Current benchmarks evaluate the performance of RAG methods from various perspectives, they share a common assumption that user queries used for retrieval are error-free. However, in real-world interactions between users and LLMs, query entry errors are frequent. The impact of these errors on current RAG methods against such errors remains largely unexplored. To bridge this gap, we propose QE-RAG, the first robust RAG benchmark designed specifically to evaluate performance against query entry errors. We analyze the impact of these errors on LLM outputs and find that corrupted queries degrade model performance, which can be mitigated through query correction and training a robust retriever for retrieving relevant documents. Based on these insights, we propose a contrastive learning-based robust retriever training method and a retrieval-augmented query correction method. Extensive experiments reveal that: (1) state-of-the-art RAG methods including sequential, branching, and iterative methods, exhibit poor robustness to query entry errors; (2) our method enhances the robustness of RAG when handling query entry errors and it's compatible with existing RAG methods, further improving their robustness.

Click-Through Rate (CTR) prediction, which estimates the probability of a user clicking on a given item, is a critical task for online information services. Existing approaches often make strong assumptions that training and test data come from the same distribution. However, the data distribution varies since user interests are constantly evolving, resulting in the Out-of-Distribution (OOD) issue. In addition, users tend to have multiple interests, some of which evolve faster than others. Toward this end, we propose Disentangled Click-Through Rate Prediction (DiseCTR), which introduces a causal perspective of recommendation and disentangles multiple aspects of user interests to alleviate the OOD issue in recommendation. We conduct a causal factorization of CTR prediction involving user interest, exposure model, and click model, based on which we develop a deep learning implementation for these three causal mechanisms. Specifically, we first design an interest encoder with sparse attention which maps raw features to user interests and then introduce a weakly supervised interest disentangler to learn independent interest embeddings, which are further integrated by an attentive interest aggregator for prediction. Experimental results on three real-world datasets show that DiseCTR achieves the best accuracy and robustness in OOD recommendation against state-of-the-art approaches, significantly improving AUC and GAUC by over 0.02 and reducing logloss by over 13.7%. Further analyses demonstrate that DiseCTR successfully disentangles user interests, which is the key to OOD generalization for CTR prediction. We have released the code and data at https://github.com/DavyMorgan/DiseCTR/ .

Video question answering plays a vital role in computer vision, and recent advances in large language models have further propelled the development of this field. However, existing video question answering techniques often face limitations in grasping fine-grained video content in spatial dimensions. It mainly stems from the fixed and low-resolution input of video frames. While some approaches using high-resolution inputs partially alleviate this problem, they introduce excessive computational burdens by encoding the entire high-resolution image. In this work, we propose a granularity-adaptive spatial evidence tokenization model for video question answering. Our method introduces multi-granular visual tokenization in the spatial dimension to produce video tokens at various granularities based on the question. It highlights spatially activated patches at low resolutions through a granularity weighting module and then adaptively encodes these activated patches at high resolution for detail supplementation. To mitigate the computational overhead associated with high-resolution frame encoding, a masking and acceleration module is developed for efficient visual tokenization. Moreover, a granularity compression module is designed to dynamically select and compress visual tokens of varying granularities based on questions. We conduct extensive experiments on 11 mainstream video question answering datasets and the experimental results demonstrate the effectiveness of our proposed method.

In search scenarios, user experience can be hindered by erroneous queries due to typos, voice errors, or knowledge gaps. Therefore, query correction is crucial for search engines. Current correction models, usually small models trained on specific data, often struggle with queries beyond their training scope or those requiring contextual understanding. While the advent of Large Language Models (LLMs) offers a potential solution, they are still limited by their pre-training data and inference cost, particularly for complex queries, making them not always effective for query correction. To tackle these, we propose Trigger3, a large-small model collaboration framework that integrates the traditional correction model and LLM for query correction, capable of adaptively choosing the appropriate correction method based on the query and the correction results from the traditional correction model and LLM. Trigger3 first employs a correction trigger to filter out correct queries. Incorrect queries are then corrected by the traditional correction model. If this fails, an LLM trigger is activated to call the LLM for correction. Finally, for queries that no model can correct, a fallback trigger decides to return the original query. Extensive experiments demonstrate Trigger3 outperforms correction baselines while maintaining efficiency.

Jingcheng Deng, Zhongtao Jiang, Liang Pang, Zihao Wei, Liwei Chen, Kun Xu, Yang Song, Huawei Shen, Xueqi Cheng. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Video generation requires modeling a vast spatiotemporal space, which demands significant computational resources and data usage. To reduce the complexity, the prevailing approaches employ a cascaded architecture to avoid direct training with full resolution latent. Despite reducing computational demands, the separate optimization of each sub-stage hinders knowledge sharing and sacrifices flexibility. This work introduces a unified pyramidal flow matching algorithm. It reinterprets the original denoising trajectory as a series of pyramid stages, where only the final stage operates at the full resolution, thereby enabling more efficient video generative modeling. Through our sophisticated design, the flows of different pyramid stages can be interlinked to maintain continuity. Moreover, we craft autoregressive video generation with a temporal pyramid to compress the full-resolution history. The entire framework can be optimized in an end-to-end manner and with a single unified Diffusion Transformer (DiT). Extensive experiments demonstrate that our method supports generating high-quality 5-second (up to 10-second) videos at 768p resolution and 24 FPS within 20.7k A100 GPU training hours. All code and models are open-sourced at https://pyramid-flow.github.io.

Retrieval aims to find the top-k items most relevant to a query/user from a large dataset. Traditional retrieval models represent queries/users and items as embedding vectors and use Approximate Nearest Neighbor (ANN) search for retrieval. Recently, researchers have proposed a generative-based retrieval method that represents items as token sequences and uses a decoder model for autoregressive training. Compared to traditional methods, this approach uses more complex models and integrates index structure during training, leading to better performance. However, these methods remain two-stage processes, where index construction is separate from the retrieval model, limiting the model's overall capacity. Additionally, existing methods construct indices by clustering pre-trained item representations in Euclidean space. However, real-world scenarios are more complex, making this approach less accurate. To address these issues, we propose a \underline{U}nified framework for \underline{R}etrieval and \underline{I}ndexing, termed \textbf{URI}. URI ensures strong consistency between index construction and the retrieval model, typically a Transformer decoder. URI simultaneously builds the index and trains the decoder, constructing the index through the decoder itself. It no longer relies on one-sided item representations in Euclidean space but constructs the index within the interactive space between queries and items. Experimental comparisons on three real-world datasets show that URI significantly outperforms existing methods.

Industrial recommendation systems (RS) rely on the multi-stage pipeline to balance effectiveness and efficiency when delivering items from a vast corpus to users. Existing RS benchmark datasets primarily focus on the exposure space, where novel RS algorithms are trained and evaluated. However, when these algorithms transition to real-world industrial RS, they face two critical challenges: (1) handling unexposed items—a significantly larger space than the exposed one, profoundly impacting their practical performance; and (2) overlooking the intricate interplay between multiple stages of the recommendation pipeline, resulting in suboptimal system performance. To bridge the gap between offline RS benchmarks and real-world online environments, we introduce RecFlow—an industrial full-flow recommendation dataset. Unlike existing datasets, RecFlow includes samples not only from the exposure space but also from unexposed items filtered at each stage of the RS funnel. RecFlow comprises 38 million interactions from 42,000 users across nearly 9 million items with additional 1.9 billion stage samples collected from 9.3 million online requests over 37 days and spanning 6 stages. Leveraging RecFlow, we conduct extensive experiments to demonstrate its potential in designing novel algorithms that enhance effectiveness by incorporating stage-specific samples. Some of these algorithms have already been deployed online at KuaiShou, consistently yielding significant gains. We propose RecFlow as the first comprehensive whole-pipeline benchmark dataset for the RS community, enabling research on algorithm design across the entire recommendation pipeline, including selection bias study, debiased algorithms, multi-stage consistency and optimality, multi-task recommendation, and user behavior modeling.

Retrieval-Augmented Generation (RAG) models are designed to incorporate external knowledge, reducing hallucinations caused by insufficient parametric (internal) knowledge. However, even with accurate and relevant retrieved content, RAG models can still produce hallucinations by generating outputs that conflict with the retrieved information. Detecting such hallucinations requires disentangling how Large Language Models (LLMs) balance external and parametric knowledge. Current detection methods often focus on one of these mechanisms or without decoupling their intertwined effects, making accurate detection difficult. In this paper, we investigate the internal mechanisms behind hallucinations in RAG scenarios. We discover hallucinations occur when the **Knowledge FFNs** in LLMs overemphasize parametric knowledge in the residual stream, while **Copying Heads** fail to effectively retain or integrate external knowledge from retrieved content. Based on these findings, we propose **ReDeEP**, a novel method that detects hallucinations by decoupling LLM’s utilization of external context and parametric knowledge. Our experiments show that ReDeEP significantly improves RAG hallucination detection accuracy. Additionally, we introduce AARF, which mitigates hallucinations by modulating the contributions of Knowledge FFNs and Copying Heads.

Recently, the personalization of Large Language Models (LLMs) to generate content that aligns with individual user preferences has garnered widespread attention. Personalized Retrieval-Augmented Generation (RAG), which retrieves relevant documents from the user's history to reflect their preferences and enhance LLM generation, is one commonly used approach for personalization. However, existing personalized RAG methods do not consider that the histories of similar users can also assist in personalized generation for the current user, meaning that collaborative information between users can also benefit personalized generation. Inspired by the application of collaborative filtering in recommender systems, we propose a method called CFRAG, which adapts Collaborative Filtering to RAG for personalized text generation. However, this presents two challenges: (1) how to incorporate collaborative information without explicit user similarity labels? (2) how to retrieve documents that support personalized LLM generation? For Challenge 1, we use contrastive learning to train user embeddings to retrieve similar users and introduce collaborative information. For Challenge 2, we design a personalized retriever and reranker to retrieve the top-k documents from these users' histories. We take into account the user's preference during retrieval and reranking. Then we leverage feedback from the LLM to fine-tune the personalized retriever and reranker, enabling them to retrieve documents that meet the personalized generation needs of the LLM. Experimental results on the Language Model Personalization (LaMP) benchmark validate the effectiveness of CFRAG. Further analysis confirms the importance of incorporating collaborative information.

Retrieval-Augmented Generation (RAG) systems for Large Language Models (LLMs) have shown promise in knowledge-intensive tasks, yet their reasoning capabilities, particularly for complex multi-step reasoning, remain limited. Although recent approaches have explored integrating RAG with chain-of-thought reasoning or incorporating test-time search with process reward model (PRM), these methods face several untrustworthy challenges, including lack of explanations, bias in PRM training data, early-step bias in PRM scores, and ignoring post-training that fails to fully optimize reasoning potential. To address these issues, we propose Retrieval-Augmented Reasoning through Trustworthy Process Rewarding (ReARTeR), a framework that enhances RAG systems' reasoning capabilities through both post-training and test-time scaling. At test time, ReARTeR introduces Trustworthy Process Rewarding via a Process Reward Model for accurate scalar scoring and a Process Explanation Model (PEM) for generating natural language explanations, enabling step refinement. During post-training, we leverage Monte Carlo Tree Search guided by Trustworthy Process Rewarding to collect high-quality step-level preference data, which is used to optimize the model through Iterative Preference Optimization. ReARTeR tackles three key challenges: (1) misalignment between PRM and PEM, addressed through off-policy preference learning; (2) bias in PRM training data, mitigated by a balanced annotation method and incorporating stronger annotations for difficult examples; and (3) early-step bias in PRM, resolved via a temporal-difference-based look-ahead search strategy. Experimental results on multi-step reasoning benchmarks demonstrate that ReARTeR significantly improves reasoning performance, highlighting its potential to advance the reasoning capability of RAG systems.

Sequential Recommendation (SR) plays a pivotal role in recommender systems by tailoring recommendations to user preferences based on their non-stationary historical interactions. Achieving high-quality performance in SR requires attention to both item representation and diversity. However, designing an SR method that simultaneously optimizes these merits remains a long-standing challenge. In this study, we address this issue by integrating recent generative Diffusion Models (DM) into SR. DM has demonstrated utility in representation learning and diverse image generation. Nevertheless, a straightforward combination of SR and DM leads to sub-optimal performance due to discrepancies in learning objectives (recommendation vs. noise reconstruction) and the respective learning spaces (non-stationary vs. stationary). To overcome this, we propose a novel framework called DimeRec (Di ffusion with multi-interest enhanced Rec ommender). DimeRec synergistically combines a guidance extraction module (GEM) and a generative diffusion aggregation module (DAM). The GEM extracts crucial stationary guidance signals from the user's non-stationary interaction history, while the DAM employs a generative diffusion process conditioned on GEM's outputs to reconstruct and generate consistent recommendations. Our numerical experiments demonstrate that DimeRec significantly outperforms established baseline methods across three publicly available datasets. Furthermore, we have successfully deployed DimeRec on a large-scale short video recommendation platform, serving hundreds of millions of users. Live A/B testing confirms that our method improves both users' time spent and result diversification.

Recent research on query generation has focused on using Large Language Models (LLMs), which, despite achieving state-of-the-art performance, also introduce hallucination issues in generated queries. In this work, we categorize these issues into relevance hallucination and factuality hallucination, proposing a new typology for hallucinations arising from LLM-based query generation. We present an effective approach to decouple content from form in LLM-generated queries, preserving the factual knowledge extracted and integrated from inputs while leveraging the LLM's linguistic capabilities to construct syntactic structures, including function words. Specifically, we introduce a model-agnostic and training-free method that transforms the Large Language Model into a Pointer-Generator (LargePiG), where the pointer attention distribution utilizes the LLM's inherent attention weights, and the copy probability is derived from the difference between the vocabulary distribution in the model's high layers and the last layer. To validate the effectiveness of LargePiG, we constructed two datasets for assessing hallucination issues in query generation, covering both document and video scenarios. Empirical studies on various LLMs demonstrated LargePiG's superiority across both datasets. Additional experiments further verified that LargePiG reduces hallucination in large vision-language models and enhances the accuracy of document-based question-answering and factuality evaluation tasks. The source code and dataset are available at https://github.com/Jeryi-Sun/LargePiG.

Recommender systems mitigate information overload by offering personalized suggestions to users. As the interactions between users and items can inherently be depicted as a bipartite graph, recent years have seen a surge in the interest in harnessing graph neural networks (GNNs) for enhancing recommender systems. However, a notable limitation of current GNN-based recommendation models is their exclusive emphasis on positive feedback, while disregarding the potentially insightful negative feedback. In this article, we introduce Pone-GNN, a novel recommendation model that synergistically integrates both Po sitive and ne gative feedback in G raph N eural N etworks. Pone-GNN utilizes dual embeddings–interest embeddings for capturing a user’s likes and disinterest embeddings for a user’s dislikes. Also, Pone-GNN performs message passing for both positive and negative feedback, and integrates two sets of embeddings through contrastive learning, which is crucial for extracting robust and discriminative embeddings. Our comprehensive experimental analysis demonstrates that Pone-GNN outperforms state-of-the-art models on diverse real-world recommendation datasets. For example, Pone-GNN achieves a 6.15% increase in relative nDCG@10 compared to the runner-up on the KuaiRec dataset.

In large-scale recommendation systems, modeling long-term user interests is progressively gaining attention among researchers and practitioners. Existing work, such as SIM and TWIN, typically employs a two-stage approach to model long-term user behavior sequences for efficiency concerns. The first stage rapidly retrieves a subset of sequences related to the target item from a long sequence using a search-based mechanism namely the General Search Unit (GSU), while the second stage calculates the interest scores using the Exact Search Unit (ESU) on the retrieved results. Given the extensive length of user behavior sequences spanning the entire life cycle, potentially reaching up to 10^6 in scale, there is currently no effective solution for fully modeling such expansive user interests. To overcome this issue, we introduced TWIN-V2, an enhancement of TWIN, where a divide-and-conquer approach is applied to compress life-cycle behaviors and uncover more accurate and diverse user interests. Specifically, a hierarchical clustering method groups items with similar characteristics in life-cycle behaviors into a single cluster during the offline phase. By limiting the size of clusters, we can compress behavior sequences well beyond the magnitude of 10^5 to a length manageable for online inference in GSU retrieval. Cluster-aware target attention extracts comprehensive and multi-faceted long-term interests of users, thereby making the final recommendation results more accurate and diverse. Extensive offline experiments on a multi-billion-scale industrial dataset and online A/B tests have demonstrated the effectiveness of TWIN-V2. Under an efficient deployment framework, TWIN-V2 has been successfully deployed to the primary traffic that serves hundreds of millions of daily active users at Kuaishou.

Recent advancements in Large Language Models (LLMs) have attracted considerable interest among researchers to leverage these models to enhance Recommender Systems (RSs). Existing work predominantly utilizes LLMs to generate knowledge-rich texts or utilizes LLM-derived embeddings as features to improve RSs. Although the extensive world knowledge embedded in LLMs generally benefits RSs, the application can only take a limited number of users and items as inputs, without adequately exploiting collaborative filtering information. Considering its crucial role in RSs, one key challenge in enhancing RSs with LLMs lies in providing better collaborative filtering information through LLMs. In this paper, drawing inspiration from the in-context learning and chain of thought reasoning in LLMs, we propose the Large Language Models enhanced Collaborative Filtering (LLM-CF) framework, which distills the world knowledge and reasoning capabilities of LLMs into collaborative filtering. We also explored a concise and efficient instruction-tuning method, which improves the recommendation capabilities of LLMs while preserving their general functionalities (e.g., not decreasing on the LLM benchmark). Comprehensive experiments on three real-world datasets demonstrate that LLM-CF significantly enhances several backbone recommendation models and consistently outperforms competitive baselines, showcasing its effectiveness in distilling the world knowledge and reasoning capabilities of LLM into collaborative filtering.

In real-world recommender systems, user engagement and subjective feedback play pivotal roles in shaping the content distribution mechanism of the platform. When platforms reach a certain scale, they often gather valuable questionnaire feedback data from users to evaluate their satisfaction with recommended items. Compared to traditional user feedback such as likes, questionnaires explicitly capture both satisfaction and dissatisfaction and are unaffected by other users' questionnaires, thus better expressing users' true preferences. In this paper, we aim to leverage the questionnaire feedback to align the recommendation model with users' true preferences. However, due to the platform distribution mechanism and divergent user attitudes toward questionnaires, the questionnaire feedback data frequently becomes sparse and exhibits selection biases, resulting in challenges in feature integration and training process. To address these issues, we introduce a novel user Satisfaction Alignment framework that effectively leverages Questionnaire feedback to enhance Recommendation, named SAQRec. SAQRec begins by training an unbiased satisfaction model to impute satisfaction, addressing selection bias and data sparsity. Then, SAQRec aligns features with users' true preferences by disentangling satisfaction and dissatisfaction from click history and categorizing clicked items into multiple satisfaction levels through the imputed satisfactions. Additionally, the imputed satisfactions from the pre-trained unbiased satisfaction model serve as pseudo-labels to align the model's outputs with users' true preferences. Extensive experiments on both public and commercial datasets demonstrate SAQRec's superior integration of questionnaire feedback in recommendation models. Online A/B testing on a short video platform confirms its effectiveness in boosting user watch time and positive-to-negative feedback ratio, enhancing overall performance and user satisfaction.

Contrastive learning, a powerful self-supervised learning paradigm, has shown its efficacy in learning embed dings from independent and identically distributed (IID) as well as non-IID data without relying on label information. Since high-quality discriminative embeddings form a rich embedding space, which benefits model performance on downstream tasks, it is necessary to study how to improve the quality of contrastive node embeddings in graph contrastive learning. However, there has been limited research on this area. In this paper, we investigate how to generate high-quality contrastive node embeddings based on an in-depth analysis of graph contrastive losses. Firstly, we propose a novel and effective method, GLATE, for estimating the temperatures in three mainstream graph contrastive losses during the training phase. Secondly, we conduct the derivation of GLATE, and the derivation results reveal the specific relationship between the quality of contrastive node embeddings and tem-peratures. Finally, the extensive experiments on 16 benchmark datasets demonstrate that GLATE consistently outperforms the state-of-the-art graph contrastive learning models in terms of both model performance and training efficiency.

In light of recent advances in multimodal Large Language Models (LLMs), there is increasing attention to scaling them from image-text data to more informative real-world videos. Compared to static images, video poses unique challenges for effective large-scale pre-training due to the modeling of its spatiotemporal dynamics. In this paper, we address such limitations in video-language pre-training with an efficient video decomposition that represents each video as keyframes and temporal motions. These are then adapted to an LLM using well-designed tokenizers that discretize visual and temporal information as a few tokens, thus enabling unified generative pre-training of videos, images, and text. At inference, the generated tokens from the LLM are carefully recovered to the original continuous pixel space to create various video content. Our proposed framework is both capable of comprehending and generating image and video content, as demonstrated by its competitive performance across 13 multimodal benchmarks in image and video understanding and generation. Our code and models are available at https://video-lavit.github.io.

Recommender systems require the simultaneous optimization of multiple objectives to accurately model user interests, necessitating the application of multi-task learning methods. However, existing multi-task learning methods in recommendations overlook the specific characteristics of recommendation scenarios, falling short in achieving proper gradient balance. To address this challenge, we set the target of multi-task learning as attaining the appropriate magnitude balance and the global direction balance, and propose an innovative methodology named GradCraft in response. GradCraft dynamically adjusts gradient magnitudes to align with the maximum gradient norm, mitigating interference from gradient magnitudes for subsequent manipulation. It then employs projections to eliminate gradient conflicts in directions while considering all conflicting tasks simultaneously, theoretically guaranteeing the global resolution of direction conflicts. GradCraft ensures the concurrent achievement of appropriate magnitude balance and global direction balance, aligning with the inherent characteristics of recommendation scenarios. Both offline and online experiments attest to the efficacy of GradCraft in enhancing multi-task performance in recommendations. The source code for GradCraft can be accessed at https://github.com/baiyimeng/GradCraft.

The diversity of recommendation is equally crucial as accuracy in improving user experience. Existing studies, e.g., Determinantal Point Process (DPP) and Maximal Marginal Relevance (MMR), employ a greedy paradigm to iteratively select items that optimize both accuracy and diversity. However, prior methods typically exhibit quadratic complexity, limiting their applications to the re-ranking stage and are not applicable to other recommendation stages with a larger pool of candidate items, such as the pre-ranking and ranking stages. In this paper, we propose Contextual Distillation Model (CDM), an efficient recommendation model that addresses diversification, suitable for the deployment in all stages of industrial recommendation pipelines. Specifically, CDM utilizes the candidate items in the same user request as context to enhance the diversification of the results. We propose a contrastive context encoder that employs attention mechanisms to model both positive and negative contexts. For the training of CDM, we compare each target item with its context embedding and utilize the knowledge distillation framework to learn the win probability of each target item under the MMR algorithm, where the teacher is derived from MMR outputs. During inference, ranking is performed through a linear combination of the recommendation and student model scores, ensuring both diversity and efficiency. We perform offline evaluations on two industrial datasets and conduct online A/B test of CDM on the short-video platform KuaiShou. The considerable enhancements observed in both recommendation quality and diversity, as shown by metrics, provide strong superiority for the effectiveness of CDM.

Scale-calibrated ranking systems are ubiquitous in real-world applications nowadays, which pursue accurate ranking quality and calibrated probabilistic predictions simultaneously.For instance, in the advertising ranking system, the predicted click-through rate (CTR) is utilized for ranking and required to be calibrated for the downstream cost-per-click ads bidding.Recently, multi-objective based methods have been wildly adopted as a standard approach for Calibrated Ranking, which incorporates the combination of two loss functions: a pointwise loss that focuses on calibrated absolute values and a ranking loss that emphasizes relative orderings.However, when applied to industrial online applications, existing multi-objective CR approaches still suffer from two crucial limitations.First, previous methods need to aggregate the full candidate list within a single mini-batch to compute the ranking loss.Such aggregation strategy violates extensive data shuffling which has long been proven beneficial for preventing overfitting, and thus degrades the training effectiveness.Second, existing multi-objective methods apply the two inherently conflicting loss functions on a single probabilistic prediction, which results in a sub-optimal trade-off between calibration and ranking.To tackle the two limitations, we propose a Self-Boosted framework for Calibrated Ranking (SBCR).In SBCR, the predicted ranking scores by the online deployed model are dumped into context features.With these additional context features, each single item can perceive the overall distribution of scores in the whole ranking list, so that the ranking loss can be constructed without the need for sample aggregation.As the deployed model is a few versions older than the training model, the dumped predictions reveal what was failed to learn and keep boosting the model to correct previously mis-predicted items.Moreover, a calibration module is introduced to decouple the point loss and ranking loss.The two losses are applied before and after the calibration module separately, which

Customizing diffusion models to generate identity-preserving images from user-provided reference images is an intriguing new problem. The prevalent approaches typically require training on extensive domain-specific images to achieve identity preservation, which lacks flexibility across different use cases. To address this issue, we exploit classifier guidance, a training-free technique that steers diffusion models using an existing classifier, for personalized image generation. Our study shows that based on a recent rectified flow framework, the major limitation of vanilla classifier guidance in requiring a special classifier can be resolved with a simple fixed-point solution, allowing flexible personalization with off-the-shelf image discriminators. Moreover, its solving procedure proves to be stable when anchored to a reference flow trajectory, with a convergence guarantee. The derived method is implemented on rectified flow with different off-the-shelf image discriminators, delivering advantageous personalization results for human faces, live subjects, and certain objects. Code is available at https://github.com/feifeiobama/RectifID.

Nowadays, many platforms provide users with both search and recommendation services as important tools for accessing information. The phenomenon has led to a correlation between user search and recommendation behaviors, providing an opportunity to model user interests in a fine-grained way. Existing approaches either model user search and recommendation behaviors separately or overlook the different transitions between user search and recommendation behaviors. In this paper, we propose a framework named UniSAR that effectively models the different types of fine-grained behavior transitions for providing users a Unified Search And Recommendation service. Specifically, UniSAR models the user transition behaviors between search and recommendation through three steps: extraction, alignment, and fusion, which are respectively implemented by transformers equipped with pre-defined masks, contrastive learning that aligns the extracted fine-grained user transitions, and cross-attentions that fuse different transitions. To provide users with a unified service, the learned representations are fed into the downstream search and recommendation models. Joint learning on both search and recommendation data is employed to utilize the knowledge and enhance each other. Experimental results on two public datasets demonstrated the effectiveness of UniSAR in terms of enhancing both search and recommendation simultaneously. The experimental analysis further validates that UniSAR enhances the results by successfully modeling the user transition behaviors between search and recommendation.

Incorporating Search and Recommendation (S&R) services within a singular application is prevalent in online platforms, leading to a new task termed open-app motivation prediction, which aims to predict whether users initiate the application with the specific intent of information searching, or to explore recommended content for entertainment. Studies have shown that predicting users' motivation to open an app can help to improve user engagement and enhance performance in various downstream tasks. However, accurately predicting open-app motivation is not trivial, as it is influenced by user-specific factors, search queries, clicked items, as well as their temporal occurrences. Furthermore, these activities occur sequentially and exhibit intricate temporal dependencies. Inspired by the success of the Neural Hawkes Process (NHP) in modeling temporal dependencies in sequences, this paper proposes a novel neural Hawkes process model to capture the temporal dependencies between historical user browsing and querying actions. The model, referred to as Neural Hawkes Process-based Open-App Motivation prediction model (NHP-OAM), employs a hierarchical transformer and a novel intensity function to encode multiple factors, and open-app motivation prediction layer to integrate time and user-specific information for predicting users' open-app motivations. To demonstrate the superiority of our NHP-OAM model and construct a benchmark for the Open-App Motivation Prediction task, we not only extend the public S&R dataset ZhihuRec but also construct a new real-world Open-App Motivation Dataset (OAMD). Experiments on these two datasets validate NHP-OAM's superiority over baseline models. Further downstream application experiments demonstrate NHP-OAM's effectiveness in predicting users' Open-App Motivation, highlighting the immense application value of NHP-OAM.

Short video recommendations often face limitations due to the quality of user feedback, which may not accurately depict user interests. To tackle this challenge, a new task has emerged: generating more dependable labels from original feedback. Existing label generation methods rely on manual rules, demanding substantial human effort and potentially misaligning with the desired objectives of the platform. To transcend these constraints, we introduce LabelCraft, a novel automated label generation method explicitly optimizing pivotal operational metrics for platform success. By formulating label generation as a higher-level optimization problem above recommender model optimization, LabelCraft introduces a trainable labeling model for automatic label mechanism modeling. Through meta-learning techniques, LabelCraft effectively addresses the bi-level optimization hurdle posed by the recommender and labeling models, enabling the automatic acquisition of intricate label generation mechanisms. Extensive experiments on real-world datasets corroborate LabelCraft's excellence across varied operational metrics, encompassing usage time, user engagement, and retention. Codes are available at https://github.com/baiyimeng/LabelCraft.

In modern recommender systems, sequential recommendation leverages chronological user behaviors to make effective next-item suggestions, which suffers from data sparsity issues, especially for new users. One promising line of work is the cross-domain recommendation, which trains models with data across multiple domains to improve the performance in data-scarce domains. Recent proposed cross-domain sequential recommendation models such as PiNet and DASL have a common drawback relying heavily on overlapped users in different domains, which limits their usage in practical recommender systems. In this paper, we propose a M ixed A ttention N etwork (MAN) with local and global attention modules to extract the domain-specific and cross-domain information. Firstly, we propose a local/global encoding layer to capture the domain-specific/cross-domain sequential pattern. Then we propose a mixed attention layer with item similarity attention, sequence-fusion attention, and group-prototype attention to capture the local/global item similarity, fuse the local/global item sequence, and extract the user groups across different domains, respectively. Finally, we propose a local/global prediction layer to further evolve and combine the domain-specific and cross-domain interests. Experimental results on two real-world datasets (each with two domains) demonstrate the superiority of our proposed model. Further study also illustrates that our proposed method and components are model-agnostic and effective, respectively. The code and data are available at https://github.com/Guanyu-Lin/MAN.

Modern personalized recommendation services often rely on user feedback, either explicit or implicit, to improve the quality of services. Explicit feedback refers to behaviors like ratings, while implicit feedback refers to behaviors like user clicks. However, in the scenario of full-screen video viewing experiences like Tiktok and Reels, the click action is absent, resulting in unclear feedback from users, hence introducing noises in modeling training. Existing approaches on de-noising recommendation mainly focus on positive instances while ignoring the noise in a large amount of sampled negative feedback. In this paper, we propose a meta-learning method to annotate the unlabeled data from loss and gradient perspectives, which considers the noises in both positive and negative instances. Specifically, we first propose anInverse Dual Loss (IDL) to boost the true label learning and prevent the false label learning. Then we further propose anInverse Gradient (IG) method to explore the correct updating gradient and adjust the updating based on meta-learning. Finally, we conduct extensive experiments on both benchmark and industrial datasets where our proposed method can significantly improve AUC by 9.25% against state-of-the-art methods. Further analysis verifies the proposed inverse learning framework is model-agnostic and can improve a variety of recommendation backbones. The source code, along with the best hyper-parameter settings, is available at this link: https://github.com/Guanyu-Lin/InverseLearning.

The Probability Ranking Principle (PRP) has been considered as the foundational standard in the design of information retrieval (IR) systems. The principle requires an IR module's returned list of results to be ranked with respect to the underlying user interests, so as to maximize the results' utility. Nevertheless, we point out that it is inappropriate to indiscriminately apply PRP through every stage of a contemporary IR system. Such systems contain multiple stages (e.g., retrieval, pre-ranking, ranking, and re-ranking stages, as examined in this paper). The selection bias inherent in the model of each stage significantly influences the results that are ultimately presented to users. To address this issue, we propose an improved ranking principle for multi-stage systems, namely the Generalized Probability Ranking Principle (GPRP), to emphasize both the selection bias in each stage of the system pipeline as well as the underlying interest of users. We realize GPRP via a unified algorithmic framework named Full Stage Learning to Rank. Our core idea is to first estimate the selection bias in the subsequent stages and then learn a ranking model that best complies with the downstream modules' selection bias so as to deliver its top ranked results to the final ranked list in the system's output. We performed extensive experiment evaluations of our developed Full Stage Learning to Rank solution, using both simulations and online A/B tests in one of the leading short-video recommendation platforms. The algorithm is proved to be effective in both retrieval and ranking stages. Since deployed, the algorithm has brought consistent and significant performance gain to the platform.

Increasing the size of embedding layers has shown to be effective in improving the performance of recommendation models, yet gradually causing their sizes to exceed terabytes in industrial recommender systems, and hence the increase of computing and storage costs. To save resources while maintaining model performances, we propose SHARK, the model compression practice we have summarized in the recommender system of industrial scenarios. SHARK consists of two main components. First, we use the novel first-order component of Taylor expansion as importance scores to prune the number of embedding tables (feature fields). Second, we introduce a new row-wise quantization method to apply different quantization strategies to each embedding. We conduct extensive experiments on both public and industrial datasets, demonstrating that each component of our proposed SHARK framework outperforms previous approaches. We conduct A/B tests in multiple models on Kuaishou, such as short video, e-commerce, and advertising recommendation models. The results of the online A/B test showed SHARK can effectively reduce the memory footprint of the embedded layer. For the short-video scenarios, the compressed model without any performance drop significantly saves 70% storage and thousands of machines, improves 30% queries per second (QPS), and has been deployed to serve hundreds of millions of users and process tens of billions of requests every day.

Short-video recommendation is one of the most important recommendation applications in today's industrial information systems. Compared with other recommendation tasks, the enormous amount of feedback is the most typical characteristic. Specifically, in short-video recommendation, the easiest-to-collect user feedback is theskipping behavior, which leads to two critical challenges for the recommendation model. First, the skipping behavior reflects implicit user preferences, and thus, it is challenging for interest extraction. Second, this kind of special feedback involves multiple objectives, such as total watching time and skipping rate, which is also very challenging. In this paper, we present our industrial solution in Kuaishou1, which serves billion-level users every day. Specifically, we deploy a feedback-aware encoding module that extracts user preferences, taking the impact of context into consideration. We further design a multi-objective prediction module which well distinguishes the relation and differences among different model objectives in the short-video recommendation. We conduct extensive online A/B tests, along with detailed and careful analysis, which verify the effectiveness of our solution.

The confluence of Search and Recommendation (S&R) services is vital to online services, including e-commerce and video platforms. The integration of S&R modeling is a highly intuitive approach adopted by industry practitioners. However, there is a noticeable lack of research conducted in this area within academia, primarily due to the absence of publicly available datasets. Consequently, a substantial gap has emerged between academia and industry regarding research endeavors in joint optimization using user behavior data from both S&R services. To bridge this gap, we introduce the first large-scale, real-world dataset KuaiSAR of integrated Search And Recommendation behaviors collected from Kuaishou, a leading short-video app in China with over 350 million daily active users. Previous research in this field has predominantly employed publicly available semi-synthetic datasets, with artificially fabricated search behaviors. Distinct from previous datasets, KuaiSAR contains genuine user behaviors, including the occurrence of each interaction within either search or recommendation service, and the users' transitions between the two services. This work aids in joint modeling of S&R, and utilizing search data for recommender systems (and recommendation data for search engines). Furthermore, due to the various feedback labels associated with user-video interactions, KuaiSAR also supports a broad range of tasks, including intent recommendation, multi-task learning, and modeling of long sequential multi-behavioral patterns. We believe this dataset will serve as a catalyst for innovative research and bridge the gap between academia and industry in understanding the S&R services in practical, real-world applications. The dataset is available at https://ethan00si.github.io/KuaiSAR/. The dataset is also shared at https://zenodo.org/record/8181109.

With the proliferation of short video applications, the significance of short video recommendations has vastly increased. Unlike other recommendation scenarios, short video recommendation systems heavily rely on feedback from watch time. Existing approaches simply treat watch time as a direct label, failing to effectively harness its extensive semantics and introduce bias, thereby limiting the potential for modeling user interests based on watch time. To overcome this challenge, we propose a framework named Debiased Multiple-semantics-extracting Labeling (DML). DML constructs labels that encompass various semantics by utilizing quantiles derived from the distribution of watch time, prioritizing relative order rather than absolute label values. This approach facilitates easier model learning while aligning with the ranking objective of recommendations. Furthermore, we introduce a method inspired by causal adjustment to refine label definitions, thereby directly mitigating bias at the label level. We substantiate the effectiveness of our DML framework through both online and offline experiments. Extensive results demonstrate that our DML could effectively leverage watch time to discover users' real interests, enhancing their engagement in our application.

Recommender systems are able to learn user preferences based on user and item representations via their historical behaviors. To improve representation learning, recent recommendation models start leveraging information from various behavior types exhibited by users. In real-world scenarios, the user behavioral graph is not only multiplex but also dynamic, i.e., the graph evolves rapidly over time, with various types of nodes and edges added or deleted, which causes the Neighborhood Disturbance. Nevertheless, most existing methods neglect such streaming dynamics and thus need to be retrained once the graph has significantly evolved, making them unsuitable in the online learning environment. Furthermore, the Neighborhood Disturbance existing in dynamic graphs deteriorates the performance of neighbor-aggregation based graph models. To this end, we propose SUPA, a novel graph neural network for dynamic multiplex heterogeneous graphs. Compared to neighbor-aggregation architecture, SUPA develops a sample-update-propagate architecture to alleviate neighborhood disturbance. Specifically, for each new edge, SUPA samples an influenced subgraph, updates the representations of the two interactive nodes, and propagates the interaction information to the sampled subgraph. Furthermore, to train SUPA incrementally online, we propose InsLearn, an efficient workflow for single-pass training of large dynamic graphs. Extensive experimental results on six real-world datasets show that SUPA has a good generalization ability and is superior to sixteen state-of-the-art baseline methods. The source code is available at https://github.com/shatter15/SUPA.

Life-long user behavior modeling, i.e., extracting a user's hidden interests from rich historical behaviors in months or even years, plays a central role in modern CTR prediction systems. Conventional algorithms mostly follow two cascading stages: a simple General Search Unit (GSU) for fast and coarse search over tens of thousands of long-term behaviors and an Exact Search Unit (ESU) for effective Target Attention (TA) over the small number of finalists from GSU. Although efficient, existing algorithms mostly suffer from a crucial limitation: the inconsistent target-behavior relevance metrics between GSU and ESU. As a result, their GSU usually misses highly relevant behaviors but retrieves ones considered irrelevant by ESU. In such case, the TA in ESU, no matter how attention is allocated, mostly deviates from the real user interests and thus degrades the overall CTR prediction accuracy. To address such inconsistency, we propose TWo-stage Interest Network (TWIN), where our Consistency-Preserved GSU (CP-GSU) adopts the identical target-behavior relevance metric as the TA in ESU, making the two stages twins. Specifically, to break TA's computational bottleneck and extend it from ESU to GSU, or namely from behavior length 102 to length 104 - 105, we build a novel attention mechanism by behavior feature splitting. For the video inherent features of a behavior, we calculate their linear projection by efficient pre-computing & caching strategies. And for the user-item cross features, we compress each into a one-dimentional bias term in the attention score calculation to save the computational cost. The consistency between two stages, together with the effective TA-based relevance metric in CP-GSU, contributes to significant performance gain in CTR prediction. Offline experiments on a 46 billion scale real production dataset from Kuaishou and an Online A/B test show that TWIN outperforms all compared SOTA algorithms. With optimized online infrastructure, we reduce the computational bottleneck by 99.3%, which contributes to the successful deployment of TWIN on Kuaishou, serving the main traffic of hundreds of millions of active users everyday.

With the increase of content pages and interactive buttons in online services such as online-shopping and video-watching websites, industrial-scale recommender systems face challenges in multi-domain and multi-task recommendations. The core of multi-task and multi-domain recommendation is to accurately capture user interests in multiple scenarios given multiple user behaviors. In this paper, we propose a plug-and-play Parameter and Embedding Personalized Network (PEPNet) for multi-domain and multi-task recommendation. PEPNet takes personalized prior information as input and dynamically scales the bottom-level Embedding and top-level DNN hidden units through gate mechanisms. Embedding Personalized Network (EPNet) performs personalized selection on Embedding to fuse features with different importance for different users in multiple domains. Parameter Personalized Network (PPNet) executes personalized modification on DNN parameters to balance targets with different sparsity for different users in multiple tasks. We have made a series of special engineering optimizations combining the Kuaishou training framework and the online deployment environment. By infusing personalized selection of Embedding and personalized modification of DNN parameters, PEPNet tailored to the interests of each individual obtains significant performance gains, with online improvements exceeding 1% in multiple task metrics across multiple domains. We have deployed PEPNet in Kuaishou apps, serving over 300 million users every day.

Graph Neural Networks (GNNs) have demonstrated promising results on exploiting node representations for many downstream tasks through supervised end-to-end training. To deal with the widespread label scarcity issue in real-world applications, Graph Contrastive Learning (GCL) is leveraged to train GNNs with limited or even no labels by maximizing the mutual information between nodes in its augmented views generated from the original graph. However, the distribution of graphs remains unconsidered in view generation, resulting in the ignorance of unseen edges in most existing literature, which is empirically shown to be able to improve GCL's performance in our experiments. To this end, we propose to incorporate graph generative adversarial networks (GANs) to learn the distribution of views for GCL, in order to i) automatically capture the characteristic of graphs for augmentations, and ii) jointly train the graph GAN model and the GCL model. Specifically, we present GACN, a novel Generative Adversarial Contrastive learning Network for graph representation learning. GACN develops a view generator and a view discriminator to generate augmented views automatically in an adversarial style. Then, GACN leverages these views to train a GNN encoder with two carefully designed self-supervised learning losses, including the graph contrastive loss and the Bayesian personalized ranking Loss. Furthermore, we design an optimization framework to train all GACN modules jointly. Extensive experiments on seven real-world datasets show that GACN is able to generate high-quality augmented views for GCL and is superior to twelve state-of-the-art baseline methods. Noticeably, our proposed GACN surprisingly discovers that the generated views in data augmentation finally conform to the well-known preferential attachment rule in online networks.

Sequential recommendation is one of the most important tasks in recommender systems, which aims to recommend the next interacted item with historical behaviors as input. Traditional sequential recommendation always mainly considers the collected positive feedback such as click, purchase, etc. However, in short-video platforms such as TikTok, video viewing behavior may not always represent positive feedback. Specifically, the videos are played automatically, and users passively receive the recommended videos. In this new scenario, users passively express negative feedback by skipping over videos they do not like, which provides valuable information about their preferences. Different from the negative feedback studied in traditional recommender systems, this passive-negative feedback can reflect users’ interests and serve as an important supervision signal in extracting users’ preferences. Therefore, it is essential to carefully design and utilize it in this novel recommendation scenario. In this work, we first conduct analyses based on a large-scale real-world short-video behavior dataset and illustrate the significance of leveraging passive feedback. We then propose a novel method that deploys the sub-interest encoder, which incorporates positive feedback and passive-negative feedback as supervision signals to learn the user’s current active sub-interest. Moreover, we introduce an adaptive fusion layer to integrate various sub-interests effectively. To enhance the robustness of our model, we then introduce a multi-task learning module to simultaneously optimize two kinds of feedback – passive-negative feedback and traditional randomly-sampled negative feedback. The experiments on two large-scale datasets verify that the proposed method can significantly outperform state-of-the-art approaches. The code is released at https://github.com/tsinghua-fib-lab/RecSys2023-SINE to benefit the community.

Modern online service providers such as online shopping platforms often provide both search and recommendation (S&R) services to meet different user needs. Rarely has there been any effective means of incorporating user behavior data from both S&R services. Most existing approaches either simply treat S&R behaviors separately, or jointly optimize them by aggregating data from both services, ignoring the fact that user intents in S&R can be distinctively different. In our paper, we propose a Search-Enhanced framework for the Sequential Recommendation (SESRec) that leverages users' search interests for recommendation, by disentangling similar and dissimilar representations within S&R behaviors. Specifically, SESRec first aligns query and item embeddings based on users' query-item interactions for the computations of their similarities. Two transformer encoders are used to learn the contextual representations of S&R behaviors independently. Then a contrastive learning task is designed to supervise the disentanglement of similar and dissimilar representations from behavior sequences of S&R. Finally, we extract user interests by the attention mechanism from three perspectives, i.e., the contextual representations, the two separated behaviors containing similar and dissimilar interests. Extensive experiments on both industrial and public datasets demonstrate that SESRec consistently outperforms state-of-the-art models. Empirical studies further validate that SESRec successfully disentangle similar and dissimilar user interests from their S&R behaviors.

Modern recommender systems often deal with a variety of user interactions, e.g., click, forward, purchase, etc., which requires the underlying recommender engines to fully understand and leverage multi-behavior data from users. Despite recent efforts towards making use of heterogeneous data, multi-behavior recommendation still faces great challenges. Firstly, sparse target signals and noisy auxiliary interactions remain an issue. Secondly, existing methods utilizing self-supervised learning (SSL) to tackle the data sparsity neglect the serious optimization imbalance between the SSL task and the target task. Hence, we propose a Multi-Behavior Self-Supervised Learning (MBSSL) framework together with an adaptive optimization method. Specifically, we devise a behavior-aware graph neural network incorporating the self-attention mechanism to capture behavior multiplicity and dependencies. To increase the robustness to data sparsity under the target behavior and noisy interactions from auxiliary behaviors, we propose a novel self-supervised learning paradigm to conduct node self-discrimination at both inter-behavior and intra-behavior levels. In addition, we develop a customized optimization strategy through hybrid manipulation on gradients to adaptively balance the self-supervised learning task and the main supervised recommendation task. Extensive experiments on five real-world datasets demonstrate the consistent improvements obtained by MBSSL over ten state-of-the-art (SOTA) baselines. We release our model implementation at: https://github.com/Scofield666/MBSSL.git.

Accurate user interest modeling is vital for recommendation scenarios. One of the effective solutions is the sequential recommendation that relies on click behaviors, but this is not elegant in the video feed recommendation where users are passive in receiving the streaming contents and return skip or no-skip behaviors. Here skip and no-skip behaviors can be treated as negative and positive feedback, respectively. With the mixture of positive and negative feedback, it is challenging to capture the transition pattern of behavioral sequence. To do so, FeedRec has exploited a shared vanilla Transformer, which may be inelegant because head interaction of multi-heads attention does not consider different types of feedback. In this paper, we propose Dual-interest Factorization-heads Attention for Sequential Recommendation (short for DFAR) consisting of feedback-aware encoding layer, dual-interest disentangling layer and prediction layer. In the feedback-aware encoding layer, we first suppose each head of multi-heads attention can capture specific feedback relations. Then we further propose factorization-heads attention which can mask specific head interaction and inject feedback information so as to factorize the relation between different types of feedback. Additionally, we propose a dual-interest disentangling layer to decouple positive and negative interests before performing disentanglement on their representations. Finally, we evolve the positive and negative interests by corresponding towers whose outputs are contrastive by BPR loss. Experiments on two real-world datasets show the superiority of our proposed method against state-of-the-art baselines. Further ablation study and visualization also sustain its effectiveness. We release the source code here: https://github.com/tsinghua-fib-lab/WWW2023-DFAR.

Recommender systems are currently widely used in various applications helping people filter information. Existing models always embed the rich information for recommendation, such as items, users, and contexts in real-value vectors, and make predictions based on these vectors. In the view of causal inference, the associations between representation vectors and user feedback are inevitably a mixture of the causal part that describes why a user prefers an item, and the non-causal part that merely reflects the statistical dependencies, for example, the display ranking position and sales promotion. However, most recommender systems assume the user-item interactions are only affected by user preferences, neglecting the striking differences between these two associations. To address this problem, we propose a model-agnostic causal learning framework called IV4Rec+ that can effectively decompose the embedding vectors into these two parts. Moreover, two strategies are proposed to utilize search queries as instrumental variables: IV4Rec+(I) only decomposes the item embeddings, while IV4Rec+(UI) decomposes both user and item embeddings. IV4Rec+ is a model-agnostic design that can be applied to many existing recommender systems, e.g., DIN, NRHUB, and SRGNN. Extensive experiments on three datasets show that IV4Rec+ significantly facilitates the performance of recommender systems and outperforms state-of-the-art frameworks.

Customer Life Time Value (LTV) is the expected total revenue that a single user can bring to a business. It is widely used in a variety of business scenarios to make operational decisions when acquiring new customers. Modeling LTV is a challenging problem, due to its complex and mutable data distribution. Existing approaches either directly learn from posterior feature distributions or leverage statistical models that make strong assumption on prior distributions, both of which fail to capture those mutable distributions. In this paper, we propose a complete set of industrial-level LTV modeling solutions. Specifically, we introduce an Order Dependency Monotonic Network (ODMN) that models the ordered dependencies between LTVs of different time spans, which greatly improves model performance. We further introduce a Multi Distribution Multi Experts (MDME) module based on the Divide-and-Conquer idea, which transforms the severely imbalanced distribution modeling problem into a series of relatively balanced sub-distribution modeling problems hence greatly reduces the modeling complexity. In addition, a novel evaluation metric Mutual Gini is introduced to better measure the distribution difference between the estimated value and the ground-truth label based on the Lorenz Curve. The ODMN framework has been successfully deployed in many business scenarios of Kuaishou, and achieved great performance. Extensive experiments on real-world industrial data demonstrate the superiority of the proposed methods compared to state-of-the-art baselines including ZILN and Two-Stage XGBoost models.

Recently, graph neural networks have shown the superiority of modeling the complex topological structures in heterogeneous network-based recommender systems. Due to the diverse interactions among nodes and abundant semantics emerging from diverse types of nodes and edges, there is a bursting research interest in learning expressive node repre-sentations in multiplex heterogeneous networks. One of the most important tasks in recommender systems is to predict the potential connection between two nodes under a specific edge type (i.e., relationship). Although existing studies utilize explicit metapaths to aggregate neighbors, practically they only consider intra-relationship metapaths and thus fail to leverage the potential uplift by inter-relationship information. Moreover, it is not always straightforward to exploit inter-relationship metapaths comprehensively under diverse relationships, espe-cially with the increasing number of node and edge types. In addition, contributions of different relationships between two nodes are difficult to measure. To address the challenges, we propose HybridGNN, an end-to-end GNN model with hybrid aggregation flows and hierarchical attentions to fully utilize the heterogeneity in the multiplex scenarios. Specifically, HybridGNN applies a randomized inter-relationship exploration module to exploit the multiplexity property among different relationships. Then, our model leverages hybrid aggregation flows under intra-relationship metapaths and randomized exploration to learn the rich semantics. To explore the importance of different aggregation flow and take advantage of the multiplexity property, we bring forward a novel hierarchical attention module which leverages both metapath-Ievel attention and relationship-level attention. Extensive experimental results on five real-world datasets suggest that HybridGNN achieves the best performance compared to several state-of-the-art baselines (p < 0.01, t-test) with statistical significance.

There has been a growing interest in developing conversational recommendation system (CRS), which provides valuable recommendations to users through conversations. Compared to the traditional recommendation, it advocates wealthier interactions and provides possibilities to obtain users’ exact preferences explicitly. Nevertheless, the corresponding research on this topic is limited due to the lack of broad-coverage dialogue corpus, especially real-world dialogue corpus. To handle this issue and facilitate our exploration, we construct E-ConvRec, an authentic Chinese dialogue dataset consisting of over 25k dialogues and 770k utterances, which contains user profile, product knowledge base (KB), and multiple sequential real conversations between users and recommenders. Next, we explore conversational recommendation in a real scene from multiple facets based on the dataset. Therefore, we particularly design three tasks: user preference recognition, dialogue management, and personalized recommendation. In the light of the three tasks, we establish baseline results on E-ConvRec to facilitate future studies.

Multimodal named entity recognition (MNER) is a vision-language task where the system is required to detect entity spans and corresponding entity types given a sentence-image pair. Existing methods capture text-image relations with various attention mechanisms that only obtain implicit alignments between entity types and image regions. To locate regions more accurately and better model cross-/within-modal relations, we propose a machine reading comprehension based framework for MNER, namely MRC-MNER. By utilizing queries in MRC, our framework can provide prior information about entity types and image regions. Specifically, we design two stages, Query-Guided Visual Grounding and Multi-Level Modal Interaction, to align fine-grained type-region information and simulate text-image/inner-text interactions respectively. For the former, we train a visual grounding model via transfer learning to extract region candidates that can be further integrated into the second stage to enhance token representations. For the latter, we design text-image and inner-text interaction modules along with three sub-tasks for MRC-MNER. To verify the effectiveness of our model, we conduct extensive experiments on two public MNER datasets, Twitter2015 and Twitter2017. Experimental results show that MRC-MNER outperforms the current state-of-the-art models on Twitter2017, and yields competitive results on Twitter2015.

Sequential recommendation aims at identifying the next item that is preferred by a user based on their behavioral history. Compared to conventional sequential models that leverage attention mechanisms and RNNs, recent efforts mainly follow two directions for improvement: multi-interest learning and graph convolutional aggregation. Specifically, multi-interest methods such as ComiRec and MIMN, focus on extracting different interests for a user by performing historical item clustering, while graph convolution methods including TGSRec and SURGE elect to refine user preferences based on multilevel correlations between historical items. Unfortunately, neither of them realizes that these two types of solutions can mutually complement each other, by aggregating multi-level user preference to achieve more precise multi-interest extraction for a better recommendation. To this end, in this paper, we propose a unified multi-grained neural model (named MGNM) via a combination of multi-interest learning and graph convolutional aggregation. Concretely, MGNM first learns the graph structure and information aggregation paths of the historical items for a user. It then performs graph convolution to derive item representations in an iterative fashion, in which the complex preferences at different levels can be well captured. Afterwards, a novel sequential capsule network is proposed to inject the sequential patterns into the multi-interest extraction process, leading to a more precise interest learning in a multi-grained manner. Experiments on three real-world datasets from different scenarios demonstrate the superiority of MGNM against several state-of-the-art baselines. The performance gain over the best baseline is up to 27.10% and 25.17% in terms of [email protected] and [email protected] respectively, which is one of the largest gains in recent development of sequential recommendation. Further analysis also demonstrates that MGNM is robust and effective at user preference understanding at multi-grained levels.

Machine-learning based recommender system(RS) has become an effective means to help people automatically discover their interests. Existing models often represent the rich information for recommendation, such as items, users, and contexts, as embedding vectors and leverage them to predict users’ feedback. In the view of causal analysis, the associations between these embedding vectors and users’ feedback are a mixture of the causal part that describes why an item is preferred by a user, and the non-causal part that merely reflects the statistical dependencies between users and items, for example, the exposure mechanism, public opinions, display position, etc. However, existing RSs mostly ignored the striking differences between the causal parts and non-causal parts when using these embedding vectors. In this paper, we propose a model-agnostic framework named IV4Rec that can effectively decompose the embedding vectors into these two parts, hence enhancing recommendation results. Specifically, we jointly consider users’ behaviors in search scenarios and recommendation scenarios. Adopting the concepts in causal analysis, we embed users’ search behaviors as instrumental variables (IVs), to help decompose original embedding vectors in recommendation, i.e., treatments. IV4Rec then combines the two parts through deep neural networks and uses the combined results for recommendation. IV4Rec is model-agnostic and can be applied to a number of existing RSs such as DIN and NRHUB. Experimental results on both public and proprietary industrial datasets demonstrate that IV4Rec consistently enhances RSs and outperforms a framework that jointly considers search and recommendation.

Modeling user’s long-term and short-term interests is crucial for accurate recommendation. However, since there is no manually annotated label for user interests, existing approaches always follow the paradigm of entangling these two aspects, which may lead to inferior recommendation accuracy and interpretability. In this paper, to address it, we propose a Contrastive learning framework to disentangle Long and Short-term interests for Recommendation (CLSR) with self-supervision. Specifically, we first propose two separate encoders to independently capture user interests of different time scales. We then extract long-term and short-term interests proxies from the interaction sequences, which serve as pseudo labels for user interests. Then pairwise contrastive tasks are designed to supervise the similarity between interest representations and their corresponding interest proxies. Finally, since the importance of long-term and short-term interests is dynamically changing, we propose to adaptively aggregate them through an attention-based network for prediction. We conduct experiments on two large-scale real-world datasets for e-commerce and short-video recommendation. Empirical results show that our CLSR consistently outperforms all state-of-the-art models with significant improvements: GAUC is improved by over 0.01, and NDCG is improved by over 4%. Further counterfactual evaluations demonstrate that stronger disentanglement of long and short-term interests is successfully achieved by CLSR. The code and data are available at https://github.com/tsinghua-fib-lab/CLSR.

Xiaoxue Zang, Lijuan Liu, Maria Wang, Yang Song, Hao Zhang, Jindong Chen. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Recently, micro-video sharing platforms such as Kuaishou and Tiktok have become a major source of information for people's lives. Thanks to the large traffic volume, short video lifespan and streaming fashion of these services, it has become more and more pressing to improve the existing recommender systems to accommodate these challenges in a cost-effective way. In this paper, we propose a novel concept-aware denoising graph neural network (named Conde) for micro-video recommendation. Conde consists of a three-phase graph convolution process to derive user and micro-video representations: warm-up propagation, graph denoising and preference refinement. A heterogeneous tripartite graph is constructed by connecting user nodes with video nodes, and video nodes with associated concept nodes, extracted from captions and comments of the videos. To address the noisy information in the graph, we introduce a user-oriented graph denoising phase to extract a subgraph which can better reflect the user's preference. Despite the main focus of micro-video recommendation in this paper, we also show that our method can be generalized to other types of tasks. Therefore, we also conduct empirical studies on a well-known public E-commerce dataset. The experimental results suggest that the proposed Conde achieves significantly better recommendation performance than the existing state-of-the-art solutions.

Sequential recommendation aims to leverage users' historical behaviors to predict their next interaction. Existing works have not yet addressed two main challenges in sequential recommendation. First, user behaviors in their rich historical sequences are often implicit and noisy preference signals, they cannot sufficiently reflect users' actual preferences. In addition, users' dynamic preferences often change rapidly over time, and hence it is difficult to capture user patterns in their historical sequences. In this work, we propose a graph neural network model called SURGE (short forSeqUential Recommendation with Graph neural nEtworks) to address these two issues. Specifically, SURGE integrates different types of preferences in long-term user behaviors into clusters in the graph by re-constructing loose item sequences into tight item-item interest graphs based on metric learning. This helps explicitly distinguish users' core interests, by forming dense clusters in the interest graph. Then, we perform cluster-aware and query-aware graph convolutional propagation and graph pooling on the constructed graph. It dynamically fuses and extracts users' current activated core interests from noisy user behavior sequences. We conduct extensive experiments on both public and proprietary industrial datasets. Experimental results demonstrate significant performance gains of our proposed method compared to state-of-the-art methods. Further studies on sequence length confirm that our method can model long behavioral sequences effectively and efficiently.

Continuously discovering novel entities in news and Web data is important for Knowledge Base (KB) maintenance. One of the key challenges is to decide whether an entity mention refers to an in-KB or out-of-KB entity. We propose a principled approach that learns a novel entity classifier by modeling mention and entity representation into multiple feature spaces, including contextual, topical, lexical, neural embedding and query spaces. Different from most previous studies that address novel entity discovery as a submodule of entity linking systems, our model is more a generalized approach and can be applied as a pre-filtering step of novel entities for any entity linking systems. Experiments on three real-world datasets show that our method significantly outperforms existing methods on identifying novel entities.

Recently, proactive systems such as Google Now and Microsoft Cortana have become increasingly popular in reforming the way users access information on mobile devices. In these systems, relevant content is presented to users based on their context without a query in the form of information cards that do not require a click to satisfy the users. As a result, prior approaches based on clicks cannot provide reliable measurements of user satisfaction with such systems. It is also unclear how much of the previous findings regarding good abandonment with reactive Web searches can be applied to these proactive systems due to the intrinsic difference in user intent, the greater variety of content types and their presentations. In this paper, we present the first large-scale analysis of viewing behavior based on the viewport (the visible fraction of a Web page) of the mobile devices, towards measuring user satisfaction with the information cards of the mobile proactive systems. In particular, we identified and analyzed a variety of factors that may influence the viewing behavior, including biases from ranking positions, the types and attributes of the information cards, and the touch interactions with the mobile devices. We show that by modeling the various factors we can better measure user satisfaction with the mobile proactive systems, enabling stronger statistical power in large-scale online A/B testing.

Modeling temporal behavior in recommendation systems is an important and challenging problem. Its challenges come from the fact that temporal modeling increases the cost of parameter estimation and inference, while requiring large amount of data to reliably learn the model with the additional time dimensions. Therefore, it is often difficult to model temporal behavior in large-scale real-world recommendation systems. In this work, we propose a novel deep neural network based architecture that models the combination of long-term static and short-term temporal user preferences to improve the recommendation performance. To train the model efficiently for large-scale applications, we propose a novel pre-train method to reduce the number of free parameters significantly. The resulted model is applied to a real-world data set from a commercial News recommendation system. We compare to a set of established baselines and the experimental results show that our method outperforms the state-of-the-art significantly.

Web search has been a reactive scenario for decades which often starts by users issuing queries. By studying the user behavior in search engine logs, we have discovered that many of the search tasks such as stock-price checking, news reading exhibit strong repeated patterns from day to day. In addition, users exhibit even stronger repetition on mobile devices. This provides us chances to perform proactive recommendations without user issuing queries. In this work, we aim at discovering and characterizing these types of tasks so that we can automatically predict when and what types of tasks will be repeated by the users in the future, through analyzing search logs from a commercial Web search engine and user interaction logs from a mobile App that offers proactive recommendations. We first introduce a set of novel features that can accurately capture task repetition. We then propose a novel deep learning framework that learns user preferences and makes automatic predictions. Our framework is capable of learning both user-independent global models as well as catering personalized models via model adaptation. The model we developed significantly outperforms other state-of-the-art predictive models by large margins. We also demonstrate the power of our model and features through an application to improve the recommendation quality of the mobile App. Results indicate a significant relevance improvement over the current production system.

In modern web-scale applications that collect data from different sources, entity conflation is a challenging task due to various data quality issues. In this paper, we propose a robust and distributed framework to perform conflation on noisy data in the Microsoft Academic Service dataset. Our framework contains two major components. In the offline component, we train a GBDT model to determine whether two papers from different sources should be conflated to the same paper entity. In the online component, we propose a scalable shingling algorithm that can apply our offline model to over 100 million instances. The result shows that our algorithm can conflate noisy data robustly and efficiently.

Recent online services rely heavily on automatic personalization to recommend relevant content to a large number of users. This requires systems to scale promptly to accommodate the stream of new users visiting the online services for the first time. In this work, we propose a content-based recommendation system to address both the recommendation quality and the system scalability. We propose to use a rich feature set to represent users, according to their web browsing history and search queries. We use a Deep Learning approach to map users and items to a latent space where the similarity between users and their preferred items is maximized. We extend the model to jointly learn from features of items from different domains and user features by introducing a multi-view Deep Learning model. We show how to make this rich-feature based user representation scalable by reducing the dimension of the inputs and the amount of training data. The rich user feature representation allows the model to learn relevant user behavior patterns and give useful recommendations for users who do not have any interaction with the service, given that they have adequate search and browsing history. The combination of different domains into a single model for learning helps improve the recommendation quality across all the domains, as well as having a more compact and a semantically richer user latent feature vector. We experiment with our approach on three real-world recommendation systems acquired from different sources of Microsoft products: Windows Apps recommendation, News recommendation, and Movie/TV recommendation. Results indicate that our approach is significantly better than the state-of-the-art algorithms (up to 49% enhancement on existing users and 115% enhancement on new users). In addition, experiments on a publicly open data set also indicate the superiority of our method in comparison with transitional generative topic models, for modeling cross-domain recommender systems. Scalability analysis show that our multi-view DNN model can easily scale to encompass millions of users and billions of item entries. Experimental results also confirm that combining features from all domains produces much better performance than building separate models for each domain.

Web search queries without hyperlink clicks are often referred to as abandoned queries. Understanding the reasons for abandonment is crucial for search engines in evaluating their performance. Abandonment can be categorized as good or bad depending on whether user information needs are satisfied by result page content. Previous research has sought to understand abandonment rationales via user surveys, or has developed models to predict those rationales using behavioral patterns. However, these models ignore important contextual factors such as the relationship between the abandoned query and prior abandonment instances. We propose more advanced methods for modeling and predicting abandonment rationales using contextual information from user search sessions by analyzing search engine logs, and discover dependencies between abandoned queries and user behaviors. We leverage these dependency signals to build a sequential classifier using a structured learning framework designed to handle such signals. Our experimental results show that our approach is 22% more accurate than the state-of-the-art abandonment-rationale classifier. Going beyond prediction, we leverage the prediction results to significantly improve relevance using instances of predicted good and bad abandonment.

Search satisfaction is a property of a user's search process. Understanding it is critical for search providers to evaluate the performance and improve the effectiveness of search engines. Existing methods model search satisfaction holistically at the search-task level, ignoring important dependencies between action-level satisfaction and overall task satisfaction. We hypothesize that searchers' latent action-level satisfaction (i.e., whether they believe they were satisfied with the results of a query or click) influences their observed search behaviors and contributes to overall search satisfaction. We conjecture that by modeling search satisfaction at the action level, we can build more complete and more accurate predictors of search-task satisfaction. To do this, we develop a latent structural learning method, whereby rich structured features and dependency relations unique to search satisfaction prediction are explored. Using in-situ search satisfaction judgments provided by searchers, we show that there is significant value in modeling action-level satisfaction in search-task satisfaction prediction. In addition, experimental results on large-scale logs from Bing.com demonstrate clear benefit from using inferred action satisfaction labels for other applications such as document relevance estimation and query suggestion.

RankNet is one of the widely adopted ranking models for web search tasks. However, adapting a generic RankNet for personalized search is little studied. In this paper, we first continue-trained a variety of RankNets with different number of hidden layers and network structures over a previously trained global RankNet model, and observed that a deep neural network with five hidden layers gives the best performance. To further improve the performance of adaptation, we propose a set of novel methods categorized into two groups. In the first group, three methods are proposed to properly assess the usefulness of each adaptation instance and only leverage the most informative instances to adapt a user-specific RankNet model. These assessments are based on KL-divergence, click entropy or a heuristic to ignore top clicks in adaptation queries. In the second group, two methods are proposed to regularize the training of the neural network in RankNet: one of these methods regularize the error back-propagation via a truncated gradient approach, while the other method limits the depth of the back propagation when adapting the neural network. We empirically evaluate our approaches using a large-scale real-world data set. Experimental results exhibit that our methods all give significant improvements over a strong baseline ranking system, and the truncated gradient approach gives the best performance, significantly better than all others.

In this paper, we introduce “task trail” to understand user search behaviors. We define a task to be an atomic user information need, whereas a task trail represents all user activities within that particular task, such as query reformulations, URL clicks. Previously, web search logs have been studied mainly at session or query level where users may submit several queries within one task and handle several tasks within one session. Although previous studies have addressed the problem of task identification, little is known about the advantage of using task over session or query for search applications. In this paper, we conduct extensive analyses and comparisons to evaluate the effectiveness of task trails in several search applications: determining user satisfaction, predicting user search interests, and suggesting related queries. Experiments on large scale data sets of a commercial search engine show that: (1) Task trail performs better than session and query trails in determining user satisfaction; (2) Task trail increases webpage utilities of end users comparing to session and query trails; (3) Task trails are comparable to query trails but more sensitive than session trails in measuring different ranking functions; (4) Query terms from the same task are more topically consistent to each other than query terms from different tasks; (5) Query suggestion based on task trail is a good complement of query suggestions based on session trail and click-through bipartite. The findings in this paper verify the need of extracting task trails from web search logs and enhance applications in search and recommendation systems.

Search engines train and apply a single ranking model across all users, but searchers' information needs are diverse and cover a broad range of topics. Hence, a single user-independent ranking model is insufficient to satisfy different users' result preferences. Conventional personalization methods learn separate models of user interests and use those to re-rank the results from the generic model. Those methods require significant user history information to learn user preferences, have low coverage in the case of memory-based methods that learn direct associations between query-URL pairs, and have limited opportunity to markedly affect the ranking given that they only re-order top-ranked items.

In this paper, we present a log-based study on user search behavior comparisons on three different platforms: desktop, mobile and tablet. We use three-month search logs in 2012 from a commercial search engine for our study. Our objective is to better understand how and to what extent mobile and tablet searchers behave differently than desktop users. Our study spans a variety of aspects including query categorization, query length, search time distribution, search location distribution, user click patterns and so on. From our data set, we reveal that there are significant differences between user search patterns in these three platforms, and therefore use the same ranking system is not an optimal solution for all of them. Consequently, we propose a framework that leverages a set of domain-specific features, along with the training data from desktop search, to further improve the search relevance for mobile and tablet platforms. Experimental results demonstrate that by transferring knowledge from desktop search, search relevance on mobile and tablet can be greatly improved.

User engagement in search refers to the frequency for users (re-)using the search engine to accomplish their tasks. Among factors that affected users' visit frequency, relevance of search results is believed to play a pivotal role. While multiple work in the past has demonstrated the correlation between search success and user engagement based on longitudinal analysis, we examine this problem from a different perspective in this work. Specifically, we carefully designed a large-scale controlled experiment on users of a large commercial Web search engine, in which users were separated into control and treatment groups, where users in treatment group were presented with search results which are deliberate degraded in relevance. We studied users' responses to the relevance degradation through tracking several behavioral metrics (such as query per user, click per session) over an extended period of time both during and following the experiment. By quantifying the relationship between user engagement and search relevance, we observe significant differences between user's short-term search behavior and long-term engagement change. By leveraging some of the key findings from the experiment, we developed a machine learning model to predict the long term impact of relevance degradation on user engagement. Overall, our model achieves over 67% of accuracy in predicting user engagement drop. Besides, our model is also capable of predicting engagement change for low-frequency users with very few user signals. We believe that insights from this study can be leveraged by search engine companies to detect and intervene search relevance degradation and to prevent long term user engagement drop.

Search tasks, comprising a series of search queries serving the same information need, have recently been recognized as an accurate atomic unit for modeling user search intent. Most prior research in this area has focused on short-term search tasks within a single search session, and heavily depend on human annotations for supervised classification model learning. In this work, we target the identification of long-term, or cross-session, search tasks (transcending session boundaries) by investigating inter-query dependencies learned from users' searching behaviors. A semi-supervised clustering model is proposed based on the latent structural SVM framework, and a set of effective automatic annotation rules are proposed as weak supervision to release the burden of manual annotation. Experimental results based on a large-scale search log collected from Bing.com confirms the effectiveness of the proposed model in identifying cross-session search tasks and the utility of the introduced weak supervision signals. Our learned model enables a more comprehensive understanding of users' search behaviors via search logs and facilitates the development of dedicated search-engine support for long-term tasks.

Personalized search systems tailor search results to the current user intent using historic search interactions. This relies on being able to find pertinent information in that user's search history, which can be challenging for unseen queries and for new search scenarios. Building richer models of users' current and historic search tasks can help improve the likelihood of finding relevant content and enhance the relevance and coverage of personalization methods. The task-based approach can be applied to the current user's search history, or as we focus on here, all users' search histories as so-called "groupization" (a variant of personalization whereby other users' profiles can be used to personalize the search experience). We describe a method whereby we mine historic search-engine logs to find other users performing similar tasks to the current user and leverage their on-task behavior to identify Web pages to promote in the current ranking. We investigate the effectiveness of this approach versus query-based matching and finding related historic activity from the current user (i.e., group versus individual). As part of our studies we also explore the use of the on-task behavior of particular user cohorts, such as people who are expert in the topic currently being searched, rather than all other users. Our approach yields promising gains in retrieval performance, and has direct implications for improving personalization in search systems.

Query suggestion is an interactive approach for search engines to better understand users information need. In this paper, we propose a novel query suggestion framework which leverages user re-query feedbacks from search engine logs. Specifically, we mined user query reformulation activities where the user only modifies part of the query by (1) adding terms after the query, (2) deleting terms within the query, or (3) modifying terms to new terms. We build a term-transition graph based on the mined data. Two models are proposed which address topic-level and term-level query suggestions, respectively. In the first topic-based unsupervised Pagerank model, we perform random walk on each of the topic-based term-transition graph and calculate the Pagerank for each term within a topic. Given a new query, we suggest relevant queries based on its topic distribution and term-transition probability within each topic. Our second model resembles the supervised learning-to-rank (LTR) framework, in which term modifications are treated as documents so that each query reformulation is treated as a training instance. A rich set of features are constructed for each (query, document) pair from Pagerank, Wikipedia, N-gram, ODP and so on. This supervised model is capable of suggesting new queries on a term level which addresses the limitation of previous methods. Experiments are conducted on a large data set from a commercial search engine. By comparing the with state-of-the-art query suggestion methods [4, 2], our proposals exhibit significant performance increase for all categories of queries.

In this paper, we introduce "task trail" as a new concept to understand user search behaviors. We define task to be an atomic user information need. Web search logs have been studied mainly at session or query level where users may submit several queries within one task and handle several tasks within one session. Although previous studies have addressed the problem of task identification, little is known about the advantage of using task over session and query for search applications. In this paper, we conduct extensive analyses and comparisons to evaluate the effectiveness of task trails in three search applications: determining user satisfaction, predicting user search interests, and query suggestion. Experiments are conducted on large scale datasets from a commercial search engine. Experimental results show that: (1) Sessions and queries are not as precise as tasks in determining user satisfaction. (2) Task trails provide higher web page utilities to users than other sources. (3) Tasks represent atomic user information needs, and therefore can preserve topic similarity between query pairs. (4) Task-based query suggestion can provide complementary results to other models. The findings in this paper verify the need to extract task trails from web search logs and suggest potential applications in search and recommendation systems.

The overwhelming amount of various monitoring and log data generated in multitier IT systems makes problem determination one of the most expensive and labor-intensive tasks in IT Services arena. Particularly the initial step of problem classification is complicated by error propagation making secondary problems surfacing on multiple dependent resources. In this paper, we propose to automate the process of problem classification by leveraging machine learning. The main focus is to categorize the problem a user experiences by recognizing the real root cause specificity leveraging available training data such as monitoring and logs across the systems. We transform the structure of the problem into a hierarchy using an existing taxonomy. We then propose an efficient hierarchical incremental learning algorithm which is capable of adjusting its internal local classifier parameters in realtime. Comparing to the traditional batch learning algorithms, this online solution decreases the computational complexity of the training process by learning from new instances on an incremental fashion. Our approach significantly reduces the memory required to store the training instances. We demonstrate the efficiency of our approach by learning hierarchical problem patterns for several issues occurring in distributed web applications. Experimental results show that our approach substantially outperforms previous methods.

Understanding the behavior of satisfied and unsatisfied Web search users is very important for improving users search experience. Collecting labeled data that characterizes search behavior is a very challenging problem. Most of the previous work used a limited amount of data collected in lab studies or annotated by judges lacking information about the actual intent. In this work, we performed a large scale user study where we collected explicit judgments of user satisfaction with the entire search task. Results were analyzed using sequence models that incorporate user behavior to predict whether the user ended up being satisfied with a search or not. We test our metric on millions of queries collected from real Web search traffic and show empirically that user behavior models trained using explicit judgments of user satisfaction outperform several other search quality metrics. The proposed model can also be used to optimize different search engine components. We propose a method that uses task level success prediction to provide a better interpretation of clickthrough data. Clickthough data has been widely used to improve relevance estimation. We use our user satisfaction model to distinguish between clicks that lead to satisfaction and clicks that do not. We show that adding new features derived from this metric allowed us to improve the estimation of document relevance.

Social bookmarking sites typically visualize user-generated tags as tag clouds. While tag clouds effectively show the relative frequency and thus popularity of tags, they fail to convey two aspects to the users: (1) the similarity between tags, and (2) the abstractness of tags. We suggest an alternative to tag clouds known as tag hierarchies. Tag hierarchies are based on a minimum evolution-based greedy algorithm for tag hierarchy construction, which iteratively includes optimal tags into the tree that introduce minimum changes to the existing taxonomy. Our algorithm also uses a global tag ranking method to order tags according to their levels of abstractness as well as popularity such that more abstract tags will appear at higher levels in the taxonomy. Based on the tag hierarchy, we derive a new tag recommendation algorithm, which is a structure-based approach that does not require heavily trained models and thus is highly efficient. User studies and quantitative analysis suggest that (1) the tag hierarchy can potentially reduce the user's tagging time in comparison to tag clouds and other tag tree structures, and (2) the tag recommendation algorithm significantly outperforms existing content-based methods in quality.

Query suggestion refers to the process of suggesting related queries to search engine users. Most existing researches have focused on improving the relevance of suggested queries. In this paper, we introduce the concept of diversifying the content of the search results from suggested queries while keeping the suggestion relevant. Our framework first retrieves a set of query candidates from search engine logs using random walk and other techniques. We then re-rank the suggested queries by ranking them in the order which maximizes the diversification function that measures the difference between the original search results and the results from suggested queries. The diversification function we proposed includes features like ODP category, URL and domain similarity and so on. One important outcome from our research which contradicts with most existing researches is that, with the increase of suggestion relevance, the similarity between the queries actually decreases. Experiments are conducted on a large set of human-labeled data, which is randomly sampled from a commercial search engine's log. Results indicate that the post-ranking framework significantly improves the relevance of suggested queries by comparing to existing models.

In this paper, we propose a new framework for searchable web sites recommendation. Given a query, our system will recommend a list of searchable web sites ranked by relevance, which can be used to complement the web page results and ads from a search engine. We model the conditional probability of a searchable web site being relevant to a given query in term of three main components: the language model of the query, the language model of the content within the web site, and the reputation of the web site searching capability (static rank). The language models for queries and searchable sites are built using information mined from client-side browsing logs. The static rank for each searchable site leverages features extracted from these client-side logs such as number of queries that are submitted to this site, and features extracted from general search engines such as the number of web pages that indexed for this site, number of clicks per query, and the dwell-time that a user spends on the search result page and on the clicked result web pages. We also learn a weight for each kind of feature to optimize the ranking performance. In our experiment, we discover 10.5 thousand searchable sites and use 5 million unique queries, extracted from one week of log data to build and demonstrate the effectiveness of our searchable web site recommendation system.

The emergence of Web 2.0 and the consequent success of social network Web sites such as Del.icio.us and Flickr introduce us to a new concept called social bookmarking, or tagging. Tagging is the action of connecting a relevant user-defined keyword to a document, image, or video, which helps the user to better organize and share their collections of interesting stuff. With the rapid growth of Web 2.0, tagged data is becoming more and more abundant on the social network Web sites. An interesting problem is how to automate the process of making tag recommendations to users when a new resource becomes available. In this article, we address the issue of tag recommendation from a machine learning perspective. From our empirical observation of two large-scale datasets, we first argue that the user-centered approach for tag recommendation is not very effective in practice. Consequently, we propose two novel document-centered approaches that are capable of making effective and efficient tag recommendations in real scenarios. The first, graph-based, method represents the tagged data in two bipartite graphs, (document, tag) and (document, word), then finds document topics by leveraging graph partitioning algorithms. The second, prototype-based, method aims at finding the most representative documents within the data collections and advocates a sparse multiclass Gaussian process classifier for efficient document classification. For both methods, tags are ranked within each topic cluster/class by a novel ranking method. Recommendations are performed by first classifying a new document into one or more topic clusters/classes, and then selecting the most relevant tags from those clusters/classes as machine-recommended tags. Experiments on real-world data from Del.icio.us, CiteULike, and BibSonomy examine the quality of tag recommendation as well as the efficiency of our recommendation algorithms. The results suggest that our document-centered models can substantially improve the performance of tag recommendations when compared to the user-centered methods, as well as topic models LDA and SVM classifiers.

Query suggestion has been an effective approach to help users narrow down to the information they need. However, most of existing studies focused on only popular/head queries. Since rare queries possess much less information (e.g., clicks) than popular queries in the query logs, it is much more difficult to efficiently suggest relevant queries to a rare query. In this paper, we propose an optimal rare query suggestion framework by leveraging implicit feedbacks from users in the query logs. Our model resembles the principle of pseudo-relevance feedback which assumes that top-returned results by search engines are relevant. However, we argue that the clicked URLs and skipped URLs contain different levels of information and thus should be treated differently. Hence, our framework optimally combines both the click and skip information from users and uses a random walk model to optimize the query correlation. Our model specifically optimizes two parameters: (1) the restarting (jumping) rate of random walk, and (2) the combination ratio of click and skip information. Unlike the Rocchio algorithm, our learning process does not involve the content of the URLs but simply leverages the click and skip counts in the query-URL bipartite graphs. Consequently, our model is capable of scaling up to the need of commercial search engines. Experimental results on one-month query logs from a large commercial search engine with over 40 million rare queries demonstrate the superiority of our framework, with statistical significance, over the traditional random walk models and pseudo-relevance feedback models.

Problem determination and resolution PDR is the process of detecting anomalies in a monitored system, locating the problems responsible for the issue, determining the root cause and fixing the cause of the problem. The cost of PDR represents a substantial part of operational costs, and faster, more effective PDR can contribute to a substantial reduction in system administration costs. In this paper, we propose to automate the process of PDR by leveraging machine learning methods. The main focus is to effectively categorize the problem a user experiences by recognizing the problem specificity leveraging all available training data such like the performance data and the logs data. Specifically, we transform the structure of the problem into a hierarchy which can be determined by existing taxonomy in advance. We then propose an efficient hierarchical incremental learning algorithm which is capable of adjusting its internal local classifier parameters in real-time. Comparing to the traditional batch learning algorithms, this online learning framework can significantly decrease the computational complexity of the training process by learning from new instances on an incremental fashion. In the same time this reduces the amount of memory required to store the training instances. We demonstrate the efficiency of our approach by learning hierarchical problem patterns for several issues occurring in distributed web applications. Experimental results show that our approach substantially outperforms previous methods.

Abstract Email spam has become a major problem for Internet users and providers. One major obstacle to its eradication is that the potential solutions need to ensure a very low false‐positive rate, which tends to be difficult in practice. We address the problem of low‐FPR classification in the context of naive Bayes, which represents one of the most popular machine learning models applied in the spam filtering domain. Drawing from the recent extensions, we propose a new term weight aggregation function, which leads to markedly better results than the standard alternatives. We identify short instances as ones with disproportionally poor performance and counter this behavior with a collaborative filtering‐based feature augmentation. Finally, we propose a tree‐based classifier cascade for which decision thresholds of the leaf nodes are jointly optimized for the best overall performance. These improvements, both individually and in aggregate, lead to substantially better detection rate of precision when compared with some of the best variants of naive Bayes proposed to date. Copyright © 2009 John Wiley & Sons, Ltd.

Tagged data is rapidly becoming more available on theWorld Wide Web. Web sites which populate tagging services offer a good way for Internet users to share their knowledge. An interesting problem is how to make tag suggestions when a new resource becomes available. In this paper, we ad-dress the issue of efficient tag suggestion. We first propose a multi-class sparse Gaussian process classification framework (SGPS) which is capable of classifying data with very few training instances. We suggest a novel prototype selection algorithm to select the best subset of points for model learn-ing. The framework is then extended to a novel multi-class multi-label classification algorithm (MMSG) that transforms tag suggestion into the problem of multi-label ranking. Ex-periments on bench-mark data sets and real-world data from Del.icio.us and BibSonomy suggest that our model can greatly improve the performance of tag suggestions when compared to the state-of-the-art. Overall, our model requires linear time to train and constant time to predict per case. The memory consumption is also significantly less than tradi-tional batch learning algorithms such as SVMs. In addition, results on tagging digital data also demonstrate that our model is capable of recommending relevant tags to images and videos by using their surrounding textual information.

We introduce dynamic correlated topic models (DCTM) for analyzing discrete data over time. This model is inspired by the hierarchical Gaussian process latent variable models (GP-LVM). DCTM is essentially a non-linear dimension reduction technique which is capable of (1) detecting topic evolution within a document corpus, (2) discovering topic correlations between document corpora, and (3) monitoring topic and correlation trends dynamically. Unlike generative aspect models such like LDA, DCTM demonstrates a much faster converging rate with better model fitting to the data. We empirically assess our approach using 268,231 scientific documents, from the year 1988 to 2005. Posterior inferences suggest that DCTM is useful for capturing topic and correlation dynamics, as well as predicting their trends.

Two major challenges in collaborative filtering are the efficiency of the algorithms and the quality of the recommendations. A variety of machine learning methods have been applied to address these two issues, including feature selection, instance selection, and clustering. Most existing methods either compromise computational complexity or prediction precision. Two novel, scalable memory-based CF algorithms are proposed, namely BS1, BS2, which combine the strengths of existing techniques while discarding their weaknesses. Experiments show that both the efficiency and performance have been improved when compared to three classical techniques: VSIM, FCBF and PD.

Tags are user-generated labels for entities. Existing research on tag recommendation either focuses on improving its accuracy or on automating the process, while ignoring the efficiency issue. We propose a highly-automated novel framework for real-time tag recommendation. The tagged training documents are treated as triplets of (words, docs, tags), and represented in two bipartite graphs, which are partitioned into clusters by Spectral Recursive Embedding (SRE). Tags in each topical cluster are ranked by our novel ranking algorithm. A two-way Poisson Mixture Model (PMM) is proposed to model the document distribution into mixture components within each cluster and aggregate words into word clusters simultaneously. A new document is classified by the mixture model based on its posterior probabilities so that tags are recommended according to their ranks. Experiments on large-scale tagging datasets of scientific documents (CiteULike) and web pages del.icio.us) indicate that our framework is capable of making tag recommendation efficiently and effectively. The average tagging time for testing a document is around 1 second, with over 88% test documents correctly labeled with the top nine tags we suggested.

To improve existing social bookmarking systems and to design new ones, researchers and practitioners need to understand how to evaluate tagging behavior. In this paper, we analyze over two years of data from CiteULike, a social bookmarking system for tagging academic papers. We propose six tag metrics-tag growth, tag reuse, tag non-obviousness, tag discrimination, tag frequency, and tag patterns-to understand the characteristics of a social bookmarking system. Using these metrics, we suggest possible design heuristics to implement a social bookmarking system for CiteSeer, a popular online scholarly digital library for computer science. We believe that these metrics and design heuristics can be applied to social bookmarking systems in other domains.

We propose a novel multiclass classification algorithm Gentle Adaptive Multiclass Boosting Learning (GAMBLE). The algorithm naturally extends the two class Gentle AdaBoost algorithm to multiclass classification by using the multiclass exponential loss and the multiclass response encoding scheme. Unlike other multiclass algorithms which reduce the K-class classification task to K binary classifications, GAMBLE handles the task directly and symmetrically, with only one committee classifier. We formally derive the GAMBLE algorithm with the quasi-Newton method, and prove the structural equivalence of the two regression trees in each boosting step. To scale up to large datasets, we utilize the generalized Query By Committee (QBC) active learning framework to focus learning on the most informative samples. Our empirical results show that with QBC-style active sample selection, we can achieve faster training time and potentially higher classification accuracy. GAMBLE's numerical superiority, structural elegance and low computation complexity make it highly competitive with state-of-the-art multiclass classification algorithms.

Name ambiguity is a special case of identity uncertainty where one person can be referenced by multiple name variations in different situations or evenshare the same name with other people. In this paper, we present an efficient framework by using two novel topic-based models, extended from Probabilistic Latent Semantic Analysis (PLSA) and Latent Dirichlet Allocation (LDA). Our models explicitly introduce a new variable for persons and learn the distribution of topics with regard to persons and words. Experiments indicate that our approach consistently outperforms other unsupervised methods including spectral and DBSCAN clustering. Scalability is addressed by disambiguating authors in over 750,000 papers from the entire CiteSeer dataset.

The issue of seeking efficient and effective methods for classifying unstructured text in large document corpora has received much attention in recent years. Traditional document representation like bag-of-words encodes documents as feature vectors, which usually leads to sparse feature spaces with large dimensionality, thus making it hard to achieve high classification accuracies. This paper addresses the problem of classifying unstructured documents on the Web. A classification approach is proposed that utilizes traditional feature reduction techniques along with a collaborative filtering method for augmenting document feature spaces. The method produces feature spaces with an order of magnitude less features compared with a baseline bag-of-words feature selection method. Experiments on both real-world data and benchmark corpus indicate that our approach improves classification accuracy over the traditional methods for both support vector machines and AdaBoost classifiers.

Email logs people's communication history which provides valuable information regarding the infrastructure of an organization. In this paper, a two-phase framework is introduced to attack the problem of leadership discovery in an organization based on email communication history among the employees. Two heuristic metrics are proposed for evaluating pair-wise leadership factors among a group of employees. We also address several issues in discovering the organization's structure through mining leadership graph constructed from the leadership factors. Experimental studies are carried out by applying the framework to Enron email corpus.