Cross-domain Sequential Recommendation (CDSR) has been proposed to enrich user-item interactions by incorporating information from various domains. Despite current progress, the domain imbalance issue and domain transition issue hinder further development of CDSR. The former presents a phenomenon where interactions in one domain dominate the entire behavior, leading to difficulty in capturing domain-specific features in the other domain. The latter points to the difficulty in capturing users' cross-domain preferences within the mixed interaction sequence, resulting in poor next-item prediction performance for specific domains. With world knowledge and powerful reasoning abilities, Large Language Models (LLMs) partially alleviate the above issues by functioning as both a generator and an encoder. However, current LLMs-enhanced CDSR methods are still under exploration, which fail to recognize the irrelevant noise and rough profiling problems. Thus, to address the aforementioned challenges, we propose an LLMs Enhanced Cross-domain Sequential Recommendation with Dual-phase Training (LLM-EDT). To address the domain imbalance issue while minimizing irrelevant noise, we propose the transferable item augmenter to adaptively generate possible cross-domain behaviors for users. Then, to alleviate the domain transition issue, we introduce a dual-phase training strategy to empower the domain-specific thread with a domain-shared background. As for the rough profiling problem, we devise a domain-aware profiling module to summarize the user's preference in each domain and adaptively aggregate them to generate comprehensive user profiles. The experiments on three public datasets validate the effectiveness of our proposed LLM-EDT. To ease reproducibility, we have released the detailed code online {https://github.com/Applied-Machine-Learning-Lab/SIGIR26_LLM-EDT}. © 2026 Copyright held by the owner/author(s).

Efficient training of large-scale heterogeneous graphs is of paramount importance in real-world applications. However, existing approaches typically explore simplified models to mitigate resource and time overhead, neglecting the crucial aspect of simplifying large-scale heterogeneous graphs from the data-centric perspective. Addressing this gap, HGCond introduces graph condensation (GC) in heterogeneous graphs and generates a small condensed graph for efficient model training. Despite its efficacy in graph generation, HGCond encounters two significant limitations. The first is low effectiveness, HGCond excessively relies on the simplest relay model for the condensation procedure, which restricts the ability to exert powerful Heterogeneous Graph Neural Networks (HGNNs) with flexible condensation ratio and limits the generalization ability. The second is low efficiency, HGCond follows the existing GC methods designed for homogeneous graphs and leverages the sophisticated optimization paradigm, resulting in a time-consuming condensing procedure. In light of these challenges, we present the first Training Free Heterogeneous Graph Condensation method, termed FreeHGC, facilitating both efficient and high-quality generation of heterogeneous condensed graphs. Specifically, we reformulate the heterogeneous graph condensation problem as a data selection issue, offering a new perspective for assessing and condensing representative nodes and edges in the heterogeneous graphs. By leveraging rich meta-paths, we introduce a new, highquality heterogeneous data selection criterion to select target-type nodes. Furthermore, two training-free condensation strategies for heterogeneous graphs are designed to condense and synthesize other-types nodes effectively. Extensive experiments demonstrate the effectiveness and efficiency of our proposed method. Besides, FreeHGC exhibits excellent generalization ability across various heterogeneous graph neural networks. Our codes are available at https://github.com/PKU-DAIR/FreeHGC.

In recent years, Graph Neural Networks (GNNs) have demonstrated remarkable capabilities in learning from graph-structured data. However, GNNs face challenges when dealing with imperfect graph structures, which often lead to performance degradation due to the underlying message propagation mechanism. In response to this issue, a class of data-centric techniques called Graph Structure Learning (GSL) has emerged, with a focus on improving the quality of graph structures. Our review of the existing GSL literature, combined with empirical studies, reveals two primary limitations: low scalability and low efficiency. To mitigate these limitations, we introduce Random Walk-based Graph Structure Learning (RWGSL), a new GSL method that utilizes random walk strategies and operates in a parameter-free manner. Extensive experiments demonstrate that Rwgsl consistently improves the classification performance of both vanilla GNNs and advanced GSL methods across various graph datasets, and Rwgsl can scale to extremely large graphs (e.g. Ogbn-Products) with acceptable time cost. In particular, the combination of Rwgsl and GCN significantly reduces the run time to approximately 5% of those observed in most GSL methods, while also achieving a superior classification accuracy. These findings validate the high scalability and robustness of Rwgsl.

Due to exceptional reasoning and understanding abilities, the Large Language Model (LLM) has revolutionized the pattern of many fields, including recommender systems (RS). There has been a handful of research that focuses on empowering the RS by LLM. Recently, considering the latency and memory costs in real-world applications, LLM-Enhanced RS (LLMERS) is highlighted. This direction pushes the LLM into the online system with a large step by eliminating the utilization of LLM during inference. As a cutting-edge field, there is a clear need for a comprehensive survey to summarize this direction. In this survey, we systematically investigate the most up-to-date works of LLM-enhanced RS to boost this direction. Based on the component of an RS model that the LLM aims to augment, the basic taxonomy includes Knowledge Enhancement, Interaction Enhancement and Model Enhancement. Additionally, we identify several promising research directions. To facilitate access to the surveyed papers, we release a repository.

Cross-domain Sequential Recommendation (CDSR) aims to extract the preference from the user's historical interactions across various domains. Despite some progress in CDSR, two problems set the barrier for further advancements, i.e., overlap dilemma and transition complexity. The former means existing CDSR methods severely rely on users who own interactions on all domains to learn cross-domain item relationships, compromising the practicability. The latter refers to the difficulties in learning the complex transition patterns from the mixed behavior sequences. With powerful representation and reasoning abilities, Large Language Models (LLMs) are promising to address these two problems by bridging the items and capturing the user's preferences from a semantic view. Therefore, we propose an LLMs Enhanced Cross-domain Sequential Recommendation model (LLM4CDSR). To obtain the semantic item relationships, we first propose an LLM-based unified representation module to represent items. Then, a trainable adapter with contrastive regularization is designed to adapt the CDSR task. Besides, a hierarchical LLMs profiling module is designed to summarize user cross-domain preferences. Finally, these two modules are integrated into the proposed tri-thread framework to derive recommendations. We have conducted extensive experiments on three public cross-domain datasets, validating the effectiveness of LLM4CDSR. We have released the code online.

Unbiased learning to rank (ULTR), which utilizes historical click logs to train ranking models, has attracted much attention in the IR community. Previous studies on ULTR have focused on mitigating a variety of biases in click logs, such as position bias, trust bias, and presentation bias, to recover the true relevance of the query-document pairs. However, they overlooked the intrinsic distribution shifts between the training data and test data. In this paper, we first validate and analyze the distribution shift problem with a real-world ULTR dataset. To solve this problem, we propose distributionally robust unbiased learning to rank (DRO-ULTR) methods. Specifically, we design two kinds of group distributionally robust optimization (group-DRO) frameworks for the existing ULTR methods, one using the pointwise click prediction loss and the other using the listwise counterfactual ranking loss. Finally, we empirically verify the effectiveness of our DRO-ULTR methods by conducting extensive experiments on the real-world dataset.

Recent advances in Retrieval-Augmented Generation (RAG) have significantly improved response accuracy and relevance by incorporating external knowledge into Large Language Models (LLMs). However, existing RAG methods primarily focus on generating text-only answers, even in Multimodal Retrieval-Augmented Generation (MRAG) scenarios, where multimodal elements are retrieved to assist in generating text answers. To address this, we introduce the Multimodal Retrieval-Augmented Multimodal Generation (MRAMG) task, in which we aim to generate multimodal answers that combine both text and images, fully leveraging the multimodal data within a corpus. Despite growing attention to this challenging task, a notable lack of a comprehensive benchmark persists for effectively evaluating its performance. To bridge this gap, we provide MRAMG-Bench, a meticulously curated, human-annotated benchmark comprising 4,346 documents, 14,190 images, and 4,800 QA pairs, distributed across six distinct datasets and spanning three domains: Web, Academia, and Lifestyle. The datasets incorporate diverse difficulty levels and complex multi-image scenarios, providing a robust foundation for evaluating the MRAMG task. To facilitate rigorous evaluation, MRAMG-Bench incorporates a comprehensive suite of both statistical and LLM-based metrics, enabling a thorough analysis of the performance of generative models in the MRAMG task. Additionally, we propose an efficient and flexible multimodal answer generation framework that can leverage LLMs/MLLMs to generate multimodal responses. Our datasets and complete evaluation results for 11 popular generative models are available at https://github.com/MRAMG-Bench/MRAMG.

Recommendation Systems have become integral to modern user experiences, but lack transparency in their decision-making processes. Existing explainable recommendation methods are hindered by reliance on a post-hoc paradigm, wherein explanation generators are trained independently of the underlying recommender models. This paradigm necessitates substantial human effort in data construction and raises concerns about explanation reliability. In this paper, we present ExpCTR, a novel framework that integrates large language model based explanation generation directly into the CTR prediction process. Inspired by recent advances in reinforcement learning, we employ two carefully designed reward mechanisms, LC alignment, which ensures explanations reflect user intentions, and IC alignment, which maintains consistency with traditional ID-based CTR models. Our approach incorporates an efficient training paradigm with LoRA and a three-stage iterative process. ExpCTR circumvents the need for extensive explanation datasets while fostering synergy between CTR prediction and explanation generation. Experimental results demonstrate that ExpCTR significantly enhances both recommendation accuracy and interpretability across three real-world datasets.

Legal case retrieval is an important task in information retrieval that aims to retrieve relevant cases for given query cases. Conversational search paradigms have been shown to improve the search experience in legal case retrieval. However, there are two challenges in applying conversational search to legal scenarios. Firstly, legal search conversations often focus on different parts of legal case documents, but existing models struggle to capture the complex structural information and extract accurate relevance signals. Secondly, collecting large-scale conversational search datasets is costly, making it difficult to build reliable conversational legal case retrieval models. To address these challenges, we propose a Structure-Aware Matching Model (SAMM) for conversational legal case retrieval. SAMM extracts matching signals between conversational utterances and segments of the legal cases to incorporate structural information. We decouple the conversational search task into three subtasks and design pre-training tasks to overcome the lack of training data. Additionally, we create ConvLegal, the largest conversational legal case retrieval dataset to the best of our knowledge, for better evaluation of different methods. We train and evaluate SAMM and baselines on both a public dataset (CLCR) and ConvLegal. Experimental results demonstrate that SAMM outperforms existing models in legal case retrieval and conversational search.

As the focus on Large Language Models (LLMs) in the field of recommendation intensifies, the optimization of LLMs for recommendation purposes (referred to as LLM4Rec) assumes a crucial role in enhancing their recommendation performance. However, existing approaches for LLM4Rec often assess performance using restricted sets of candidates, which may not accurately reflect the models’ overall ranking capabilities. In this article, our objective is to pursue LLM4Rec models with comprehensive ranking capacity and propose a two-step grounding framework known as BIGRec (Bi-step Grounding Paradigm for Recommendation). BIGRecm initially grounds LLMs to the recommendation space by fine-tuning them to generate meaningful tokens for items and subsequently identifies appropriate actual items that correspond to the generated tokens. By conducting extensive experiments on two datasets, we substantiate the superior performance, capacity for handling few-shot scenarios, and versatility across multiple domains exhibited by BIGRec. Furthermore, we observe that the marginal benefits derived from increasing the quantity of training samples are modest for BIGRec, implying that LLMs possess the limited capability to assimilate statistical information, such as popularity and collaborative filtering, due to their robust semantic priors. These findings also underline the efficacy of integrating diverse statistical information into the LLM4Rec framework, thereby pointing towards a potential avenue for future research. Finally, we conduct analysis utilizing BIGRec to explore the characteristics of incorporating recommendations into LLMs, thereby offering prospective insights for the advancement of the field. Our code and data are available at https://github.com/SAI990323/Grounding4Rec .

Adapting Large Language Models for Recommendation (LLM4Rec) has shown promising results. However, the challenges of deploying LLM4Rec in real-world scenarios remain largely unexplored. In particular, recommender models need incremental adaptation to evolving user preferences, while the suitability of traditional incremental learning methods within LLM4Rec remains ambiguous due to the unique characteristics of Large Language Models (LLMs).

In the information retrieval (IR) area, dense retrieval (DR) models use deep learning techniques to encode queries and passages into embedding space to compute their semantic relations. It is important for DR models to balance both efficiency and effectiveness. Pre-trained language models (PLMs), especially Transformer-based PLMs, have been proven to be effective encoders of DR models. However, the self-attention component in Transformer-based PLM results in a computational complexity that grows quadratically with sequence length, and thus exhibits a slow inference speed for long-text retrieval. Some recently proposed non-Transformer PLMs, especially the Mamba architecture PLMs, have demonstrated not only comparable effectiveness to Transformer-based PLMs on generative language tasks but also better efficiency due to linear time scaling in sequence length. This paper implements the Mamba Retriever to explore whether Mamba can serve as an effective and efficient encoder of DR model for IR tasks. We fine-tune the Mamba Retriever on the classic short-text MS MARCO passage ranking dataset and the long-text LoCoV0 dataset. Experimental results show that (1) on the MS MARCO passage ranking dataset and BEIR, the Mamba Retriever achieves comparable or better effectiveness compared to Transformer-based retrieval models, and the effectiveness grows with the size of the Mamba model; (2) on the long-text LoCoV0 dataset, the Mamba Retriever can extend to longer text length than its pre-trained length after fine-tuning on retrieval task, and it has comparable or better effectiveness compared to other long-text retrieval models; (3) the Mamba Retriever has superior inference speed for long-text retrieval. In conclusion, Mamba Retriever is both effective and efficient, making it a practical model, especially for long-text retrieval.

As a representative information retrieval task, site recommendation, which aims at predicting the optimal sites for a brand or an institution to open new branches in an automatic data-driven way, is beneficial and crucial for brand development in modern business. However, there is no publicly available dataset so far and most existing approaches are limited to an extremely small scope of brands, which seriously hinders the research on site recommendation. Therefore, we collect, construct and release an open comprehensive dataset, namely OpenSiteRec, to facilitate and promote the research on site recommendation. Specifically, OpenSiteRec leverages a heterogeneous graph schema to represent various types of real-world entities and relations in four international metropolises. To evaluate the performance of the existing general methods on the site recommendation task, we conduct benchmarking experiments of several representative recommendation models on OpenSiteRec. Furthermore, we also highlight the potential application directions to demonstrate the wide applicability of OpenSiteRec. We believe that our OpenSiteRec dataset is significant and anticipated to encourage the development of advanced methods for site recommendation. OpenSiteRec is available online at https://OpenSiteRec.github.io/.

Legal case retrieval, which aims to find relevant cases based on a short case description, serves as an important part of modern legal systems. Despite the success of existing retrieval methods based on Pretrained Language Models, there are still two issues in legal case retrieval that have not been well considered before. First, existing methods underestimate the semantics associations among legal elements, e.g., law articles and crimes, which played an essential role in legal case retrieval. These methods only adopt the pre-training language model to encode the whole legal case, instead of distinguishing different legal elements in the legal case. They randomly split a legal case into different segments, which may break the completeness of each legal element. Second, due to the difficulty in annotating the relevant labels of similar cases, legal case retrieval inevitably faces the problem of lacking training data. In this paper, we propose a counterfactual graph learning framework for legal case retrieval. Concretely, to overcome the above challenges, we transform the legal case document into a graph and model the semantics of the legal elements through a graph neural network. To alleviate the low resource and learn the causal relationship between the semantics of legal elements and relevance, a counterfactual data generator is designed to augment counterfactual data and enhance legal case representation. Extensive experiments based on two publicly available legal benchmarks demonstrate that our CFGL-LCR can significantly outperform previous state-of-the-art methods in legal case retrieval.

Legal case retrieval, which aims to find relevant cases for a query case, plays a core role in the intelligent legal system. Despite the success that pre-training has achieved in ad-hoc retrieval tasks, effective pre-training strategies for legal case retrieval remain to be explored. Compared with general documents, legal case documents are typically long text sequences with intrinsic logical structures. However, most existing language models have difficulty understanding the long-distance dependencies between different structures. Moreover, in contrast to the general retrieval, the relevance in the legal domain is sensitive to key legal elements. Even subtle differences in key legal elements can significantly affect the judgement of relevance. However, existing pre-trained language models designed for general purposes have not been equipped to handle legal elements.

Session search is a widely adopted technique in search engines that seeks to leverage the complete interaction history of a search session to better understand the information needs of users and provide more relevant ranking results. The vast majority of existing methods model a search session as a sequence of queries and previously clicked documents. However, if we simply represent a search session as a sequence we will lose the topological information in the original search session. It is non-trivial to model the intra-session interactions and complicated structural patterns among the previously issued queries, clicked documents, as well as the terms or entities that appeared in them. To solve this problem, in this paper, we propose a novel Session Search with Graph Classification Model (SSGC), which regards session search as a graph classification task on a heterogeneous graph that represents the search history in each session. To improve the performance of the graph classification, we design a specific pre-training strategy for our proposed GNN-based classification model. Extensive experiments on two public session search datasets demonstrate the effectiveness of our model in the session search task.

Overfitting is a common problem in machine learning, which means the model too closely fits the training data while performing poorly in the test data. Among various methods of coping with overfitting, dropout is one of the representative ways. From randomly dropping neurons to dropping neural structures, dropout has achieved great success in improving model performances. Although various dropout methods have been designed and widely applied in past years, their effectiveness, application scenarios, and contributions have not been comprehensively summarized and empirically compared by far. It is the right time to make a comprehensive survey. In this paper, we systematically review previous dropout methods and classify them into three major categories according to the stage where dropout operation is performed. Specifically, more than seventy dropout methods published in top AI conferences or journals (e.g., TKDE, KDD, TheWebConf, SIGIR) are involved. The designed taxonomy is easy to understand and capable of including new dropout methods. Then, we further discuss their application scenarios, connections, and contributions. To verify the effectiveness of distinct dropout methods, extensive experiments are conducted on recommendation scenarios with abundant heterogeneous information. Finally, we propose some open problems and potential research directions about dropout that worth to be further explored.

Recommendation systems play an important role in alleviating the information overload issue. Generally, a recommendation model is trained to discern between positive (liked) and negative (disliked) instances for each user. However, under the open-world assumption, there are only positive instances but no negative instances from users’ implicit feedback, which poses the imbalanced learning challenge of lacking negative samples. To address this, two types of learning strategies have been proposed before, the negative sampling strategy and non-sampling strategy. The first strategy samples negative instances from missing data (i.e., unlabeled data), while the non-sampling strategy regards all the missing data as negative. Although learning strategies are known to be essential for algorithm performance, the in-depth comparison of negative sampling and non-sampling has not been sufficiently explored by far. To bridge this gap, we systematically analyze the role of negative sampling and non-sampling for implicit recommendation in this work. Specifically, we first theoretically revisit the objection of negative sampling and non-sampling. Then, with a careful setup of various representative recommendation methods, we explore the performance of negative sampling and non-sampling in different scenarios. Our results empirically show that although negative sampling has been widely applied to recent recommendation models, it is non-trivial for uniform sampling methods to show comparable performance to non-sampling learning methods. Finally, we discuss the scalability and complexity of negative sampling and non-sampling and present some open problems and future research topics that are worth being further explored.

Sequential recommendation has become a trending research topic for its capability to capture dynamic user intents based on historical interaction sequence. To train a sequential recommendation model, it is a common practice to optimize the next-item recommendation task with a pairwise ranking loss. In this paper, we revisit this typical training method from the perspective of contrastive learning and find it can be taken as a specialized contrastive learning task conceptually and mathematically, named context-target contrast . Further, to leverage other self-supervised signals in user interaction sequences, we propose another contrastive learning task to encourage sequences after augmentation, as well as sequences with the same target item, to have similar representations, called context-context contrast . A general framework, ContraRec, is designed to unify the two kinds of contrast signals, leading to a holistic joint-learning framework for sequential recommendation with different contrastive learning tasks. Besides, various sequential recommendation methods (e.g., GRU4Rec, Caser, and BERT4Rec) can be easily integrated as the base sequence encoder in our ContraRec framework. Extensive experiments on three public datasets demonstrate that ContraRec achieves superior performance compared to state-of-the-art sequential recommendation methods.

Collaborative filtering (CF) plays a critical role in the development of recommender systems. Most CF methods utilize an encoder to embed users and items into the same representation space, and the Bayesian personalized ranking (BPR) loss is usually adopted as the objective function to learn informative encoders. Existing studies mainly focus on designing more powerful encoders (e.g., graph neural network) to learn better representations. However, few efforts have been devoted to investigating the desired properties of representations in CF, which is important to understand the rationale of existing CF methods and design new learning objectives. In this paper, we measure the representation quality in CF from the perspective of alignment and uniformity on the hypersphere. We first theoretically reveal the connection between the BPR loss and these two properties. Then, we empirically analyze the learning dynamics of typical CF methods in terms of quantified alignment and uniformity, which shows that better alignment or uniformity both contribute to higher recommendation performance. Based on the analyses results, a learning objective that directly optimizes these two properties is proposed, named DirectAU. We conduct extensive experiments on three public datasets, and the proposed learning framework with a simple matrix factorization model leads to significant performance improvements compared to state-of-the-art CF methods.

Recommender systems provide essential web services by learning users’ personal preferences from collected data. However, in many cases, systems also need to forget some training data. From the perspective of privacy, users desire a tool to erase the impacts of their sensitive data from the trained models. From the perspective of utility, if a system’s utility is damaged by some bad data, the system needs to forget such data to regain utility. While unlearning is very important, it has not been well-considered in existing recommender systems. Although there are some researches have studied the problem of machine unlearning, existing methods can not be directly applied to recommendation as they are unable to consider the collaborative information.

Knowledge Graph (KG) is a flexible structure that is able to describe the complex relationship between data entities. Currently, most KG embedding models are trained based on negative sampling, i.e., the model aims to maximize some similarity of the connected entities in the KG, while minimizing the similarity of the sampled disconnected entities. Negative sampling helps to reduce the time complexity of model learning by only considering a subset of negative instances, which may fail to deliver stable model performance due to the uncertainty in the sampling procedure. To avoid such deficiency, we propose a new framework for KG embedding—Efficient Non-Sampling Knowledge Graph Embedding (NS-KGE). The basic idea is to consider all of the negative instances in the KG for model learning, and thus to avoid negative sampling. The framework can be applied to square-loss based knowledge graph embedding models or models whose loss can be converted to a square loss. A natural side-effect of this non-sampling strategy is the increased computational complexity of model learning. To solve the problem, we leverage mathematical derivations to reduce the complexity of non-sampling loss function, which eventually provides us both better efficiency and better accuracy in KG embedding compared with existing models. Experiments on benchmark datasets show that our NS-KGE framework can achieve a better performance on efficiency and accuracy over traditional negative sampling based models, and that the framework is applicable to a large class of knowledge graph embedding models.

User intention is an important factor to be considered for recommender systems, which always changes dynamically in different contexts. Recent studies (represented by sequential recommendation) begin to focus on predicting what users want beyond what users like, which are better at capturing user intention and have attracted a surge of interest. However, user intention modeling is non-trivial, because it is generally influenced by various factors, among which item relations and their temporal evolutionary effects are of great importance. For example, consumption of a cellphone will have varying impacts on the demands for its relational items: For complements, the demands are likely to be promoted in the short term; while for substitutes, the long-term effect may take advantage, because users do not need another cellphone immediately. Moreover, the temporal evolutions of different relational effects vary across different domains, which makes it challenging to adaptively take them into consideration. As a result, most existing studies only loosely incorporate item relations by encoding their semantics into embeddings, neglecting fine-grained time-aware effects. In this work, we propose Knowledge-aware Dynamic Attention (KDA) to take both relational effects and their temporal evolutions into consideration. Specifically, to model dynamic impacts of historical relational interactions on user intention, we aggregate the history sequence into relation-specific embeddings, where the attention weight consists of two parts. First, we measure the relational intensity between historical items and the target item to model the absolute degree of influence in terms of each relation. Second, to model how the relational effects drift with time, we innovatively introduce Fourier transform with learnable frequency-domain embeddings to estimate temporal decay functions of different relations adaptively. Subsequently, the self-attention mechanism is leveraged to derive the final representation of the whole history sequence, which reflects the dynamic user intention and will be applied to generate the recommendation list. Extensive experiments in three real-world datasets indicate the proposed KDA model significantly outperforms the state-of-the-art methods on the Top- K recommendation task. Moreover, the proposed Fourier-based method opens up a new avenue to adaptively integrate temporal dynamics into general neural models.

Knowledge graph (KG) contains well-structured external information and has shown to be effective for high-quality recommendation. However, existing KG enhanced recommendation methods have largely focused on exploring advanced neural network architectures to better investigate the structural information of KG. While for model learning, these methods mainly rely on Negative Sampling (NS) to optimize the models for both KG embedding task and recommendation task. Since NS is not robust (e.g., sampling a small fraction of negative instances may lose lots of useful information), it is reasonable to argue that these methods are insufficient to capture collaborative information among users, items, and entities.

To provide more accurate recommendation, it is a trending topic to go beyond modeling user-item interactions and take context features into account. Factorization Machines (FM) with negative sampling is a popular solution for context-aware recommendation. However, it is not robust as sampling may lost important information and usually leads to non-optimal performances in practical. Several recent efforts have enhanced FM with deep learning architectures for modelling high-order feature interactions. While they either focus on rating prediction task only, or typically adopt the negative sampling strategy for optimizing the ranking performance. Due to the dramatic fluctuation of sampling, it is reasonable to argue that these sampling-based FM methods are still suboptimal for context-aware recommendation.

Recommendation systems play a vital role to keep users engaged with personalized contents in modern online platforms. Recently, deep learning has revolutionized many research fields and there is a surge of interest in applying it for recommendation. However, existing studies have largely focused on exploring complex deep-learning architectures for recommendation task, while typically applying the negative sampling strategy for model learning. Despite effectiveness, we argue that these methods suffer from two important limitations: (1) the methods with complex network structures have a substantial number of parameters, and require expensive computations even with a sampling-based learning strategy; (2) the negative sampling strategy is not robust, making sampling-based methods difficult to achieve the optimal performance in practical applications. In this work, we propose to learn neural recommendation models from the whole training data without sampling. However, such a non-sampling strategy poses strong challenges to learning efficiency. To address this, we derive three new optimization methods through rigorous mathematical reasoning, which can efficiently learn model parameters from the whole data (including all missing data) with a rather low time complexity. Moreover, based on a simple Neural Matrix Factorization architecture, we present a general framework named ENMF, short for Efficient Neural Matrix Factorization . Extensive experiments on three real-world public datasets indicate that the proposed ENMF framework consistently and significantly outperforms the state-of-the-art methods on the Top-K recommendation task. Remarkably, ENMF also shows significant advantages in training efficiency, which makes it more applicable to real-world large-scale systems.

Many previous studies attempt to utilize information from other domains to achieve better performance of recommendation. Recently, social information has been shown effective in improving recommendation results with transfer learning frameworks, and the transfer part helps to learn users' preferences from both item domain and social domain. However, two vital issues have not been well-considered in existing methods: 1) Usually, a static transfer scheme is adopted to share a user's common preference between item and social domains, which is not robust in real life where the degrees of sharing and information richness are varied for different users. Hence a non-personalized transfer scheme may be insufficient and unsuccessful. 2) Most previous neural recommendation methods rely on negative sampling in training to increase computational efficiency, which makes them highly sensitive to sampling strategies and hence difficult to achieve optimal results in practical applications.

Social connections are known to be helpful for modeling users' potential preferences and improving the performance of recommender systems. However, in social-aware recommendations, there are two issues which influence the inference of users' preferences, and haven't been well-studied in most existing methods: First, the preferences of a user may only partially match that of his friends in certain aspects, especially when considering a user with diverse interests. Second, for an individual, the influence strength of his friends might be different, as not all friends are equally helpful for modeling his preferences in the system. To address the above issues, in this paper, we propose a novel Social Attentional Memory Network (SAMN) for social-aware recommendation. Specifically, we first design an attention-based memory module to learn user-friend relation vectors, which can capture the varying aspect attentions that a user share with his different friends. Then we build a friend-level attention component to adaptively select informative friends for user modeling. The two components are fused together to mutually enhance each other and lead to a finer extended model. Experimental results on three publicly available datasets show that the proposed SAMN model consistently and significantly outperforms the state-of-the-art recommendation methods. Furthermore, qualitative studies have been made to explore what the proposed attention-based memory module and friend-level attention have learnt, which provide insights into the model's learning process.

Reviews information is dominant for users to make online purchasing decisions in e-commerces. However, the usefulness of reviews is varied. We argue that less-useful reviews hurt model's performance, and are also less meaningful for user's reference. While some existing models utilize reviews for improving the performance of recommender systems, few of them consider the usefulness of reviews for recommendation quality. In this paper, we introduce a novel attention mechanism to explore the usefulness of reviews, and propose a Neural Attentional Regression model with Review-level Explanations (NARRE) for recommendation. Specifically, NARRE can not only predict precise ratings, but also learn the usefulness of each review simultaneously. Therefore, the highly-useful reviews are obtained which provide review-level explanations to help users make better and faster decisions. Extensive experiments on benchmark datasets of Amazon and Yelp on different domains show that the proposed NARRE model consistently outperforms the state-of-the-art recommendation approaches, including PMF, NMF, SVD++, HFT, and DeepCoNN in terms of rating prediction, by the proposed attention model that takes review usefulness into consideration. Furthermore, the selected reviews are shown to be effective when taking existing review-usefulness ratings in the system as ground truth. Besides, crowd-sourcing based evaluations reveal that in most cases, NARRE achieves equal or even better performances than system's usefulness rating method in selecting reviews. And it is flexible to offer great help on the dominant cases in real e-commerce scenarios when the ratings on review-usefulness are not available in the system.