the appropriate level of granularity for clustering. We present AppGrouper, an alternative approach to algorithmic-only solutions, incorporating human input in a knowledge-graph-based clustering process. AppGrouper provides an interactive interface that lets domain experts steer the clustering process in early, mid, and late stages. We deployed and evaluated AppGrouper with internal experts. We found that AppGroup improved quality of algorithm-generated app clusters on 56 out of 82 search queries. We also found that the internal experts made more changes in early and mid stages for lower-quality algorithmic results, focusing more on narrow queries. Our result suggests, in some contexts, machine learning systems can greatly benefit from steering from human experts, creating a symbiotic working relationship.

Researchers have recently been focusing on understanding online communities in social networks that offer easy access to new audiences. How do online communities function within these social networks? In this work, we conducted a mixed-method study of public Google+ Communities and found two major types evident in both how users talk about them and how they appear to use them: plazas to meet new people, and topic boards to discuss common interests. This reflects two common motivations users cite in describing Communities: "meeting like minded people" and "finding great content". We characterize these two types of Communities within Google+ using mixed methods including surveys, interviews, and quantitative analytics, and expose differences in user behaviors between them.

Crowdsourcing has become a popular and indispensable component of many problem-solving pipelines in the research literature, with crowd workers often treated as computational resources that can reliably solve problems that computers have trouble with, such as image labeling/classification, natural language processing, or document writing. Yet, obviously crowd workers are human, and long sequences of the same monotonous tasks might intuitively reduce the amount of good quality work done by the workers. Here we propose an investigation into how we can use diversions containing small amounts of entertainment to improve crowd workers' experiences. We call these small period of entertainment ``micro-diversions", which we hypothesize to provide timely relief to workers during long sequences of micro-tasks. We hope to improve productivity by retaining workers to work on our tasks longer and to either improve or retain the quality of work. We experimentally test micro-diversions on Amazon's Mechanical Turk, a large paid-crowdsourcing platform. We find that micro-diversions can significantly improve worker retention rate while retaining the same work quality.

Search and recommendation engines are increasingly more intelligent. They have become more personalized and social as well as more interactive. No longer just offering ten blue links, search engines have increasingly been integrated with task and item recommenders directly, for example, to offer news, movie, music, and dining suggestions. Vice versa, recommendation systems have increasingly became more search-like by offering capabilities that enable users to tune and direct recommendation results instantly. As the two technologies evolve toward each other, there is increasingly a blurring of the boundary between these two approaches to interactive information seeking. On the search side, this is driven by the merging of question answering capabilities with search, led by systems like Google Now and Apple Siri that move search toward intelligent personal assistants. On the recommendation side, there has been a merging of techniques from not just keyword search but also faceted search, along with user-based and item-based collaborative filtering techniques and other more proactive recommenders.

Many recommenders aim to provide relevant recommendations to users by building personal topic interest profiles and then using these profiles to find interesting contents for the user. In social media, recommender systems build user profiles by directly combining users' topic interest signals from a wide variety of consumption and publishing behaviors, such as social media posts they authored, commented on, +1'd or liked. Here we propose to separately model users' topical interests that come from these various behavioral signals in order to construct better user profiles. Intuitively, since publishing a post requires more effort, the topic interests coming from publishing signals should be more accurate of a user's central interest than, say, a simple gesture such as a +1. By separating a single user's interest profile into several behavioral profiles, we obtain better and cleaner topic interest signals, as well as enabling topic prediction for different types of behavior, such as topics that the user might +1 or comment on, but might never write a post on that topic.

As news reading becomes more social, how do different types of annotations affect people's selection of news articles? This paper reports on results from two experiments looking at social annotations in two different news reading contexts. The first experiment simulates a logged-out experience with annotations from strangers, a computer agent, and a branded company. Results indicate that, perhaps unsurprisingly, annotations by strangers have no persuasive effects. However, surprisingly, unknown branded companies still had a persuasive effect. The second experiment simulates a logged-in experience with annotations from friends, finding that friend annotations are both persuasive and improve user satisfaction over their article selections. In post-experiment interviews, we found that this increased satisfaction is due partly because of the context that annotations add. That is, friend annotations both help people decide what to read, and provide social context that improves engagement. Interviews also suggest subtle expertise effects. We discuss implications for design of social annotation systems and suggestions for future research.

Tabbed web browsing interfaces enable users to multi-task and easily switch between open web pages. However, tabbed browsing is difficult for mobile web browsers due to the limited screen space and the reduced precision of touch. We present an experiment comparing Safari's pages-based switching interface using horizontal swiping gestures with the stacked cards-based switching interface using vertical scrolling gestures, introduced by Chrome. The results of our experiment show that cards-based switching interface allows for faster switching and is less frustrating, with no significant effect on error rates. We generalize these findings, and provide design implications for mobile information spaces.

Consumer review sites and recommender systems typically rely on a large volume of user-contributed ratings, which makes rating acquisition an essential component in the design of such systems. User ratings are then summarized to provide an aggregate score representing a popular evaluation of an item. An inherent problem in such summarization is potential bias due to raters self-selection and heterogeneity in terms of experience, tastes and rating scale interpretation. There are two major approaches to collecting ratings, which have different advantages and disadvantages. One is to allow a large number of volunteers to choose and rate items directly (a method employed by e.g. Yelp and Google Places). Alternatively, a panel of raters may be maintained and invited to rate a predefined set of items at regular intervals (such as in Zagat Survey). The latter approach arguably results in more consistent reviews and reduced selection bias, however, at the expense of much smaller coverage (fewer rated items).

Summary form only given. Social interactions have always been an important part of human learning and experience. We now know that social interactions are critical in many knowledge and information processes. Research has shown results ranging from influences on our behavior from social networks [Aral2012] to our understanding of social belonging on health [Walton2011], as well as how conflicts and coordination play out in Wikipedia [Kittur2007]. Interestingly, social scientists have studied social interactions for many years, but it wasn't until very recently that researchers are able to study these mechanisms through the explosion of services and data available on web-based social systems. In this talk, I plan to illustrate a model-driven approach to researching social interactions on the Web. Our research methods and systems are informed by models such as information scent, sense making, information theory, probabilistic models, and evolutionary dynamic models. These models have been used to understand a wide variety of user behaviors, from individuals interacting with social bookmarks in Delicious to groups of people working on articles in Wikipedia. These models range in complexity from a simple set of assumptions to complex equations describing human and group behaviors. By using this model-driven approach, we further our understanding of how knowledge is fundamentally constructed in a social context, and a path forward for further social interaction research. An interesting question is whether a machine can assist humans in being creative and inspire a user during the creation of 3D models or a shape in general. One possible means to achieve this is through a design gallery which presents a variety of computed suggestive designs from which the user can pick the ones he likes the best. The ensuing challenge is how to come up with intriguing suggestions which inspire creativity, rather than banal suggestions which stall the design process. In my talk I will discuss about the notion of creative modeling, synthesis of inspiring examples, the analysis of a set, and show a number of recent works that step towards this end.

As web search increasingly becomes reliant on social signals, it is imperative for us to understand the effect of these signals on users' behavior. There are multiple ways in which social signals can be used in search: (a) to surface and rank important social content; (b) to signal to users which results are more trustworthy and important by placing annotations on search results. We focus on the latter problem of understanding how social annotations affect user behavior.

Online social networks have become indispensable tools for information sharing, but existing 'all-or-nothing' models for sharing have made it difficult for users to target information to specific parts of their networks. In this paper, we study Google+, which enables users to selectively share content with specific 'Circles' of people. Through a combination of log analysis with surveys and interviews, we investigate how active users organize and select audiences for shared content. We find that these users frequently engaged in selective sharing, creating circles to manage content across particular life facets, ties of varying strength, and interest-based groups. Motivations to share spanned personal and informational reasons, and users frequently weighed ''limiting' factors (e.g. privacy, relevance, and social norms) against the desire to reach a large audience. Our work identifies implications for the design of selective sharing mechanisms in social networks.

We ask how to best present social annotations on search results, and attempt to find an answer through mixed-method eye-tracking and interview experiments. Current practice is anchored on the assumption that faces and names draw attention; the same presentation format is used independently of the social connection strength and the search query topic. The key findings of our experiments indicate room for improvement. First, only certain social contacts are useful sources of information, depending on the search topic. Second, faces lose their well-documented power to draw attention when rendered small as part of a social search result annotation. Third, and perhaps most surprisingly, social annotations go largely unnoticed by users in general due to selective, structured visual parsing behaviors specific to search result pages. We conclude by recommending improvements to the design and content of social annotations to make them more noticeable and useful.

Context-aware information is widely available in various ways and is becoming more and more important for enhancing retrieval performance and recommendation results. The current main issue to cope with is not only recommending or retrieving the most relevant items and content, but defining them ad hoc. Other relevant issues include personalizing and adapting the information and the way it is displayed to the user's current situation and interests. Ubiquitous computing further provides new means for capturing user feedback on items and providing information.

Conversation is a key element in online social streams such as Twitter and Facebook. However, finding interesting conversations to read is often a challenge, due to information overload and differing user preferences. In this work we explored five algorithms that recommend conversations to Twitter users, utilizing thread length, topic and tie-strength as factors. We compared the algorithms through an online user study and gathered feedback from real Twitter users. In particular, we investigated how users' purposes of using Twitter affect user preferences for different types of conversations and the performance of different algorithms. Compared to a random baseline, all algorithms recommended more interesting conversations. Further, tie-strength based algorithms performed significantly better for people who use Twitter for social purposes than for people who use Twitter for informational purpose only.

Little research exists on one of the most common, oldest, and most utilized forms of online social geographic information: the 'location' field found in most virtual community user profiles. We performed the first in-depth study of user behavior with regard to the location field in Twitter user profiles. We found that 34% of users did not provide real location information, frequently incorporating fake locations or sarcastic comments that can fool traditional geographic information tools. When users did input their location, they almost never specified it at a scale any more detailed than their city. In order to determine whether or not natural user behaviors have a real effect on the 'locatability' of users, we performed a simple machine learning experiment to determine whether we can identify a user's location by only looking at what that user tweets. We found that a user's country and state can in fact be determined easily with decent accuracy, indicating that users implicitly reveal location information, with or without realizing it. Implications for location-based services and privacy are discussed.

Enterprise wikis commonly see low adoption rates, preventing them from reaching the critical mass that is needed to make them valuable. The high interaction costs for contributing content to these wikis is a key factor impeding wiki adoption. Much of the collaboration among knowledge workers continues to occur in email, which causes useful information to stay siloed in personal inboxes. In this demo we present Mail2Wiki, a system that enables easy contribution and initial curation of content from the personal space of email to the shared repository of a wiki.

Corporate wikis are affected by poor adoption rates. The high interaction costs required to organize and maintain information in these wikis are a key factor that limits broader adoption. We present VisualWikiCurator, a wiki extension designed to lower such costs by (a) recommending new content to easily update a wiki page, and (b) extracting structured data from the wiki page while providing new alternative visualizations of the data. The visualizations of extracted semantic data act both as alternative views and as tools to organize the page content. Since no information extraction algorithm is perfect with generic unstructured data, we use a mixed-initiative approach to allow users to refine machine-extracted metadata and easily re-organize the content in wiki pages.

This paper presents a research project on the design of a cross-channel information management tool for knowledge workers: we focus on IT services professionals in a large enterprise who work in multiple ad hoc task forces. Through three rounds of investigation, we characterized their work practices and needs, specified their requirements for a cross-channel information management tool, and designed and evaluated a prototype to address these needs. We found that these workers shared the problem of managing information across multiple channels, requiring better support for aggregating, filtering, and organizing this information. We report the requirements elicited and the prototypes built during the design process.

More and more web users keep up with newest information through information streams such as the popular micro-blogging website Twitter. In this paper we studied content recommendation on Twitter to better direct user attention. In a modular approach, we explored three separate dimensions in designing such a recommender: content sources, topic interest models for users, and social voting. We implemented 12 recommendation engines in the design space we formulated, and deployed them to a recommender service on the web to gather feedback from real Twitter users. The best performing algorithm improved the percentage of interesting content to 72% from a baseline of 33%. We conclude this work by discussing the implications of our recommender design and how our design can generalize to other information streams.

Information overload is a growing threat to the productivity of today's knowledge workers, who need to keep track of multiple streams of information from various sources. RSS feed readers are a popular choice for syndicating information streams, but current tools tend to contribute to the overload problem instead of solving it. We introduce FeedWinnower, an enhanced feed aggregator that helps readers to filter feed items by four facets (topic, people, source, and time), thus facilitating feed triage. The combination of the four facets provides a powerful way for users to slice and dice their personal feeds. In addition, we present a formative evaluation of the prototype conducted with 15 knowledge workers in two different organizations.

Web 2.0 tools are penetrating into organizations after their successful adoption in the consumer domain (e.g., social networking; sharing of photos, videos, tags, or bookmarks; wiki-based editing). Some of these new tools and the collaborative processes that they support on the large scale are often referred to as Collective Intelligence (CI). The workshop brings together leading researchers and designers who are studying or developing CI tools aimed at workers in organizations. The goal is to further articulate the emerging research agenda for this new CSCW area and define new observed forms of CI in organization. Studies of communities, CI tools, and new methods are discussed.

Twitter streams are on overload: active users receive hundreds of items per day, and existing interfaces force us to march through a chronologically-ordered morass to find tweets of interest. We present an approach to organizing a user's own feed into coherently clustered trending topics for more directed exploration. Our Twitter client, called Eddi, groups tweets in a user's feed into topics mentioned explicitly or implicitly, which users can then browse for items of interest. To implement this topic clustering, we have developed a novel algorithm for discovering topics in short status updates powered by linguistic syntactic transformation and callouts to a search engine. An algorithm evaluation reveals that search engine callouts outperform other approaches when they employ simple syntactic transformation and backoff strategies. Active Twitter users evaluated Eddi and found it to be a more efficient and enjoyable way to browse an overwhelming status update feed than the standard chronological interface.

How can we build systems that enable users to mix and match tools together? How will we know whether we have done a good job in creating usable visual interactive systems that help users accomplish a wide variety of goals? How can people share the results of their explorations with each other, and for innovative tools to be remixed? Widely-used tools such as Web Browsers, Wikis, spreadsheets, and analytics environments like R all contain models of how people mix and combine operators and functionalities. In my own research, system developments are very much informed by models such as information scent, sense making, information theory, probabilistic models, and more recently, evolutionary dynamic models. These models have been used to understand a wide-variety of user behaviors in human-centric computing, from individuals interacting with a search system like MrTaggy.com to groups of people working on articles in Wikipedia. These models range in complexity from a simple set of assumptions to complex equations describing human and group behavior. In this talk, I will attempt to illustrate how a model-driven approach to answering the above questions should help to illuminate the path forward for Human-Centric Computing.

A common practice in work groups is to share links to interesting web pages. Moreover, passages in these web pages are often cut-and-pasted, and used in various other contexts. In this paper we report how we explore the idea of paragraph fingerprinting to achieve the goal of annotate once, appear anywhere"in a social annotation system called SparTag.us. This work was motivated by the prominence of redundant contents with different URLs on the Web and shared documents that are read and re-read within enterprises. Our technique attaches users' annotations to the contents of paragraphs, enabling annotations to move along with the paragraphs within dynamic live pages and travel across page boundary to other pages as long as the paragraph contents remain intact. We also describe how we use paragraph fingerprinting to facilitate the social sharing of information nuggets among our users.

Social tagging arose out of the need to organize found content that is worth revisiting. A significant side effect has been the use of social tagging sites as navigational signposts for interesting content. The collective behavior of users who tagged contents seems to offer a good basis for exploratory search interfaces, even for users who are not using social bookmarking sites. In this paper, we present the design of a tag-based exploratory system and detail an experiment in understanding its effectiveness. The tag-based search system allows users to utilize relevance feedback on tags to indicate their interest in various topics, enabling rapid exploration of the topic space. The experiment shows that the system seems to provide a kind of scaffold for users to learn new topics.

Wikipedia is an online encyclopedia which has undergone tremendous growth. However, this same growth has made it difficult to characterize its content and coverage. In this paper we develop measures to map Wikipedia using its socially annotated, hierarchical category structure. We introduce a mapping technique that takes advantage of socially-annotated hierarchical categories while dealing with the inconsistencies and noise inherent in the distributed way that they are generated. The technique is demonstrated through two applications: mapping the distribution of topics in Wikipedia and how they have changed over time; and mapping the degree of conflict found in each topic area. We also discuss the utility of the approach for other applications and datasets involving collaboratively annotated category hierarchies.

In prior work we reported on the design of a social annotation system, SparTag.us, for use in sensemaking activities such as work-group reading and report writing. Previous studies of note-taking systems have demonstrated behavioral differences in social annotation practices, but are not clear in the actual performance gains provided by social features. This paper presents a laboratory study aimed at evaluating the learning effect of social features in SparTag.us. We found significant learning gains, and consider implications for design and for understanding the underlying mechanisms in play when people use social annotation systems.

We are experiencing a new Social Web, where people share, communicate, commiserate, and conflict with each other. As evidenced by systems like Wikipedia, twitter, and delicious.com, these environments are turning people into social information foragers and sharers. Groups interact to resolve conflicts and jointly make sense of topic areas from "Obama vs. Clinton" to "Islam."

Tagging systems such as del.icio.us and Diigo have become important ways for users to organize information gathered from the Web. However, despite their popularity among early adopters, tagging still incurs a relatively high interaction cost for the general users. We introduce a new tagging system called SparTag.us, which uses an intuitive Click2Tag technique to provide in situ, low cost tagging of web content. SparTag.us also lets users highlight text snippets and automatically collects tagged or highlighted paragraphs into a system-created notebook, which can be later browsed and searched. We report several user studies aimed at evaluating Click2Tag and SparTag.us.

This paper presents a context-aware mobile recommender system, codenamed Magitti. Magitti is unique in that it infers user activity from context and patterns of user behavior and, without its user having to issue a query, automatically generates recommendations for content matching. Extensive field studies of leisure time practices in an urban setting (Tokyo) motivated the idea, shaped the details of its design and provided data describing typical behavior patterns. The paper describes the fieldwork, user interface, system components and functionality, and an evaluation of the Magitti prototype.

User studies are important for many aspects of the design process and involve techniques ranging from informal surveys to rigorous laboratory studies. However, the costs involved in engaging users often requires practitioners to trade off between sample size, time requirements, and monetary costs. Micro-task markets, such as Amazon's Mechanical Turk, offer a potential paradigm for engaging a large number of users for low time and monetary costs. Here we investigate the utility of a micro-task market for collecting user measurements, and discuss design considerations for developing remote micro user evaluation tasks. Although micro-task markets have great potential for rapidly collecting user measurements at low costs, we found that special care is needed in formulating tasks in order to harness the capabilities of the approach.

Wikis are collaborative systems in which virtually anyone can edit anything. Although wikis have become highly popular in many domains, their mutable nature often leads them to be distrusted as a reliable source of information. Here we describe a social dynamic analysis tool called WikiDashboard which aims to improve social transparency and accountability on Wikipedia articles. Early reactions from users suggest that the increased transparency afforded by the tool can improve the interpretation, communication, and trustworthiness of Wikipedia articles.

Social networking promises individuals new dimensions of freedom to interact, associate, and give expression to their talents. Recently, systems such as Mechanical Turk have started to facilitate self-organizing collaboration on work-related tasks. Such developments raise interesting questions. Is it possible to create (and sustain) businesses that do not have traditional, formal structure - without traditional "employees"? Can we find and organize (and optimize) talent on the web for task-oriented work - spontaneously and efficiently? How do people relate to one another in possibly evanescent workgroups? One aspect of the challenge in the Social Workspace is understanding and modeling the user behavior and the economic basis for creating, preserving, and exchanging value in the marketplace when workgroup identity, orientations to property, recruiting and managing appropriate talent are not organized under traditional company structures. Another aspect is the technology needed to support virtual organizations and work. The panel will discuss trends in social work and the evolving (scientific) basis of our understanding of new models of workers and organizations.

Search engine researchers typically depict search as the solitary activity of an individual searcher. In contrast, results from our critical-incident survey of 150 users on Amazon's Mechanical Turk service suggest that social interactions play an important role throughout the search process. Our main contribution is that we have integrated models from previous work in sensemaking and information seeking behavior to present a canonical social model of user activities before, during, and after search, suggesting where in the search process both explicitly and implicitly shared information may be valuable to individual searchers.

Wikipedia has become one of the most important information resources on the Web by promoting peer collaboration and enabling virtually anyone to edit anything. However, this mutability also leads many to distrust it as a reliable source of information. Although there have been many attempts at developing metrics to help users judge the trustworthiness of content, it is unknown how much impact such measures can have on a system that is perceived as inherently unstable. Here we examine whether a visualization that exposes hidden article information can impact readers' perceptions of trustworthiness in a wiki environment. Our results suggest that surfacing information relevant to the stability of the article and the patterns of editor behavior can have a significant impact on users' trust across a variety of page types.

Wikipedia, a wiki-based encyclopedia, has become one of the most successful experiments in collaborative knowledge building on the Internet. As Wikipedia continues to grow, the potential for conflict and the need for coordination increase as well. This article examines the growth of such non-direct work and describes the development of tools to characterize conflict and coordination costs in Wikipedia. The results may inform the design of new collaborative knowledge systems.

Researchers have noticed that readers are increasingly skimming instead of reading in depth. Skimming also occur in re-reading activities, where the goal is to recall specific topical facts. Bookmarks and highlighters were invented precisely to achieve this goal. For skimming activities, readers need effective ways to direct their attention toward the most relevant passages within text. We describe how we have enhanced skimming activity by conceptually highlighting sentences within electronic text that relate to search keywords. We perform the conceptual highlighting by computing what conceptual keywords are related to each other via word co-occurrence and spreading activation. Spreading activation is a cognitive model developed in psychology to simulate how memory chunks and conceptual items are retrieved in our brain. We describe the method used, and illustrate the idea with realistic scenarios using our system.

This paper describes the 3Book, a 3D interactive visualization of a codex book as a component for various digital library and sensemaking systems. The book is designed to hold large books and to support sensemaking operations by readers. The book includes methods in which the automatic semantic analysis of the book's content is used to dynamically tailor access.

Ubiquitous and Wearable Computing both have the goal of pushing the computer into the background, supporting all kinds of human activities. Application areas include areas such as everyday environments (e.g. clothing, home, office), promoting new forms of creative learning via physical/virtual objects, and new tools for interactive design. In this paper, we thrust ubiquitous computing into the extremely hostile environment of the sparring ring of a martial art competition. Our system uses piezoelectric force sensors that transmit signals wirelessly to enable the detection of when a significant impact has been delivered to a competitor's body. The objective is to support the judges in scoring the sparring matches accurately, while preserving the goal of merging and blending into the background of the activity. The system therefore must take into account of the rules of the game, be responsive in real-time asynchronously, and often cope with untrained operators of the system. We present a pilot study of the finished prototype and detail our experience.

According to usability experts, the top user issue for Web sites is difficult navigation. We have been developing auto-mated usability tools for several years, and here we describe a prototype service called InfoScent™ Bloodhound Simula-tor, a push-button navigation analysis system, which auto-matically analyzes the information cues on a Web site to produce a usability report. We further build upon previous algorithms to create a method called Information Scent Absorption Rate, which measures the navigability of a site by computing the probability of users reaching the desired destinations on the site. Lastly, we present a user study involving 244 subjects over 1385 user sessions that show how Bloodhound correlates with real users surfing for in-formation on four Web sites. The hope is that, by using a simulation of user surfing behavior, we can reduce the need for human labor during usability testing, thus dramatically lower testing costs, and ultimately improving user experience. The Bloodhound Project is unique in that we apply a concrete HCI theory directly to a real-world prob-lem. The lack of empirically validated HCI theoretical model has plagued the development of our field, and this is a step toward that direction.

The two predominant paradigms for finding information on the Web are browsing and keyword searching. While they exhibit complementary advantages, neither paradigm alone is adequate for complex information goals that lend themselves partially to browsing and partially to searching. To integrate browsing and searching smoothly into a single interface, we introduce a novel approach called ScentTrails. Based on the concept of information scent developed in the context of information foraging theory, ScentTrails highlights hyperlinks to indicate paths to search results. This interface enables users to interpolate smoothly between searching and browsing to locate content matching complex information goals effectively. In a preliminary user study, ScentTrails enabled subjects to find information more quickly than by either searching or browsing alone.

Visualization can be viewed as a process that transforms raw data (value) into views. There has been two major category of data process models that have been proposed to model the visualization transformation process. This paper seeks to compare the Data Flow Models and the Data State Models. Specifically, it proves that, in terms of expressiveness, anything that can represented using the Data Flow Model can also be represented using the Data State Model, and vice versa.

Understanding user behaviors on Web sites enables site owners to make sites more usable, ultimately helping users to achieve their goals more quickly. Accordingly, researchers have devised methods for categorizing user sessions in hopes of revealing user interests. These techniques build user profiles by combining users' navigation paths with other data features, such as page viewing time, hyperlink structure, and page content. Previously, we have presented complex techniques of combining many of these data features to cluster user profiles. In this paper, we introduce a user study and a systematic evaluation of these different data features and their associated weighting schemes. We present the results of our study, including accuracy measures for a number of clustering approaches, and offer recommendations for Web analysts. While further investigation over more sites is needed to definitively settle on a robust scheme, we have characterized this analytic space

On the Web, users typically forage for information by navigating from page to page along Web links. Their surfing patterns or actions are guided by their information needs. Researchers need tools to explore the complex interactions between user needs, user actions, and the structures and contents of the Web. In this paper, we describe two computational methods for understanding the relationship between user needs and user actions. First, for a particular pattern of surfing, we seek to infer the associated information need. Second, given an information need, and some pages as starting pints, we attempt to predict the expected surfing patterns. The algorithms use a concept called “information scent”, which is the subjective sense of value and cost of accessing a page based on perceptual cues. We present an empirical evaluation of these two algorithms, and show their effectiveness.

Designers and researchers of users' interactions with the World Wide Web need tools that permit the rapid exploration of hypotheses about complex interactions of user goals, user behaviors, and Web site designs. We present an architecture and system for the analysis and prediction of user behavior and Web site usability. The system integrates research on human information foraging theory, a reference model of information visualization and Web data-mining techniques. The system also incorporates new methods of Web site visualization (Dome Tree, Usage Based Layouts), a new predictive modeling technique for Web site use (Web User Flow by Information Scent, WUFIS), and new Web usability metrics.

Digital books can significantly enhance the reading experience, providing many functions not available in printed books. In this paper we study a particular augmentation of digital books that provides readers with customized recommendations. We systematically explore the application of spreading activation over text and citation data to generate useful recommendations. Our findings reveal that for the tasks performed in our corpus, spreading activation over text is more useful than citation data. Further, fusing text and citation data via spreading activation results in the most useful recommendations. The fused spreading activation techniques outperform traditional text-based retrieval methods. Finally, we introduce a preliminary user interface for the display of recommendations from these algorithms.

Several visualizations have emerged which attempt to visualize all or part of the World Wide Web.Those visualizations, however, fail to present the dynamically changing ecology of users and documents on the Web.We present new techniques for Web Ecology and Evolution Visualization (WEEV).Disk Trees represent a discrete time slice of the Web ecology.A collection of Disk Trees forms a Time Tube, representing the evolution of the Web over longer periods of time.These visualizations are intended to aid authors and webmasters with the production and organization of content, assist Web surfers making sense of information, and help researchers understand the Web.

In information visualization, as the volume and complexity of the data increases, researchers require more powerful visualization tools that enable them to more effectively explore multidimensional datasets. In this paper, we discuss the general utility of a novel visualization spreadsheet framework. Just as a numerical spreadsheetenables exploration of numbers, a visualization spreadsheet enables exploration of visual forms of information. We show that the spreadsheet approach facilitates certain information visualization tasks that are more difficult using other approaches. Unlike traditional spreadsheets, which store only simple data elements and formulas in each cell, a visualization spreadsheetcell can hold anentire complex data set, selection criteria, viewing specifications, and other information needed for a full-fledged information visualization. Similarly, inter-cell operations are far more complex, stretching beyond simple arithmetic and string operations to encompass a range of domain-specific operators. We have built two prototype systems that illustrate some of these research issues. The underlying approach in our work allows domain experts to define new data types and data operations, and enables visualization experts to incorporate new visualizations, viewing parameters, and view operations. 1

Information visualization faces challenges presented by the need to represent abstract data and the relationships within the data. Previously, we presented a system for visualizing similarities between a single DNA sequence and a large database of other DNA sequences (E.H. Chi et al., 1995). Similarity algorithms generate similarity information in textual reports that can be hundreds or thousands of pages long. Our original system visualized the most important variables from these reports. However, the biologists we work with found this system so useful they requested visual representations of other variables. We present an enhanced system for interactive exploration of this multivariate data. We identify a larger set of useful variables in the information space. The new system involves more variables, so it focuses on exploring subsets of the data. We present an interactive system allowing mapping of different variables to different axes, incorporating animation using a time axis, and providing tools for viewing subsets of the data. Detail-on-demand is preserved by hyperlinks to the analysis reports. We present three case studies illustrating the use of these techniques. The combined technique of applying a time axis with a 3D scatter plot and query filters to visualization of biological sequence similarity data is both powerful and novel.

Biological sequence similarity analysis presents visualization challenges, primarily because of the massive amounts of discrete, multi dimensional data. Genomic data generated by molecular biologists is analyzed by algorithms that search for similarity to known sequences in large genomic databases. The output from these algorithms can be several thousand pages of text, and is difficult to analyze because of its length and complexity. We developed and implemented a novel graphical representation for sequence similarity search results, which visually reveals features that are difficult to find in textual reports. The method opens new possibilities in the interpretation of this discrete, multidimensional data by enabling interactive investigation of the graphical representation.