Self-supervised learning (SSL) has shown great promise in graph representation learning. However, most existing graph SSL methods are developed and evaluated under a single-dataset setting, leaving their cross-dataset transferability largely unexplored and limiting their ability to leverage knowledge transfer and large-scale pretraining, factors that are critical for developing generalized intelligence beyond fitting training data. To address this gap and advance foundation model research for graphs, we present GSTBench, the first systematic benchmark for evaluating the transferability of graph SSL methods. We conduct large-scale pretraining on ogbn-papers100M and evaluate five representative SSL methods across a diverse set of target graphs. Our standardized experimental setup decouples confounding factors such as model architecture, dataset characteristics, and adaptation protocols, enabling rigorous comparisons focused solely on pretraining objectives. Surprisingly, we observe that most graph SSL methods struggle to generalize, with some performing worse than random initialization. In contrast, GraphMAE, a masked autoencoder approach, consistently improves transfer performance. We analyze the underlying factors that drive these differences and offer insights to guide future research on transferable graph SSL, laying a solid foundation for the ''pretrain-then-transfer'' paradigm in graph learning. Our code is available at https://github.com/SongYYYY/GSTBench.

In this paper, we systemically disentangle BHG, a graph deep learning framework for daily users' ads recommendations. BHG mainly relies on two pillars: (1) graph tokenization to convert the input temporal heterogeneous graph into sequences of tokens, and (2) graph MLP-Mixer neural architecture to learn node representations on sequences of tokens via a mini-batch manner. In general, BHG embraces three advantages: (1) flexibility, i.e., BHG can be seamlessly integrated with any existing industrial recommendation model by treating the learned node embeddings as additional features that encode interactions, (2) efficiency, i.e., the graph tokenization allows sampling the neighborhood both locally and globally, and reduces the number of nodes considered for aggregations, and (3) model simplicity, i.e., the graph MLP-Mixer does not require self-attention for aggregating nodes and hence enjoys the simplicity. We demonstrate the superior performance of the proposed BHG on two internal datasets and one public dataset. We hope this paper can share insights and explain large-scale graph deep learning deployments for researchers, engineers, and practitioners.

Link Prediction (LP) is a critical task in graph machine learning. While Graph Neural Networks (GNNs) have significantly advanced LP performance recently, existing methods face key challenges including limited supervision from sparse connectivity, sensitivity to initialization, and poor generalization under distribution shifts. We explore pretraining as a solution to address these challenges. Unlike node classification, LP is inherently a pairwise task, which requires the integration of both node- and edge-level information. In this work, we present the first systematic study on the transferability of these distinct modules and propose a late fusion strategy to effectively combine their outputs for improved performance. To handle the diversity of pretraining data and avoid negative transfer, we introduce a Mixture-of-Experts (MoE) framework that captures distinct patterns in separate experts, facilitating seamless application of the pretrained model on diverse downstream datasets. For fast adaptation, we develop a parameter-efficient tuning strategy that allows the pretrained model to adapt to unseen datasets with minimal computational overhead. Experiments on 16 datasets across two domains demonstrate the effectiveness of our approach, achieving state-of-the-art performance on low-resource link prediction while obtaining competitive results compared to end-to-end trained methods, with over 10,000x lower computational overhead.

Content-based recommendation systems play a crucial role in delivering personalized content to users in the digital world. In this work, we introduce EmbSum, a novel framework that enables offline pre-computations of users and candidate items while capturing the interactions within the user engagement history. By utilizing the pretrained encoder-decoder model and poly-attention layers, EmbSum derives User Poly-Embedding (UPE) and Content Poly-Embedding (CPE) to calculate relevance scores between users and candidate items. EmbSum actively learns the long user engagement histories by generating user-interest summary with supervision from large language model (LLM). The effectiveness of EmbSum is validated on two datasets from different domains, surpassing state-of-the-art (SoTA) methods with higher accuracy and fewer parameters. Additionally, the model’s ability to generate summaries of user interests serves as a valuable by-product, enhancing its usefulness for personalized content recommendations.

Conversational recommendation system (CRS) is able to obtain fine-grained and dynamic user preferences based on interactive dialogue. Previous CRS assumes that the user has a clear target item, which often deviates from the real scenario. The user may have a clear single preference for some attribute types (e.g., brand) of items, while for other attribute types (e.g., color), the user may have multiple preferences or even no clear preferences, which leads to multiple acceptable items under multiple combinations of attribute instances. Furthermore, previous works assume that users would provide clear responses to any questions asked by the system. And, they also assume that users would be dedicated to the target item, that is, user would answer “yes” to the attribute corresponding to the target item and answer “no” to other attributes. However, users’ responses to attributes are not completely dependent on target items, but also influenced by users’ inherent interests. Besides, for some over-specific or equivocal questions, the feedback of user might not be clear (“yes”/“no”) and user might give some fuzzy response like “I don’t know”. To address the aforementioned issues, we first propose a more realistic conversational recommendation learning setting, namely Multi-Interest Multi-round Conversational Recommendation (MIMCR), where users may have multiple interests in attribute instance combinations and accept multiple items with partially overlapped combinations of attribute instances. To effectively cope with MIMCR, we propose a novel learning framework, namely Multiple Choice questions based on Multi-Interest Policy Learning. Moreover, we further propose a more realistic User-centric User Simulator with Fuzzy Feedback (UUSFF), which naturally calibrates the user response with additional fuzzy feedback based on user’s inherent preference. To better match the new scenario UUSFF, we propose a simple but effective adaption method for different backbones. Extensive experimental results on several datasets demonstrate the superiority of our methods for the proposed settings.

Ranking model plays an essential role in e-commerce search and recommendation. An effective ranking model should give a personalized ranking list for each user according to the user preference. Existing algorithms usually extract a user representation vector from the user behavior sequence, then feed the vector into a feed-forward network (FFN) together with other features for feature interactions, and finally produce a personalized ranking score. Despite tremendous progress in the past, there is still room for improvement. Firstly, the personalized patterns of feature interactions for different users are not explicitly modeled. Secondly, most of existing algorithms have poor personalized ranking results for long-tail users with few historical behaviors due to the data sparsity.To overcome the two challenges, we propose Attention Weighted Mixture of Experts (AW-MoE) with contrastive learning for personalized ranking. Firstly, AW-MoE leverages the MoE framework to capture personalized feature interactions for different users. To model the user preference, the user behavior sequence is simultaneously fed into expert networks and the gate network. Within the gate network, one gate unit and one activation unit are designed to adaptively learn the fine-grained activation vector for experts using an attention mechanism. Secondly, a random masking strategy is applied to the user behavior sequence to simulate long-tail users, and an auxiliary contrastive loss is imposed to the output of the gate network to improve the model generalization for these users. This is validated by a higher performance gain on the long-tail user test set.Experiment results on a JD real production dataset and a public dataset demonstrate the effectiveness of AW-MoE, which significantly outperforms state-of-art methods. Notably, AW-MoE has been successfully deployed in the JD e-commerce search engine, serving the real traffic of hundreds of millions of active users.

Providing personalization in product search has attracted increasing attention in both industry and research communities. Most existing personalized product search methods model users' individual search interests based on their historical search logs to generate personalized search results. However, the search logs may be sparse or noisy in the real scenario, which is difficult for existing methods to learn accurate and robust user representations. To address this issue, we propose a contrastive learning framework CoPPS that aims to learn high-quality user representations for personalized product search. Specifically, we design three data augmentation and contrastive learning strategies to construct self-supervision signals from the original search behaviours. The contrastive learning tasks utilize an external knowledge graph and exploit the correlations within and between user sequences, thereby facilitating the discovery of more meaningful search patterns and ultimately enhancing the quality of personalized search. Experimental results on the public Amazon datasets verify the effectiveness of our approach.

Conversational recommender systems (CRS) explicitly solicit users' preferences for improved recommendations on the fly. Most existing CRS solutions count on a single policy trained by reinforcement learning for a population of users. However, for users new to the system, such a global policy becomes ineffective to satisfy them, i.e., the cold-start challenge. In this paper, we study CRS policy learning for cold-start users via meta reinforcement learning. We propose to learn a meta policy and adapt it to new users with only a few trials of conversational recommendations. To facilitate fast policy adaptation, we design three synergetic components. Firstly, we design a meta-exploration policy dedicated to identifying user preferences via a few exploratory conversations, which accelerates personalized policy adaptation from the meta policy. Secondly, we adapt the item recommendation module for each user to maximize the recommendation quality based on the collected conversation states during conversations. Thirdly, we propose a Transformer-based state encoder as the backbone to connect the previous two components. It provides comprehensive state representations by modeling complicated relations between positive and negative feedback during the conversation. Extensive experiments on three datasets demonstrate the advantage of our solution in serving new users, compared with a rich set of state-of-the-art CRS solutions.

Social recommendation based on social network has achieved great success in improving the performance of the recommendation system. Since social network (user-user relations) and user-item interactions are both naturally represented as graph-structured data, Graph Neural Networks (GNNs) have thus been widely applied for social recommendation. Despite the superior performance of existing GNNs-based methods, there are still several severe limitations: (i) Few existing GNNs-based methods have considered a single heterogeneous global graph which takes into account user-user relations, user-item interactions, and item-item similarities simultaneously. That may lead to a lack of complex semantic information and rich topological information when encoding users and items based on GNN. (ii) Furthermore, previous methods tend to overlook the reliability of the original user-user relations which may be noisy and incomplete. (iii) More importantly, the item-item connections established by a few existing methods merely using initial rating attributes or extra attributes (such as category) of items, may be inaccurate or sub-optimal with respect to social recommendation. In order to address these issues, we propose an end-to-end heterogeneous global graph learning framework, namely Graph Learning Augmented Heterogeneous Graph Neural Network (GL-HGNN) for social recommendation. GL-HGNN aims to learn a heterogeneous global graph that makes full use of user-user relations, user-item interactions and item-item similarities in a unified perspective. To this end, we design a Graph Learner (GL) method to learn and optimize user-user and item-item connections separately. Moreover, we employ a Heterogeneous Graph Neural Network (HGNN) to capture the high-order complex semantic relations from our learned heterogeneous global graph. To scale up the computation of graph learning, we further present the Anchor-based Graph Learner (AGL) to reduce computational complexity. Extensive experiments on four real-world datasets demonstrate the effectiveness of our model.

BERT-style models pre-trained on the general corpus (e.g., Wikipedia) and fine-tuned on specific task corpus, have recently emerged as breakthrough techniques in many NLP tasks: question answering, text classification, sequence labeling and so on. However, this tech- nique may not always work, especially for two scenarios: a corpus that contains very different text from the general corpus Wikipedia, or a task that learns embedding spacial distribution for a specific purpose (e.g., approximate nearest neighbor search). In this paper, to tackle the above two scenarios that we have encountered in an industrial e-commerce search system, we propose customized and novel pre-training tasks for two critical modules: user intent detec- tion and semantic embedding retrieval. The customized pre-trained models after fine-tuning, being less than 10% of BERT-base's size in order to be feasible for cost-efficient CPU serving, significantly improve the other baseline models: 1) no pre-training model and 2) fine-tuned model from the official pre-trained BERT using general corpus, on both offline datasets and online system. We have open sourced our datasets 1 for the sake of reproducibility and future works.

Automatic product description generation for e-commerce has witnessed significant advancement in the past decade. Product copy- writing aims to attract users' interest and improve user experience by highlighting product characteristics with textual descriptions. As the services provided by e-commerce platforms become diverse, it is necessary to adapt the patterns of automatically-generated descriptions dynamically. In this paper, we report our experience in deploying an E-commerce Prefix-based Controllable Copywriting Generation (EPCCG) system into the JD.com e-commerce product recommendation platform. The development of the system contains two main components: 1) copywriting aspect extraction; 2) weakly supervised aspect labelling; 3) text generation with a prefix-based language model; and 4) copywriting quality control. We conduct experiments to validate the effectiveness of the proposed EPCCG. In addition, we introduce the deployed architecture which cooperates the EPCCG into the real-time JD.com e-commerce recommendation platform and the significant payoff since deployment. The codes for implementation are provided at https://github.com/xguo7/Automatic-Controllable-Product-Copywriting-for-E-Commerce.git.

Product images are essential for providing desirable user experience in an e-commerce platform. For a platform with billions of products, it is extremely time-costly and labor-expensive to manually pick and organize qualified images. Furthermore, there are the numerous and complicated image rules that a product image needs to comply in order to be generated/selected. To address these challenges, in this paper, we present a new learning framework in order to achieve Automatic Generation of Product-Image Sequence (AGPIS) in e-commerce. To this end, we propose a Multi-modality Unified Image-sequence Classifier (MUIsC), which is able to simultaneously detect all categories of rule violations through learning. MUIsC leverages textual review feedback as the additional training target and utilizes product textual description to provide extra semantic information. %Without using prior knowledge or manually-crafted task, a single MUIsC model is able to learn the holistic knowledge of image reviewing and detect all categories of rule violations simultaneously. Based on offline evaluations, we show that the proposed MUIsC significantly outperforms various baselines. Besides MUIsC, we also integrate some other important modules in the proposed framework, such as primary image selection, non-compliant content detection, and image deduplication. With all these modules, our framework works effectively and efficiently in JD.com recommendation platform. By Dec 2021, our AGPIS framework has generated high-standard images for about 1.5 million products and achieves 13.6% in reject rate. Code of this work is available at https://github.com/efan3000/muisc.

Predicting the next interaction of a short-term interaction session is a challenging task in session-based recommendation. Almost all existing works rely on item transition patterns, and neglect user historical sessions while modeling user preference, which often leads to non-personalized recommendation. And existing personalized session-based recommenders are limited to sessions of the current user, and ignore the useful item-transition patterns from other user's historical sessions. To address these issues, we propose a novel Heterogeneous Global Graph Neural Networks (HG-GNN) to exploit the item transitions over all sessions in a subtle manner for better inferring user preference from the current and historical sessions. To effectively exploit the item transitions over all sessions from users, our global graph contains item transitions of sessions, user-item interactions and global co-occurrence items. Moreover, to capture user preference from sessions comprehensively, we propose a graph augmented preference encoder to learn the session representation. Specifically, we design a novel heterogeneous graph neural network (HGNN) on heterogeneous global graph to learn long-term user preference and item representations with rich semantics. Based on the HGNN, we propose the Personalized Session Encoder to combine the general user preference and temporal interest of the current session to generate the personalized session representation for recommendation. Extensive experimental results on three real-world datasets show that our model outperforms other state-of-the-art methods.

Conversational recommendation system (CRS) is able to obtain fine-grained and dynamic user preferences based on interactive dialogue. Previous CRS assumes that the user has a clear target item, which often deviates from the real scenario, that is for many users who resort to CRS, they might not have a clear idea about what they really like. Specifically, the user may have a clear single preference for some attribute types (e.g. brand) of items, while for other attribute types (e.g. color), the user may have multiple preferences or even no clear preferences, which leads to multiple acceptable attribute instances (e.g. black and red) of one attribute type. Therefore, the users could show their preferences over items under multiple combinations of attribute instances rather than a single item with unique combination of all attribute instances. As a result, we first propose a more realistic conversational recommendation learning setting, namely Multi-Interest Multi-round Conversational Recommendation (MIMCR), where users may have multiple interests in attribute instance combinations and accept multiple items with partially overlapped combinations of attribute instances. To effectively cope with the new CRS learning setting, in this paper, we propose a novel learning framework, namely Multiple Choice questions based Multi-Interest Policy Learning (MCMIPL). In order to obtain user preferences more efficiently, the agent generates multiple choice questions rather than binary yes/no ones on specific attribute instance. Furthermore, we propose a union set strategy to select candidate items instead of existing intersection set strategy in order to overcome over-filtering items during the conversation. Finally, we design a Multi-Interest Policy Learning (MIPL) module, which utilizes captured multiple interests of the user to decide next action, either asking attribute instances or recommending items. Extensive experimental results on four datasets demonstrate the superiority of our method for the proposed MIMCR setting.

Recent years have seen a significant amount of interests in Sequential Recommendation (SR), which aims to understand and model the sequential user behaviors and the interactions between users and items over time. Surprisingly, despite the huge success Sequential Recommendation has achieved, there is little study on Sequential Search (SS), a twin learning task that takes into account a user's current and past search queries, in addition to behavior on historical query sessions. The SS learning task is even more important than the counterpart SR task for most of E-commence companies due to its much larger online serving demands as well as traffic volume.

Nowadays e-commerce search has become an integral part of many people's shopping routines. One critical challenge in today's e-commerce search is the semantic matching problem where the relevant items may not contain the exact terms in the user query. In this paper, we propose a novel deep neural network based approach to query rewriting, in order to tackle this problem. Specifically, we formulate query rewriting into a cyclic machine translation problem to leverage abundant click log data. Then we introduce a novel cyclic consistent training algorithm in conjunction with state-of-the-art machine translation models to achieve the optimal performance in terms of query rewriting accuracy. In order to make it practical in industrial scenarios, we optimize the syntax tree construction to reduce computational cost and online serving latency. Offline experiments show that the proposed method is able to rewrite hard user queries into more standard queries that are more appropriate for the inverted index to retrieve. Comparing with human curated rule-based method, the proposed model significantly improves query rewriting diversity while maintaining good relevancy. Online A/B experiments show that it improves core e-commerce business metrics significantly. Since the summer of 2020, the proposed model has been launched into our search engine production, serving hundreds of millions of users.

Graph convolution networks (GCN), which recently becomes new state-of-the-art method for graph node classification, recommendation and other applications, has not been successfully applied to industrial-scale search engine yet. In this proposal, we introduce our approach, namely SearchGCN, for embedding-based candidate retrieval in one of the largest e-commerce search engine in the world. Empirical studies demonstrate that SearchGCN learns better embedding representations than existing methods, especially for long tail queries and items. Thus, SearchGCN has been deployed into JD.com's search production since July 2020.

We propose a novel domain-specific generative pre-training (DSGPT) method for text generation and apply it to the product title and review summarization problems on E-commerce mobile display. First, we adopt a decoder-only transformer architecture, which fits well for fine-tuning tasks by combining input and output all together. Second, we demonstrate utilizing only small amount of pre-training data in related domains is powerful. Pre-training a language model from a general corpus such as Wikipedia or the Common Crawl requires tremendous time and resource commitment, and can be wasteful if the downstream tasks are limited in variety. Our DSGPT is pre-trained on a limited dataset, the Chinese short text summarization dataset (LCSTS). Third, our model does not require product-related human-labeled data. For title summarization task, the state of art explicitly uses additional background knowledge in training and predicting stages. In contrast, our model implicitly captures this knowledge and achieves significant improvement over other methods, after fine-tuning on the public Taobao.com dataset. For review summarization task, we utilize JD.com in-house dataset, and observe similar improvement over standard machine translation methods which lack the flexibility of fine-tuning. Our proposed work can be simply extended to other domains for a wide range of text generation tasks.

Embedding index that enables fast approximate nearest neighbor(ANN) search, serves as an indispensable component for state-of-the-art deep retrieval systems. Traditional approaches, often separating the two steps of embedding learning and index building, incur additional indexing time and decayed retrieval accuracy. In this paper, we propose a novel method called Poeem, which stands for product quantization based embedding index jointly trained with deep retrieval model, to unify the two separate steps within an end-to-end training, by utilizing a few techniques including the gradient straight-through estimator, warm start strategy, optimal space decomposition and Givens rotation. Extensive experimental results show that the proposed method not only improves retrieval accuracy significantly but also reduces the indexing time to almost none. We have open sourced our approach for the sake of comparison and reproducibility.

Practical large-scale recommender systems usually contain thousands of feature fields from users, items, contextual information, and their interactions. Most of them empirically allocate a unified dimension to all feature fields, which is memory inefficient. Thus it is highly desired to assign various embedding dimensions to different feature fields according to their importance and predictability. Due to the large amounts of feature fields and the nuanced relationship between embedding dimensions with feature distributions and neural network architectures, manually allocating embedding dimensions in practical recommender systems can be challenging. To this end, we propose an AutoML-based framework (AutoDim) in this paper, which can automatically select dimensions for different feature fields in a data-driven fashion. Specifically, we first proposed an end-to-end differentiable framework that can calculate the weights over various dimensions in a soft and continuous manner for feature fields, and an AutoML-based optimization algorithm; then, we derive a hard and discrete embedding component architecture according to the maximal weights and retrain the whole recommender framework. We conduct extensive experiments on benchmark datasets to validate the effectiveness of AutoDim.

Ranking is the most important component in a search system. Most search systems deal with large amounts of natural language data, hence an effective ranking system requires a deep understanding of text semantics. Recently, deep learning based natural language processing (deep NLP) models have generated promising results on ranking systems. BERT is one of the most successful models that learn contextual embedding, which has been applied to capture complex query-document relations for search ranking. However, this is generally done by exhaustively interacting each query word with each document word, which is inefficient for online serving in search product systems. In this paper, we investigate how to build an efficient BERT-based ranking model for industry use cases. The solution is further extended to a general ranking framework, DeText, that is open sourced and can be applied to various ranking productions. Offline and online experiments of DeText on three real-world search systems present significant improvement over state-of-the-art approaches.

As the largest professional network, LinkedIn hosts millions of user profiles and job postings. Users effectively find what they need by entering search queries. However, finding what they are looking for can be a challenge, especially if they are unfamiliar with specific keywords from their industry. Query Suggestion is a popular feature where a search engine can suggest alternate, related queries. At LinkedIn, we have productionized a deep learning Seq2Seq model to transform an input query into several alternatives. This model is trained by examining search history directly typed by users. Once online, we can determine whether or not users clicked on suggested queries. This new feedback data indicates which suggestions caught the user's attention. In this work, we propose training a model with both the search history and user feedback datasets. We examine several ways to incorporate feedback without any architectural change, including adding a novel pairwise ranking loss term during training. The proposed new training technique produces the best combined score out of several alternatives in offline metrics. Deployed in the LinkedIn search engine, it significantly outperforms the control model with respect to key business metrics.

Query Auto Completion (QAC), as the starting point of information retrieval tasks, is critical to user experience. Generally it has two steps: generating completed query candidates according to query prefixes, and ranking them based on extracted features. Three major challenges are observed for a query auto completion system: (1) QAC has a strict online latency requirement. For each keystroke, results must be returned within tens of milliseconds, which poses a significant challenge in designing sophisticated language models for it. (2) For unseen queries, generated candidates are of poor quality as contextual information is not fully utilized. (3) Traditional QAC systems heavily rely on handcrafted features such as the query candidate frequency in search logs, lacking sufficient semantic understanding of the candidate.

In this talk, we will go over the components of personalized search and recommender systems and demonstrate the applications of various deep learning techniques along the way.

Anomaly detection has been widely studied and used in diverse applications. Building an effective anomaly detection system requires researchers and developers to learn complex structure from noisy data, identify dynamic anomaly patterns, and detect anomalies with limited labels. Recent advancements in deep learning techniques have greatly improved anomaly detection performance, in comparison with classical approaches, and have extended anomaly detection to a wide variety of applications. This tutorial will help the audience gain a comprehensive understanding of deep learning based anomaly detection techniques in various application domains. First, we give an overview of the anomaly detection problem, introducing the approaches taken before the deep model era and listing out the challenges they faced. Then we survey the state-of-the-art deep learning models that range from building block neural network structures such as MLP, CNN, and LSTM, to more complex structures such as autoencoder, generative models (VAE, GAN, Flow-based models), to deep one-class detection models, etc. In addition, we illustrate how techniques such as transfer learning and reinforcement learning can help amend the label sparsity issue in anomaly detection problems and how to collect and make the best use of user labels in practice. Second to last, we discuss real world use cases coming from and outside LinkedIn. The tutorial concludes with a discussion of future trends.

Search and recommender systems process rich natural language text data such as user queries and documents. Achieving high-quality search and recommendation results requires processing and understanding such information effectively and efficiently, where natural language processing (NLP) technologies are widely deployed. In recent years, the rapid development of deep learning technology has been proven successful for improving various NLP tasks, indicating their great potential of promoting search and recommender systems.

With the exponential growth of information on the internet, users have been relying on search engines for finding the precise documents. However, user queries are often short. The inherent ambiguity of short queries imposes great challenges for search engines to understand user intent. Query suggestion is one key technique for search engines to augment user queries so that they can better understand user intent. In the past, query suggestions have been relying on either term-frequency--based methods with little semantic understanding of the query, or word-embedding--based methods with little personalization efforts. Here, we present a sequence-to-sequence-model--based query suggestion framework that is capable of modeling structured, personalized features and unstructured query texts naturally. This capability opens up the opportunity to better understand query semantics and user intent at the same time. As the largest professional network, LinkedIn has the advantage of utilizing a rich amount of accurate member profile information to personalize query suggestions. We applied this framework in the LinkedIn production traffic and showed that personalized query suggestions significantly improved member search experience as measured by key business metrics at LinkedIn.

Anomaly detection has been widely studied and used in diverse applications. Building an effective anomaly detection system requires the researchers/developers to learn the complex structure from noisy data, identify the dynamic anomaly patterns and detect anomalies while lacking sufficient labels. Recent advancement in deep learning techniques has made it possible to largely improve anomaly detection performance compared to the classical approaches. This tutorial will help the audience gain a comprehensive understanding of deep learning-based anomaly detection techniques in various application domains. First, it introduces what is the anomaly detection problem, the approaches taken before the deep model era and the challenges it faced. Then it surveys the state-of-the-art deep learning models extensively and discusses the techniques used to overcome the limitations from traditional algorithms. Second to last, it studies deep model anomaly detection techniques in real world examples from LinkedIn production systems. The tutorial concludes with a discussion of future trends.

Search and recommender systems share many fundamental components including language understanding, retrieval and ranking, and language generation. Building powerful search and recommender systems requires processing natural language effectively and efficiently. Recent rapid growth of deep learning technologies has presented both opportunities and challenges in this area. This tutorial offers an overview of deep learning based natural language processing (NLP) for search and recommender systems from an industry perspective. It first introduces deep learning based NLP technologies, including language understanding and language generation. Then it details how those technologies can be applied to common tasks in search and recommender systems, including query and document understanding, retrieval and ranking, and language generation. Applications in LinkedIn production systems are presented. The tutorial concludes with discussion of future trend.

Deep learning models have been very successful in many natural language processing tasks. Search engine works with rich natural language data, e.g., queries and documents, which implies great potential of applying deep natural language processing on such data to improve search performance. Furthermore, it opens an unprecedented opportunity to explore more advanced search experience, such as conversational search and chatbot. This tutorial offers an overview on deep learning based natural language processing for search systems from an industry perspective. We focus on how deep natural language processing powers search systems in practice. The tutorial introduces basic concepts, elaborates associated challenges, reviews the state-of-the-art approaches, covers end-to-end tasks in search systems with examples, and discusses the future trend.

User feedback like clicks and ratings on recommended items provides important information for recommender systems to predict users' interests in unseen items. Most systems rely on models trained using a single type of feedback, e.g., ratings for movie recommendation and clicks for online news recommendation. However, in addition to the primary feedback, many systems also allow users to provide other types of feedback, e.g., liking or sharing an article, or hiding all articles from a source. These additional feedback potentially provides extra information for the recommendation models. To optimize user experience and business objectives, it is important for a recommender system to use both the primary feedback and additional feedback. This paper presents an empirical study on various training methods for incorporating multiple user feedback types based on LinkedIn recommendation products. We study three important problems that we face at LinkedIn: (1) Whether to send an email based on clicks and complaints, (2) how to rank updates in LinkedIn feeds based on clicks and hides and (3) how jointly optimize for viral actions and clicks in LinkedIn feeds. Extensive offline experiments on historical data show the effectiveness of these methods in different situations. Online A/B testing results further demonstrate the impact of these methods on LinkedIn production systems.

Learning to rank arises in many data mining applications, ranging from web search engine, online advertising to recommendation system. In learning to rank, the performance of a ranking model is strongly affected by the number of labeled examples in the training set; on the other hand, obtaining labeled examples for training data is very expensive and time-consuming. This presents a great need for the active learning approaches to select most informative examples for ranking learning; however, in the literature there is still very limited work to address active learning for ranking. In this paper, we propose a general active learning framework, expected loss optimization (ELO), for ranking. The ELO framework is applicable to a wide range of ranking functions. Under this framework, we derive a novel algorithm, expected discounted cumulative gain (DCG) loss optimization (ELO-DCG), to select most informative examples. Then, we investigate both query and document level active learning for raking and propose a two-stage ELO-DCG algorithm which incorporate both query and document selection into active learning. Furthermore, we show that it is flexible for the algorithm to deal with the skewed grade distribution problem with the modification of the loss function. Extensive experiments on real-world web search data sets have demonstrated great potential and effectiveness of the proposed framework and algorithms.

We describe LASER, a scalable response prediction platform currently used as part of a social network advertising system. LASER enables the familiar logistic regression model to be applied to very large scale response prediction problems, including ones beyond advertising. Though the underlying model is well understood, we apply a whole-system approach to address model accuracy, scalability, explore-exploit, and real-time inference. To facilitate training with both large numbers of training examples and high dimensional features on commodity clustered hardware, we employ the Alternating Direction Method of Multipliers (ADMM). Because online advertising applications are much less static than classical presentations of response prediction, LASER employs a number of techniques that allows it to adapt in real time. LASER models can be divided into components with different re-training frequencies, allowing us to learn from changes in ad campaign performance frequently without incurring the cost of retraining larger, more stable sections of the model. Thompson sampling during online inference further helps by efficiently balancing exploration of new ads with exploitation of long running ones. To enable predictions made with the most recent feature data, we employ a range of techniques, including extensive caching and lazy evaluation, to permit real time, low latency scoring. LASER models are defined using a configuration language that ties together the training, validation, and inference pieces and permits even non-programming analysts to experiment with different model structures without modifications to code or interruptions to running servers. Finally, we show via extensive offline experiments and online A/B tests that this system provides significant benefits to prediction accuracy, gains in revenue and CTR, and reductions in system latency.

Web portal services have become an important medium to deliver digital content (e.g. news, advertisements, etc.) to Web users in a timely fashion. To attract more users to various content modules on the Web portal, it is necessary to design a recommender system that can effectively achieve Web portal content optimization by automatically estimating content item attractiveness and relevance to user interests. The state-of-the-art online learning methodology adapts dedicated pointwise models to independently estimate the attractiveness score for each candidate content item. Although such pointwise models can be easily adapted for online recommendation, there still remain a few critical problems. First, this pointwise methodology fails to use invaluable user preferences between content items. Moreover, the performance of pointwise models decreases drastically when facing the problem of sparse learning samples. To address these problems, we propose exploring a new dynamic pairwise learning methodology for Web portal content optimization in which we exploit dynamic user preferences extracted based on users' actions on portal services to compute the attractiveness scores of content items. In this article, we introduce two specific pairwise learning algorithms, a straightforward graph-based algorithm and a formalized Bayesian modeling one. Experiments on large-scale data from a commercial Web portal demonstrate the significant improvement of pairwise methodologies over the baseline pointwise models. Further analysis illustrates that our new pairwise learning approaches can benefit personalized recommendation more than pointwise models, since the data sparsity is more critical for personalized content optimization.

With the rapid growth of E-Commerce on the Internet, online product search service has emerged as a popular and effective paradigm for customers to find desired products and select transactions. Most product search engines today are based on adaptations of relevance models devised for information retrieval. However, there is still a big gap between the mechanism of finding products that customers really desire to purchase and that of retrieving products of high relevance to customers' query. In this paper, we address this problem by proposing a new ranking framework for enhancing product search based on dynamic best-selling prediction in E-Commerce. Specifically, we first develop an effective algorithm to predict the dynamic best-selling, i.e. the volume of sales, for each product item based on its transaction history. By incorporating such best-selling prediction with relevance, we propose a new ranking model for product search, in which we rank higher the product items that are not only relevant to the customer's need but with higher probability to be purchased by the customer. Results of a large scale evaluation, conducted over the dataset from a commercial product search engine, demonstrate that our new ranking method is more effective for locating those product items that customers really desire to buy at higher rank positions without hurting the search relevance.

We study the problem of labeling the edges of a social network graph (e.g., acquaintance connections in Facebook) as either positive (i.e., trust, true friendship) or negative (i.e., distrust, possible frenemy) relations. Such signed relations provide much stronger signal in tying the behavior of online users than the unipolar Homophily effect, yet are largely unavailable as most social graphs only contain unsigned edges.

Learning to rank arises in many information retrieval applications, ranging from Web search engine, online advertising to recommendation systems. Traditional ranking mainly focuses on one type of data source, and effective modeling relies on a sufficiently large number of labeled examples, which require expensive and time-consuming labeling process. However, in many real-world applications, ranking over multiple related heterogeneous domains becomes a common situation, where in some domains we may have a relatively large amount of training data while in some other domains we can only collect very little. Theretofore, how to leverage labeled information from related heterogeneous domain to improve ranking in a target domain has become a problem of great interests. In this paper, we propose a novel probabilistic model, pairwise cross-domain factor model, to address this problem. The proposed model learns latent factors(features) for multi-domain data in partially-overlapped heterogeneous feature spaces. It is capable of learning homogeneous feature correlation, heterogeneous feature correlation, and pairwise preference correlation for cross-domain knowledge transfer. We also derive two PCDF variations to address two important special cases. Under the PCDF model, we derive a stochastic gradient based algorithm, which facilitates distributed optimization and is flexible to adopt different loss functions and regularization functions to accommodate different data distributions. The extensive experiments on real world data sets demonstrate the effectiveness of the proposed model and algorithm.

This article studies evolutionary clustering, a recently emerged hot topic with many important applications, noticeably in dynamic social network analysis. In this article, based on the recent literature on nonparametric Bayesian models, we have developed two generative models: DPChain and HDP-HTM. DPChain is derived from the Dirichlet process mixture (DPM) model, with an exponential decaying component along with the time. HDP-HTM combines the hierarchical dirichlet process (HDP) with a hierarchical transition matrix (HTM) based on the proposed Infinite hierarchical Markov state model (iHMS). Both models substantially advance the literature on evolutionary clustering, in the sense that not only do they both perform better than those in the existing literature, but more importantly, they are capable of automatically learning the cluster numbers and explicitly addressing the corresponding issues. Extensive evaluations have demonstrated the effectiveness and the promise of these two solutions compared to the state-of-the-art literature.

This paper investigates the general problem of pattern change discovery between high-dimensional data sets. Current methods either mainly focus on magnitude change detection of low-dimensional data sets or are under supervised frameworks. In this paper, the notion of the principal angles between the subspaces is introduced to measure the subspace difference between two high-dimensional data sets. Principal angles bear a property to isolate subspace change from the magnitude change. To address the challenge of directly computing the principal angles, we elect to use matrix factorization to serve as a statistical framework and develop the principle of the dominant subspace mapping to transfer the principal angle based detection to a matrix factorization problem. We show how matrix factorization can be naturally embedded into the likelihood ratio test based on the linear models. The proposed method is of an unsupervised nature and addresses the statistical significance of the pattern changes between high-dimensional data sets. We have showcased the different applications of this solution in several specific real-world applications to demonstrate the power and effectiveness of this method.

Combining correlated information from multiple contexts can significantly improve predictive accuracy in recommender problems. Such information from multiple contexts is often available in the form of several incomplete matrices spanning a set of entities like users, items, features, and so on. Existing methods simultaneously factorize these matrices by sharing a single set of factors for entities across all contexts. We show that such a strategy may introduce significant bias in estimates and propose a new model that ameliorates this issue by positing local, context-specific factors for entities. To avoid over-fitting in contexts with sparse data, the local factors are connected through a shared global model. This sharing of parameters allows information to flow across contexts through multivariate regressions among local factors, instead of enforcing exactly the same factors for an entity, everywhere. Model fitting is done in an EM framework, we show that the E-step can be fitted through a fast multi-resolution Kalman filter algorithm that ensures scalability. Experiments on benchmark and real-world Yahoo! datasets clearly illustrate the usefulness of our approach. Our model significantly improves predictive accuracy, especially in cold-start scenarios.

While a user's preference is directly reflected in the interactive choice process between her and the recommender, this wealth of information was not fully exploited for learning recommender models. In particular, existing collaborative filtering (CF) approaches take into account only the binary events of user actions but totally disregard the contexts in which users' decisions are made. In this paper, we propose Collaborative Competitive Filtering (CCF), a framework for learning user preferences by modeling the choice process in recommender systems. CCF employs a multiplicative latent factor model to characterize the dyadic utility function. But unlike CF, CCF models the user behavior of choices by encoding a local competition effect. In this way, CCF allows us to leverage dyadic data that was previously lumped together with missing data in existing CF models. We present two formulations and an efficient large scale optimization algorithm. Experiments on three real-world recommendation data sets demonstrate that CCF significantly outperforms standard CF approaches in both offline and online evaluations.

Targeting interest to match a user with services (e.g. news, products, games, advertisements) and predicting friendship to build connections among users are two fundamental tasks for social network systems. In this paper, we show that the information contained in interest networks (i.e. user-service interactions) and friendship networks (i.e. user-user connections) is highly correlated and mutually helpful. We propose a framework that exploits homophily to establish an integrated network linking a user to interested services and connecting different users with common interests, upon which both friendship and interests could be efficiently propagated. The proposed friendship-interest propagation (FIP) framework devises a factor-based random walk model to explain friendship connections, and simultaneously it uses a coupled latent factor model to uncover interest interactions. We discuss the flexibility of the framework in the choices of loss objectives and regularization penalties and benchmark different variants on the Yahoo! Pulse social networking system. Experiments demonstrate that by coupling friendship with interest, FIP achieves much higher performance on both interest targeting and friendship prediction than systems using only one source of information.

Learning to rank arises in many information retrieval applications, ranging from Web search engine, online advertising to recommendation system. In learning to rank, the performance of a ranking function heavily depends on the number of labeled examples in the training set; on the other hand, obtaining labeled examples for training data is very expensive and time-consuming. This presents a great need for making use of available auxiliary data, i.e., the within-domain unlabeled data and the out-of-domain labeled data. In this paper, we propose a general framework for ranking with auxiliary data, which is applicable to various ranking applications. Under this framework, we derive a generic ranking algorithm to effectively make use of both the within-domain unlabeled data and the out-of-domain labeled data. The proposed algorithm iteratively learns ranking functions for target domain and source domains and enforces their consensus on the unlabeled data in the target domain.

Learning to rank arises in many information retrieval applications, ranging from Web search engine, online advertising to recommendation system. In learning to rank, the performance of a ranking model is strongly affected by the number of labeled examples in the training set; on the other hand, obtaining labeled examples for training data is very expensive and time-consuming. This presents a great need for the active learning approaches to select most informative examples for ranking learning; however, in the literature there is still very limited work to address active learning for ranking. In this paper, we propose a general active learning framework, Expected Loss Optimization (ELO), for ranking. The ELO framework is applicable to a wide range of ranking functions. Under this framework, we derive a novel algorithm, Expected DCG Loss Optimization (ELO-DCG), to select most informative examples. Furthermore, we investigate both query and document level active learning for raking and propose a two-stage ELO-DCG algorithm which incorporate both query and document selection into active learning. Extensive experiments on real-world Web search data sets have demonstrated great potential and effective-ness of the proposed framework and algorithms.

Supervised learning algorithms usually require high quality labeled training set of large volume. It is often expensive to obtain such labeled examples in every domain of an application. Domain adaptation aims to help in such cases by utilizing data available in related domains. However transferring knowledge from one domain to another is often non trivial due to different data distributions among the domains. Moreover, it is usually very hard to measure and formulate these distribution differences. Hence we introduce a new concept of label-relation function to transfer knowledge among different domains without explicitly formulating the data distribution differences. A novel learning framework, Domain Transfer Risk Minimization (DTRM), is proposed based on this concept. DTRM simultaneously minimizes the empirical risk for the target and the regularized empirical risk for source domain. Under this framework, we further derive a generic algorithm called Domain Adaptation by Label Relation (DALR) that is applicable to various applications in both classification and regression settings. DALR iteratively updates the target hypothesis function and outputs for the source domain until it converges. We provide an in-depth theoretical analysis of DTRM and establish fundamental error bounds. We also experimentally evaluate DALR on the task of ranking search results using real-world data. Our experimental results show that the proposed algorithm effectively and robustly utilizes data from source domains under various conditions: different sizes for source domain data; different noise levels for source domain data, and different difficulty levels for target domain data.

In bibliographies like DBLP and Citeseer, there are three kinds of entity-name problems that need to be solved. First, multiple entities share one name, which is called the name sharing problem. Second, one entity has different names, which is called the name variant problem. Third, multiple entities share multiple names, which is called the name mixing problem. We aim to solve these problems based on one model in this paper. We call this task complete entity resolution. Different from previous work, our work use global information based on data with two types of information, words and author names. We propose a generative latent topic model that involves both author names and words - the LDA-dual model, by extending the LDA (Latent Dirichlet Allocation) model. We also propose a method to obtain model parameters that is global information. Based on obtained model parameters, we propose two algorithms to solve the three problems mentioned above. Experimental results demonstrate the effectiveness and great potential of the proposed model and algorithms.

Evolutionary Clustering has emerged as an important research topic in recent literature of data mining, and solutions to this problem have found a wide spectrum of applications, particularly in social network analysis. In this paper, based on the recent literature on Dirichlet processes, we have developed two different and specific models as solutions to this problem: DPChain and HDP-EVO. Both models substantially advance the literature on evolutionary clustering in the sense that not only they both perform better than the existing literature, but more importantly they are capable of automatically learning the cluster numbers and structures during the evolution. Extensive evaluations have demonstrated the effectiveness and promise of these models against the state-of-the-art literature.

This paper studies evolutionary clustering, which is a recently hot topic with many important applications, noticeably in social network analysis. In this paper, based on the recent literature on Hierarchical Dirichlet Process (HDP) and Hidden Markov Model (HMM), we have developed a statistical model HDP-HTM that combines HDP with a Hierarchical Transition Matrix (HTM) based on the proposed Infinite Hierarchical Hidden Markov State model (iH2MS) as an effective solution to this problem. The HDP-HTM model substantially advances the literature on evolutionary clustering in the sense that not only it performs better than the existing literature, but more importantly it is capable of automatically learning the cluster numbers and structures and at the same time explicitly addresses the correspondence issue during the evolution. Extensive evaluations have demonstrated the effectiveness and promise of this solution against the state-of-the-art literature.

Multiple view data, which have multiple representations from different feature spaces or graph spaces, arise in various data mining applications such as information retrieval, bioinformatics and social network analysis. Since different representations could have very different statistical properties, how to learn a consensus pattern from multiple representations is a challenging problem. In this paper, we propose a general model for multiple view unsupervised learning. The proposed model introduces the concept of mapping function to make the different patterns from different pattern spaces comparable and hence an optimal pattern can be learned from the multiple patterns of multiple representations. Under this model, we formulate two specific models for two important cases of unsupervised learning, clustering and spectral dimensionality reduction; we derive an iterating algorithm for multiple view clustering, and a simple algorithm providing a global optimum to multiple spectral dimensionality reduction. We also extend the proposed model and algorithms to evolutionary clustering and unsupervised learning with side information. Empirical evaluations on both synthetic and real data sets demonstrate the effectiveness of the proposed model and algorithms.

Learning communities from a graph is an important problem in many domains. Different types of communities can be generalized as link-pattern based communities. In this paper, we propose a general model based on graph approximation to learn link-pattern based community structures from a graph. The model generalizes the traditional graph partitioning approaches and is applicable to learning various community structures. Under this model, we derive a family of algorithms which are flexible to learn various community structures and easy to incorporate the prior knowledge of the community structures. Experimental evaluation and theoretical analysis show the effectiveness and great potential of the proposed model and algorithms.

Relational clustering has attracted more and more attention due to its phenomenal impact in various important applications which involve multi-type interrelated data objects, such as Web mining, search marketing, bioinformatics, citation analysis, and epidemiology. In this paper, we propose a probabilistic model for relational clustering, which also provides a principal framework to unify various important clustering tasks including traditional attributes-based clustering, semi-supervised clustering, co-clustering and graph clustering. The proposed model seeks to identify cluster structures for each type of data objects and interaction patterns between different types of objects. Under this model, we propose parametric hard and soft relational clustering algorithms under a large number of exponential family distributions. The algorithms are applicable to relational data of various structures and at the same time unifies a number of stat-of-the-art clustering algorithms: co-clustering algorithms, the k-partite graph clustering, Bregman k-means, and semi-supervised clustering based on hidden Markov random fields.

Various data mining applications involve data objects of multiple types that are related to each other, which can be naturally formulated as a k-partite graph. However, the research on mining the hidden structures from a k-partite graph is still limited and preliminary. In this paper, we propose a general model, the relation summary network, to find the hidden structures (the local cluster structures and the global community structures) from a k-partite graph. The model provides a principal framework for unsupervised learning on k-partite graphs of various structures. Under this model, we derive a novel algorithm to identify the hidden structures of a k-partite graph by constructing a relation summary network to approximate the original k-partite graph under a broad range of distortion measures. Experiments on both synthetic and real data sets demonstrate the promise and effectiveness of the proposed model and algorithm. We also establish the connections between existing clustering approaches and the proposed model to provide a unified view to the clustering approaches.

Combining multiple clusterings arises in various important data mining scenarios. However, finding a consensus clustering from multiple clusterings is a challenging task because there is no explicit correspondence between the classes from different clusterings. We present a new framework based on soft correspondence to directly address the correspondence problem in combining multiple clusterings. Under this framework, we propose a novel algorithm that iteratively computes the consensus clustering and correspondence matrices using multiplicative updating rules. This algorithm provides a final consensus clustering as well as correspondence matrices that gives intuitive interpretation of the relations between the consensus clustering and each clustering from clustering ensembles. Extensive experimental evaluations also demonstrate the effectiveness and potential of this framework as well as the algorithm for discovering a consensus clustering from multiple clusterings.

Dyadic data matrices, such as co-occurrence matrix, rating matrix, and proximity matrix, arise frequently in various important applications. A fundamental problem in dyadic data analysis is to find the hidden block structure of the data matrix. In this paper, we present a new coclustering framework, block value decomposition(BVD), for dyadic data, which factorizes the dyadic data matrix into three components, the row-coefficient matrix R, the block value matrix B, and the column-coefficient matrix C. Under this framework, we focus on a special yet very popular case – non-negative dyadic data, and propose a specific novel co-clustering algorithm that iteratively computes the three decomposition matrices based on the multiplicative updating rules. Extensive experimental evaluations also demonstrate the effectiveness and potential of this framework as well as the specific algorithms for co-clustering, and in particular, for discovering the hidden block structure in the dyadic data.