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.

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.

Dynamic Partially Reconfiguration (DPR) on FPGA has attracted significant research interest in recent years since it provides benefits such as reduced area and flexible functionality. However, due to the lack of supporting synthesis tools in the current DPR design flow, leveraging benefits from DPR requires specific design expertise with laborious manual design effort. Considering the complicated concurrency relations among various functions, it is challenging to select appropriate Partial Reconfiguration Modules (PR Modules) and cluster them into proper groups with a proper reconfiguration schedule so that the hardware modules can be swapped in and out correctly during the run time. Furthermore, the design of PR Modules also impacts reconfiguration latency and resource utilization greatly. In this paper, we propose a Maximum-Weight Independent Set model to formulate the PR Module selection and clustering problem so that the original manual exploration can be solved efficiently and automatically. We also propose a step-wise adjustment configuration prefetching strategy incorporated in our model to generate optimized reconfiguration schedules. Our proposed approach not only supports various design constraints but also can consider multiple objectives such as area and reconfiguration delay. Experimental results show that our approach can optimize resource utilization and reduce reconfiguration delay with good scalability. Especially, the implementation of the real design case shows that our approach can be embedded in Xilinx's DPR design flow successfully.

This paper explores learning rich self-supervised entity representations from large amounts of associated text. Once pre-trained, these models become applicable to multiple entity-centric tasks such as ranked retrieval, knowledge base completion, question answering, and more. Unlike other methods that harvest self-supervision signals based merely on a local context within a sentence, we radically expand the notion of context to include any available text related to an entity. This enables a new class of powerful, high-capacity representations that can ultimately distill much of the useful information about an entity from multiple text sources, without any human supervision. We present several training strategies that, unlike prior approaches, learn to jointly predict words and entities – strategies we compare experimentally on downstream tasks in the TV-Movies domain, such as MovieLens tag prediction from user reviews and natural language movie search. As evidenced by results, our models match or outperform competitive baselines, sometimes with little or no fine-tuning, and are also able to scale to very large corpora. Finally, we make our datasets and pre-trained models publicly available. This includes Reviews2Movielens, mapping the ~1B word corpus of Amazon movie reviews (He and McAuley, 2016) to MovieLens tags (Harper and Konstan, 2016), as well as Reddit Movie Suggestions with natural language queries and corresponding community recommendations.

Modeling the complex interactions between users and items is at the core of designing successful recommender systems. One key task consists of predicting users’ personalized sequential behavior, where the challenge mainly lies in modeling ‘third-order’ interactions between a user, her previously visited item(s), and the next item to consume. In this paper, we propose a unified method, TransRec, to model such interactions for largescale sequential prediction. Methodologically, we embed items into a ‘transition space’ where users are modeled as translation vectors operating on item sequences. Empirically, this approach outperforms the state-of-the-art on a wide spectrum of real-world datasets.

Recent advancements in deep neural networks for graph-structured data have led to state-of-the-art performance on recommender system benchmarks. However, making these methods practical and scalable to web-scale recommendation tasks with billions of items and hundreds of millions of users remains an unsolved challenge. Here we describe a large-scale deep recommendation engine that we developed and deployed at Pinterest. We develop a data-efficient Graph Convolutional Network (GCN) algorithm, which combines efficient random walks and graph convolutions to generate embeddings of nodes (i.e., items) that incorporate both graph structure as well as node feature information. Compared to prior GCN approaches, we develop a novel method based on highly efficient random walks to structure the convolutions and design a novel training strategy that relies on harder-and-harder training examples to improve robustness and convergence of the model. We also develop an efficient MapReduce model inference algorithm to generate embeddings using a trained model. Overall, we can train on and embed graphs that are four orders of magnitude larger than typical GCN implementations. We show how GCN embeddings can be used to make high-quality recommendations in various settings at Pinterest, which has a massive underlying graph with 3 billion nodes representing pins and boards, and 17 billion edges. According to offline metrics, user studies, as well as A/B tests, our approach generates higher-quality recommendations than comparable deep learning based systems. To our knowledge, this is by far the largest application of deep graph embeddings to date and paves the way for a new generation of web-scale recommender systems based on graph convolutional architectures.

Dealing with sparse, long-tailed datasets, and cold-start problems is always a challenge for recommender systems. These issues can partly be dealt with by making predictions not in isolation, but by leveraging information from related events; such information could include signals from social relationships or from the sequence of recent activities. Both types of additional information can be used to improve the performance of state-of-the-art matrix factorization-based techniques. In this paper, we propose new methods to combine both social and sequential information simultaneously, in order to further improve recommendation performance. We show these techniques to be particularly effective when dealing with sparsity and cold-start issues in several large, real-world datasets.

Modeling the complex interactions between users and items as well as amongst items themselves is at the core of designing successful recommender systems. One classical setting is predicting users' personalized sequential behavior (or 'next-item' recommendation), where the challenges mainly lie in modeling 'third-order' interactions between a user, her previously visited item(s), and the next item to consume. Existing methods typically decompose these higher-order interactions into a combination of pairwise relationships, by way of which user preferences (user-item interactions) and sequential patterns (item-item interactions) are captured by separate components. In this paper, we propose a unified method, TransRec, to model such third-order relationships for large-scale sequential prediction. Methodologically, we embed items into a 'transition space' where users are modeled as translation vectors operating on item sequences. Empirically, this approach outperforms the state-of-the-art on a wide spectrum of real-world datasets. Data and code are available at https://sites.google.com/a/eng.ucsd.edu/ruining-he/.

Inverted list compression is a topic that has been studied for 50 years due to its fundamental importance in numerous applications including information retrieval, databases, and graph analytics. Typically, an inverted list compression algorithm is evaluated on its space overhead and query processing time. Earlier list compression designs mainly focused on minimizing the space overhead to reduce expensive disk I/O time in disk-oriented systems. But the recent trend is shifted towards reducing query processing time because the underlying systems tend to be memory-resident. Although there are many highly optimized compression approaches in main memory, there is still a considerable performance gap between query processing over compressed lists and uncompressed lists, which motivates this work. In this work, we set out to bridge this performance gap for the first time by proposing a new compression scheme, namely, MILC (memory inverted list compression). MILC relies on a series of techniques including offset-oriented fixed-bit encoding, dynamic partitioning, in-block compression, cache-aware optimization, and SIMD acceleration. We conduct experiments on three real-world datasets in information retrieval, databases, and graph analytics to demonstrate the high performance and low space overhead of MILC. We compare MILC with 12 recent compression algorithms and experimentally show that MILC improves the query performance by up to 13.2× and reduces the space overhead by up to 4.7×.

Modern recommender systems model people and items by discovering or `teasing apart' the underlying dimensions that encode the properties of items and users' preferences toward them. Critically, such dimensions are uncovered based on user feedback, often in implicit form (such as purchase histories, browsing logs, etc.); in addition, some recommender systems make use of side information, such as product attributes, temporal information, or review text.However one important feature that is typically ignored by existing personalized recommendation and ranking methods is the visual appearance of the items being considered. In this paper we propose a scalable factorization model to incorporate visual signals into predictors of people's opinions, which we apply to a selection of large, real-world datasets. We make use of visual features extracted from product images using (pre-trained) deep networks, on top of which we learn an additional layer that uncovers the visual dimensions that best explain the variation in people's feedback. This not only leads to significantly more accurate personalized ranking methods, but also helps to alleviate cold start issues, and qualitatively to analyze the visual dimensions that influence people's opinions.

Predicting personalized sequential behavior is a key task for recommender systems. In order to predict user actions such as the next product to purchase, movie to watch, or place to visit, it is essential to take into account both long-term user preferences and sequential patterns (i.e., short-term dynamics). Matrix Factorization and Markov Chain methods have emerged as two separate but powerful paradigms for modeling the two respectively. Combining these ideas has led to unified methods that accommodate long-and short-term dynamics simultaneously by modeling pairwise user-item and item-item interactions. In spite of the success of such methods for tackling dense data, they are challenged by sparsity issues, which are prevalent in real-world datasets. In recent years, similarity-based methods have been proposed for (sequentially-unaware) item recommendation with promising results on sparse datasets. In this paper, we propose to fuse such methods with Markov Chains to make personalized sequential recommendations. We evaluate our method, Fossil, on a variety of large, real-world datasets. We show quantitatively that Fossil outperforms alternative algorithms, especially on sparse datasets, and qualitatively that it captures personalized dynamics and is able to make meaningful recommendations.

Identifying relationships between items is a key task of an online recommender system, in order to help users discover items that are functionally complementary or visually compatible. In domains like clothing recommendation, this task is particularly challenging since a successful system should be capable of handling a large corpus of items, a huge amount of relationships among them, as well as the high-dimensional and semantically complicated features involved. Furthermore, the human notion of "compatibility" to capture goes beyond mere similarity: For two items to be compatible-whether jeans and a t-shirt, or a laptop and a charger-they should be similar in some ways, but systematically different in others. In this paper we propose a novel method, Monomer, to learn complicated and heterogeneous relationships between items in product recommendation settings. Recently, scalable methods have been developed that address this task by learning similarity metrics on top of the content of the products involved. Here our method relaxes the metricity assumption inherent in previous work and models multiple localized notions of 'relatedness,' so as to uncover ways in which related items should be systematically similar, and systematically different. Quantitatively, we show that our system achieves state-of-the-art performance on large-scale compatibility prediction tasks, especially in cases where there is substantial heterogeneity between related items.

Understanding users' interactions with highly subjective content---like artistic images---is challenging due to the complex semantics that guide our preferences. On the one hand one has to overcome `standard' recommender systems challenges, such as dealing with large, sparse, and long-tailed datasets. On the other, several new challenges present themselves, such as the need to model content in terms of its visual appearance, or even social dynamics, such as a preference toward a particular artist that is independent of the art they create. In this paper we build large-scale recommender systems to model the dynamics of a vibrant digital art community, Behance, consisting of tens of millions of interactions (clicks and 'appreciates') of users toward digital art. Methodologically, our main contributions are to model (a) rich content, especially in terms of its visual appearance; (b) temporal dynamics, in terms of how users prefer 'visually consistent' content within and across sessions; and (c) social dynamics, in terms of how users exhibit preferences both towards certain art styles, as well as the artists themselves.

Building a successful recommender system depends on understanding both the dimensions of people's preferences as well as their dynamics. In certain domains, such as fashion, modeling such preferences can be incredibly difficult, due to the need to simultaneously model the visual appearance of products as well as their evolution over time. The subtle semantics and non-linear dynamics of fashion evolution raise unique challenges especially considering the sparsity and large scale of the underlying datasets. In this paper we build novel models for the One-Class Collaborative Filtering setting, where our goal is to estimate users' fashion-aware personalized ranking functions based on their past feedback. To uncover the complex and evolving visual factors that people consider when evaluating products, our method combines high-level visual features extracted from a deep convolutional neural network, users' past feedback, as well as evolving trends within the community. Experimentally we evaluate our method on two large real-world datasets from Amazon.com, where we show it to outperform state-of-the-art personalized ranking measures, and also use it to visualize the high-level fashion trends across the 11-year span of our dataset.

Binary translation makes it convenient to emulate one instruction set by another. Nowadays, it is growing in popularity in various applications, especially the embedded platforms. When it comes to the test of binary translators, traditional methodologies which still mainly rely on manual unit test is costly, labor intensive and often not adequate to test complicated algorithms in the translators. Some standard benchmark suites, like SPEC CPU2006, are compiled with different compilation options for further tests. However, the translation modules still have over 30% of their code unexecuted after such tests, according to our experimental results. Methodologies based on randomization can generate a vast variety of tests, thus improve the code coverage in the translation system. In this paper, we propose such an approach named EATBit. Test binaries are generated with randomly selected instructions and operands. The binaries and a large amount of input data are then refined to exclude invalid ones. Experimental results on a real binary translator demonstrate that EATBit can not only improve code coverage by over 20%, but also find some new bugs in the translator successfully.

Dynamic Partial Reconfiguration (DPR) on FPGAs has attracted significant research interests in recent years since it provides benefits such as reduced area and flexible functionality. However, due to the lack of supporting synthesis tools in current DPR design flow, leveraging these benefits requires specific designer expertise with laborious manual design effort. Considering the complicated concurrency relations among functions, it is challenging to properly select Partial Reconfiguration Modules (PR Modules) and partition them into groups so that the hardware modules can be swapped in and out during the run time. What's more, the design of PR Modules also impacts reconfiguration latency and resource utilization greatly. In this paper, we formulate the PR Module generation problem into a standard Maximum-Weight Independent Set Problem (MWISP) so that the original manual exploration can be solved optimally and automatically. Our proposed algorithm not only supports various design constraints, but also has the ability to consider multiple objectives such as area and reconfiguration delay. Experimental results show that our approach can optimize resource utilization and reduce reconfiguration delay with good scalability. Especially, the implementation of the real design case shows that our approach can be embedded in the Xilinx's DPR design flow successfully and it can save around 70% reconfiguration latency overhead compared with the heuristic PR Module generation approaches.