The development of high-quality datasets is crucial for benchmarking and advancing research in Graphical User Interface (GUI) agents. Despite their importance, existing datasets are often constructed under idealized conditions, overlooking the diverse anomalies frequently encountered in real-world deployments. To address this limitation, we introduce GUI-Robust, a novel dataset designed for comprehensive GUI agent evaluation, explicitly incorporating seven common types of anomalies observed in everyday GUI interactions. Furthermore, we propose a semi-automated dataset construction paradigm that collects user action sequences from natural interactions via RPA tools and then generate corresponding step and task descriptions for these actions with the assistance of MLLMs. This paradigm significantly reduces annotation time cost by a factor of over 19 times. Finally, we assess state-of-the-art GUI agents using the GUI-Robust dataset, revealing their substantial performance degradation in abnormal scenarios. We anticipate that our work will highlight the importance of robustness in GUI agents and inspires more future research in this direction. The dataset and code are available at https://github.com/chessbean1/GUI-Robust.

Traditional sequential recommendation (SR) models learn low-dimensional item ID embeddings from user-item interactions, often overlooking textual information such as item titles or descriptions. Recent advances in Large Language Models (LLMs) have inspired a surge of research that encodes item textual information with high-dimensional semantic embeddings, and designs transformation methods to inject such embeddings into SR models. These embedding transformation strategies can be categorized into two types, both of which exhibits notable drawbacks: 1) adapter-based methods suffer from pronounced dimension collapse, concentrating information into a few dominant dimensions; 2) SVD-based methods are rigid and manual, considering only a few principal spectral components while discarding rich information in the remaining spectrum.

Recent years have seen a rapid surge in research leveraging Large Language Models (LLMs) for recommendation. These methods typically employ supervised fine-tuning (SFT) to adapt LLMs to recommendation scenarios, and utilize beam search during inference to efficiently retrieve B top-ranked recommended items. However, we identify a critical training-inference inconsistency: while SFT optimizes the overall probability of positive items, it does not guarantee that such items will be retrieved by beam search even if they possess high overall probabilities. Due to the greedy pruning mechanism, beam search can prematurely discard a positive item once its prefix probability is insufficient.

Recommendation systems (RS) aim to retrieve the top-K items most relevant to users, with metrics such as Precision@K and Recall@K commonly used to assess effectiveness. The architecture of an RS model acts as an inductive bias, shaping the patterns the model is inclined to learn. In recent years, numerous recommendation architectures have emerged, spanning traditional matrix factorization, deep neural networks, and graph neural networks. However, their designs are often not explicitly aligned with the top-K objective, thereby limiting their effectiveness. To address this limitation, we propose TopKGAT, a novel recommendation architecture directly derived from a differentiable approximation of top-K metrics. The forward computation of a single TopKGAT layer is intrinsically aligned with the gradient ascent dynamics of the Precision@K metric, enabling the model to naturally improve top-K recommendation accuracy. Structurally, TopKGAT resembles a graph attention network and can be implemented efficiently. Extensive experiments on four benchmark datasets demonstrate that TopKGAT consistently outperforms state-of-the-art baselines. The code is available at https://github.com/StupidThree/TopKGAT.

Click-through rate (CTR) prediction is a fundamental task in modern recommender systems. In recent years, the integration of large language models (LLMs) has been shown to effectively enhance the performance of traditional CTR methods. However, existing LLM-enhanced methods often require extensive processing of detailed textual descriptions for large-scale instances or user/item entities, leading to substantial computational overhead. To address this challenge, this work introduces LLaCTR, a novel and lightweight LLM-enhanced CTR method that employs a field-level enhancement paradigm. Specifically, LLaCTR first utilizes LLMs to distill crucial and lightweight semantic knowledge from small-scale feature fields through self-supervised field-feature fine-tuning. Subsequently, it leverages this field-level semantic knowledge to enhance both feature representation and feature interactions. In our experiments, we integrate LLaCTR with six representative CTR models across four datasets, demonstrating its superior performance in terms of both effectiveness and efficiency compared to existing LLM-enhanced methods. Our code is available at https://github.com/istarryn/LLaCTR.

Sequential recommenders generate recommendations based on users’ historical interaction sequences. However, in practice, these sequences are often contaminated by noisy interactions, which can arise from various factors such as clickbait, the influence of prominently positioned items, or accidental interactions. Such noise can significantly degrade recommendation performance. Accurately identifying such noisy interactions without additional information is particularly challenging due to the absence of explicit supervisory signals indicating noise. Large Language Models (LLMs), equipped with extensive open knowledge and semantic reasoning abilities, offer a promising avenue to bridge this information gap. However, employing LLMs for denoising in sequential recommendation presents notable challenges: (1) Direct application of pretrained LLMs may not be competent for the denoising task, frequently generating nonsensical responses; (2) Fine-tuning on the denoising task can partially mitigate the issue of generating nonsensical outputs. However, even after fine-tuning, the reliability of LLM outputs remains questionable, especially given the complexity of the denoising task and the inherent hallucination issue of LLMs. To tackle these challenges, we propose LLM4DSR, a tailored approach for denoising sequential recommendation using LLMs. We constructed a self-supervised fine-tuning task to activate LLMs’ capabilities to identify noisy items and suggest replacements. Furthermore, we developed an uncertainty estimation module that ensures only high-confidence responses are utilized for sequence corrections. Remarkably, LLM4DSR is model-agnostic, allowing corrected sequences to be flexibly applied across various recommendation models. To the best of our knowledge, this is the first work that employs LLMs for sequential recommendation denoising while addressing the unique challenges of adapting LLMs to this task. Extensive experiments conducted on three real-world datasets across two noise settings validate the effectiveness of LLM4DSR, demonstrating an average improvement of 12.9% in NDCG@20. The code is available at https://github.com/WANGBohaO-jpg/LLM4DSR .

In the realm of recommender systems (RS), Top-K ranking metrics such as NDCG@K are the gold standard for evaluating recommendation performance. However, during the training of recommendation models, optimizing NDCG@K poses significant challenges due to its inherent discontinuous nature and the intricate Top-K truncation. Recent efforts to optimize NDCG@K have either overlooked the Top-K truncation or suffered from high computational costs and training instability. To overcome these limitations, we propose SoftmaxLoss@K (SL@K), a novel recommendation loss tailored for NDCG@K optimization. Specifically, we integrate the quantile technique to handle Top-K truncation and derive a smooth upper bound for optimizing NDCG@K to address discontinuity. The resulting SL@K loss has several desirable properties, including theoretical guarantees, ease of implementation, computational efficiency, gradient stability, and noise robustness. Extensive experiments on four real-world datasets and three recommendation backbones demonstrate that SL@K outperforms existing losses with a notable average improvement of 6.03%. The code is available at https://github.com/Tiny-Snow/IR-Benchmark.

Large language models (LLMs), known for their comprehension capabilities and extensive knowledge, have been increasingly applied to recommendation systems (RS).Given the fundamental gap between the mechanism of LLMs and the requirement of RS, researchers have focused on fine-tuning LLMs with recommendationspecific data to enhance their performance.Language Modeling Loss (LML), originally designed for language generation tasks, is commonly adopted.However, we identify two critical limitations of LML: 1) it exhibits significant divergence from the recommendation objective; 2) it erroneously treats all fictitious item descriptions as negative samples, introducing misleading training signals.To address these limitations, we propose a novel Masked Softmax Loss (MSL) tailored for fine-tuning LLMs on recommendation.MSL improves LML by identifying and masking invalid tokens that could lead to fictitious item descriptions during loss computation.This strategy can effectively avoid the interference from erroneous

Recommendation Systems (RS) are often plagued by popularity bias. When training a recommendation model on a typically long-tailed dataset, the model tends to not only inherit this bias but often exacerbate it, resulting in over-representation of popular items in the recommendation lists. This study conducts comprehensive empirical and theoretical analyses to expose the root causes of this phenomenon, yielding two core insights: 1) Item popularity is memorized in the principal spectrum of the score matrix predicted by the recommendation model; 2) The dimension reduction phenomenon amplifies the relative prominence of the principal spectrum, thereby intensifying the popularity bias.

Learning effective representations for Continuous-Time Dynamic Graphs (CTDGs) has garnered significant research interest, largely due to its powerful capabilities in modeling complex interactions between nodes. A fundamental and crucial requirement for representation learning in CTDGs is the appropriate estimation and preservation of proximity. However, due to the sparse and evolving characteristics of CTDGs, the spatial-temporal properties inherent in high-order proximity remain largely unexplored. Despite its importance, this property presents significant challenges due to the computationally intensive nature of personalized interaction intensity estimation and the dynamic attributes of CTDGs. To this end, we propose a novel Correlated Spatial-Temporal Positional encoding that incorporates a parameter-free personalized interaction intensity estimation under the weak assumption of the Poisson Point Process. Building on this, we introduce the Dynamic Graph Transformer with Correlated Spatial-Temporal Positional Encoding (CorDGT), which efficiently retains the evolving spatial-temporal high-order proximity for effective node representation learning in CTDGs. Extensive experiments on seven small and two large-scale datasets demonstrate the superior performance and scalability of the proposed CorDGT. The code is available at: https://github.com/wangz3066/CorDGT.

Recommender Systems (RS) aim to generate personalized ranked lists for each user and are evaluated using ranking metrics. Although personalized ranking is a fundamental aspect of RS, this critical property is often overlooked in the design of model architectures. To address this issue, we propose Rankformer, a ranking-inspired recommendation model. The architecture of Rankformer is inspired by the gradient of the ranking objective, embodying a unique (graph) transformer architecture --- it leverages global information from all users and items to produce more informative representations and employs specific attention weights to guide the evolution of embeddings towards improved ranking performance. We further develop an acceleration algorithm for Rankformer, reducing its complexity to a linear level with respect to the number of positive instances. Extensive experimental results demonstrate that Rankformer outperforms state-of-the-art methods. The code is available at https://github.com/StupidThree/Rankformer.

Graph Neural Networks (GNNs) have become a prominent approach for learning from graph-structured data. However, their effectiveness can be significantly compromised when the graph structure is suboptimal. To address this issue, Graph Structure Learning (GSL) has emerged as a promising technique that refines node connections adaptively. Nevertheless, we identify two key limitations in existing GSL methods: 1) Most methods primarily focus on node similarity to construct relationships, while overlooking the quality of node information. Blindly connecting low-quality nodes and aggregating their ambiguous information can degrade the performance of other nodes. 2) The constructed graph structures are often constrained to be symmetric, which may limit the model's flexibility and effectiveness.

Memory-based temporal graph neural network (MTGNN) models are effective for predicting temporal graphs by using node memory and message-passing modules to capture temporal and structural information, respectively. However, distributed training for large graphs presents challenges such as accuracy loss and decreased efficiency due to remote features and memory transmission. Despite improvements in MTGNN system optimizations, issues like dynamic load imbalances, communication overhead, and memory staleness persist. To tackle these challenges, we introduce MemShare, a distributed MTGNN system. MemShare introduces a novel shared node memory paradigm that utilizes a small subset of shared nodes across machines and GPUs to reduce distributed communication for memory management. It incorporates techniques like shared nodes-centric graph partitioning, shared nodes-aware boundary decay sampling, and shared nodes-targeted synchronous smoothing aggregation. Experiments show that MemShare outperforms existing distributed MTGNN systems in accuracy and training efficiency.

Knowledge Graph Embedding (KGE), which projects entities and relations into continuous vector spaces, has garnered significant attention. Although high-dimensional KGE methods offer better performance, they come at the expense of significant computation and memory overheads. Decreasing embedding dimensions significantly deteriorates model performance. While several recent efforts utilize knowledge distillation or non-Euclidean representation learning to augment the effectiveness of low-dimensional KGE, they either necessitate a pre-trained high-dimensional teacher model or involve complex non-Euclidean operations, thereby incurring considerable additional computational costs. To address this, this work proposes Confidence-aware Self-Knowledge Distillation (CSD) that learns from the model itself to enhance KGE in a low-dimensional space. Specifically, CSD extracts knowledge from embeddings in previous iterations, which would be utilized to supervise the learning of the model in the next iterations. Moreover, a specific semantic module is developed to filter reliable knowledge by estimating the confidence of previously learned embeddings. This straightforward strategy bypasses the need for time-consuming pre-training of teacher models and can be integrated into various KGE methods to improve their performance. Our comprehensive experiments on six KGE backbones and four datasets underscore the effectiveness of the proposed CSD.

Given the ubiquity of Uncertain Graphs (UGs), the field of UG mining has garnered increasing attention. Among various mining tasks, query processing stands out as the most fundamental and crucial. Current methods for query answering on UGs primarily rely on Monte-Carlo sampling and heuristic approaches. However, these techniques either struggle with a significant efficiency-accuracy trade-off or lack generalization over different graphs and queries. To circumvent these limitations, this work proposes a novel index-based method for query answering on UGs. We construct a labeling index framework, which can answer queries by pre-computed and stored operators. To the best of our knowledge, this is the first index frame-work that can deal with reliability, expected reliable distance and distance-constrained reliability queries, providing lower or upper bounded query answer results. By transferring the time consuming sampling process into the offline index operator computation, the query answering only needs to traverse a limited number of operators, which accelerates the response time of query answering with several orders of magnitude. We further utilize the vertex cover and its h-hop extension to prune the index structure, thereby reducing the space complexity. Experimental results on five real-world datasets demonstrate that the proposed index framework is both effective and efficient.

In recommender systems, most graph-based methods focus on positive user feedback, while overlooking the valuable negative feedback. Integrating both positive and negative feedback to form a signed graph can lead to a more comprehensive understanding of user preferences. However, the existing efforts to incorporate both types of feedback are sparse and face two main limitations: 1) They process positive and negative feedback separately, which fails to holistically leverage the collaborative information within the signed graph; 2) They rely on MLPs or GNNs for information extraction from negative feedback, which may not be effective. To overcome these limitations, we introduceSIGformer, a new method that employs the transformer architecture to sign-aware graph-based recommendation. SIGformer incorporates two innovative positional encodings that capture the spectral properties and path patterns of the signed graph, enabling the full exploitation of the entire graph. Our extensive experiments across five real-world datasets demonstrate the superiority of SIGformer over state-of-the-art methods. The code is available at https://github.com/StupidThree/SIGformer.

With the capacity to capture high-order collaborative signals, Graph Neural Networks (GNNs) have emerged as powerful methods in Recommender Systems (RS). However, their efficacy often hinges on the assumption that training and testing data share the same distribution (\aka IID assumption), and exhibits significant declines under distribution shifts. Distribution shifts commonly arises in RS, often attributed to the dynamic nature of user preferences or ubiquitous biases during data collection in RS. Despite its significance, researches on GNN-based recommendation against distribution shift are still sparse. To bridge this gap, we propose Distributionally Robust GNN (DR-GNN) that incorporates Distributional Robust Optimization (DRO) into the GNN-based recommendation. DR-GNN addresses two core challenges: 1) To enable DRO to cater to graph data intertwined with GNN, we reinterpret GNN as a graph smoothing regularizer, thereby facilitating the nuanced application of DRO; 2) Given the typically sparse nature of recommendation data, which might impede robust optimization, we introduce slight perturbations in the training distribution to expand its support. Notably, while DR-GNN involves complex optimization, it can be implemented easily and efficiently. Our extensive experiments validate the effectiveness of DR-GNN against three typical distribution shifts. The code is available at https://github.com/WANGBohaO-jpg/DR-GNN.

Implicit feedback (e.g., user clicks) is widely used in building recommender systems (RS). However, the inherent notorious exposure bias significantly affects recommendation performance. Exposure bias refers a phenomenon that implicit feedback is influenced by user exposure and does not precisely reflect user preference. Current methods for addressing exposure bias primarily reduce confidence in unclicked data, employ exposure models, or leverage propensity scores. Regrettably, these approaches often lead to biased estimations or elevated model variance, yielding sub-optimal results. To overcome these limitations, we propose a new method ReCRec that Reasons the C auses behind the implicit feedback for debiased R ec ommendation . ReCRec identifies three scenarios behind unclicked data—i.e., unexposed, dislike, or a combination of both. A reasoning module is employed to infer the category to which each instance pertains. Consequently, the model is capable of extracting reliable positive and negative signals from unclicked data, thereby facilitating more accurate learning of user preferences. We also conduct thorough theoretical analyses to demonstrate the debiased nature and low variance of ReCRec. Extensive experiments on both semi-synthetic and real-world datasets validate its superiority over state-of-the-art methods.

In recommendation systems (RS), user behavior data is observational rather than experimental, resulting in widespread bias in the data. Consequently, tackling bias has emerged as a major challenge in the field of recommendation systems. Recently, Doubly Robust Learning (DR) has gained significant attention due to its remarkable performance and robust properties. However, our experimental findings indicate that existing DR methods are severely impacted by the presence of so-called Poisonous Imputation, where the imputation significantly deviates from the truth and becomes counterproductive.

Knowledge distillation is a technique to enhance the generalization ability of a student model by exploiting outputs from a teacher model. Recently, feature-map based variants explore knowledge transfer between manually assigned teacher-student pairs in intermediate layers for further improvement. However, layer semantics may vary in different neural networks, resulting in performance degeneration due to negative regularization from semantic mismatch in manual layer associations. To address this issue, we propose semantic calibration for cross-layer knowledge distillation (SemCKD), which automatically assigns proper target layers of the teacher model for each student layer with an attention mechanism. With a learned attention distribution, each student layer distills knowledge contained in multiple teacher layers rather than a specific intermediate layer for appropriate cross-layer supervision. We further provide theoretical analysis of the association weights and conduct extensive experiments to demonstrate the effectiveness of our approach. On average, SemCKD improves the student Top-1 classification accuracy by 4.27% across twelve different teacher-student model combinations on CIFAR-100. Code is available athttps://github.com/DefangChen/SemCKD.

Recommendation fromimplicit feedbackis a highly challenging task due to the lack of reliable negative feedback data. Existing methods address this challenge by treating all the un-observed data as negative (dislike) but downweight the confidence of these data. However, this treatment causes two problems: (1) Confidence weights of the unobserved data are usually assigned manually, which lack flexibility and may create empirical bias on evaluating user's preference. (2) To handle massive volume of the unobserved feedback data, most of the existing methods rely on stochastic inference and data sampling strategies. However, since a user is only aware of a very small fraction of items in a large dataset, it is difficult for existing samplers to selectinformativetraining instances in which the user really dislikes the item rather than does not know it. To address the above two problems, we propose two novel recommendation methods SamWalker and SamWalker++ that support both adaptive confidence assignment and efficient model learning. SamWalker models data confidence with a social network-aware function, which can adaptively specify different weights to different data according to users’social contexts. However, the social network information may not be available in many recommender systems, which hinders application of SamWalker. Thus, we further propose SamWalker++, which does not require any side information and models data confidence with a constructed pseudo-social network. In the pseudo-social network, similar users are connected with specific item nodes or community nodes. This way, the inference of one's data confidence can benefit from the knowledge from other similar users. We also develop fast random-walk-based sampling strategies for our SamWalker and SamWalker++ to adaptively draw informative training instances, which can speed up gradient estimation and reduce sampling variance. Extensive experiments on five real-world datasets demonstrate the superiority of the proposed SamWalker and SamWalker++.

Graph Neural Networks (GNNs), which generalize deep neural networks to graph-structured data, have drawn considerable attention and achieved state-of-the-art performance in numerous graph related tasks. However, existing GNN models mainly focus on designing graph convolution operations. The graph pooling (or downsampling) operations, that play an important role in learning hierarchical representations, are usually overlooked. In this paper, we proposed a novel multi-view graph pooling operator dubbed as MVPool, which ranks nodes across different views with different contextual graph information. Meanwhile, attention mechanism is utilized to promote the collaboration of different views for generating robust node rankings. Then the pooling operation adaptively selects a subset of nodes to form an induced subgraph based on the ranking list. To preserve the underlying graph topological information, we further introduce a structure learning mechanism to learn a refined graph structure for the pooled graph at each layer. The proposed MVPool operator is a general strategy that can be integrated into various graph neural network architectures. By combining MVPool operator with graph neural networks, we perform hierarchical representation learning for both node and graph level classification as well as clustering tasks. Experimental results on nine widely used benchmarks demonstrate the effectiveness of our proposed model.

Learning low-dimensional representations for Heterogeneous Information Networks (HINs) has drawn increasing attention recently for its effectiveness in real-world applications. Compared with homogeneous networks, HINs are characterized by meta-paths connecting different types of nodes with semantic meanings. Existing methods mainly follow the prototype of independently learning meta-path-based embeddings and integrating them into a unified embedding. However, meta-paths in a HIN are inherently correlated since they reflect different perspectives of the same object. If each meta-path is treated as an isolated semantic data resource and the correlations among them are disregarded, sub-optimality in the both the meta-path based embedding and final embedding will be resulted. To address this issue, we make the first attempt to explicitly model the correlation among meta-paths by proposing Collaborative Knowledge Distillation for Heterogeneous Information Network Embedding (CKD). More specifically, we model the knowledge in each meta-path with two different granularities: regional knowledge and global knowledge. We learn the meta-path-based embeddings by collaboratively distill the knowledge from intra-meta-path and inter-meta-path simultaneously. Experiments conducted on six real-world HIN datasets demonstrates the effectiveness of the CKD method.

User recommendation aims at recommending users with potential interests in the social network. Previous works have mainly focused on the undirected social networks with symmetric relationship such as friendship, whereas recent advances have been made on the asymmetric relationship such as the following and followed by relationship. Among the few existing direction-aware user recommendation methods, the random walk strategy has been widely adopted to extract the asymmetric proximity between users. However, according to our analysis on real-world directed social networks, we argue that the asymmetric proximity captured by existing random walk based methods are insufficient due to the inbalance in-degree and out-degree of nodes. To tackle this challenge, we propose InfoWalk, a novel informative walk strategy to efficiently capture the asymmetric proximity solely based on random walks. By transferring the direction information into the weights of each step, InfoWalk is able to overcome the limitation of edges while simultaneously maintain both the direction and proximity. Based on the asymmetric proximity captured by InfoWalk, we further propose the qualitative (DNE-L) and quantitative (DNE-T) directed network embedding methods, capable of preserving the two properties in the embedding space. Extensive experiments conducted on six real-world benchmark datasets demonstrate the superiority of the proposed DNE model over several state-of-the-art approaches in various tasks.

Graph similarity learning, which measures the similarities between a pair of graph-structured objects, lies at the core of various machine learning tasks such as graph classification, similarity search, etc. In this paper, we devise a novel graph neural network based framework to address this challenging problem, motivated by its great success in graph representation learning. As the vast majority of existing graph neural network models mainly concentrate on learning effective node or graph level representations of a single graph, little effort has been made to jointly reason over a pair of graph-structured inputs for graph similarity learning. To this end, we propose Hierarchical Hypergraph Matching Networks (H2sup>MN) to calculate the similarities between graph pairs with arbitrary structure. Specifically, our proposed H2MN learns graph representation from the perspective of hypergraph, and takes each hyperedge as a subgraph to perform subgraph matching, which could capture the rich substructure similarities across the graph. To enable hierarchical graph representation and fast similarity computation, we further propose a hyperedge pooling operator to transform each graph into a coarse graph of reduced size. Then, a multi-perspective cross-graph matching layer is employed on the coarsened graph pairs to extract the inter-graph similarity. Comprehensive experiments on five public datasets empirically demonstrate that our proposed model can outperform state-of-the-art baselines with different gains for graph-graph classification and regression tasks.

Sampling strategies have been widely applied in many recommendation systems to accelerate model learning from implicit feedback data. A typical strategy is to draw negative instances with uniform distribution, which, however, will severely affect a model’s convergence, stability, and even recommendation accuracy. A promising solution for this problem is to over-sample the “difficult” (a.k.a. informative) instances that contribute more on training. But this will increase the risk of biasing the model and leading to non-optimal results. Moreover, existing samplers are either heuristic, which require domain knowledge and often fail to capture real “difficult” instances, or rely on a sampler model that suffers from low efficiency. To deal with these problems, we propose CoSam, an efficient and effective collaborative sampling method that consists of (1) a collaborative sampler model that explicitly leverages user-item interaction information in sampling probability and exhibits good properties of normalization, adaption, interaction information awareness, and sampling efficiency, and (2) an integrated sampler-recommender framework, leveraging the sampler model in prediction to offset the bias caused by uneven sampling. Correspondingly, we derive a fast reinforced training algorithm of our framework to boost the sampler performance and sampler-recommender collaboration. Extensive experiments on four real-world datasets demonstrate the superiority of the proposed collaborative sampler model and integrated sampler-recommender framework.

Graph embedding has become the research focus in both academic and industrial communities due to its powerful capabilities. The majority of existing work overwhelmingly learn node embeddings in the context of static, plain or attributed, homogeneous graphs. However, many real-world applications frequently involve bipartite graphs with temporal and attributed interaction edges, named temporal interaction graphs. The temporal interactions usually imply different facets of interest and might even evolve over time, thus putting forward huge challenges in learning effective node representations. In this paper, we propose a novel framework named TigeCMN to learn node representations from a sequence of temporal interactions. Specifically, we devise two coupled memory networks to store and update node embeddings in external matrices explicitly and dynamically, which forms deep matrix representations and could enhance the expressiveness of the node embeddings. We conduct experiments on two real-world datasets and the experimental results empirically demonstrate that TigeCMN can outperform the state-of-the-arts with different gains.

The e-commerce era is witnessing a rapid development of various annual online promotions, such as Black Friday, Cyber Monday, and Alibaba's 11.11, etc. S ales P redictions for O nline P romotions (SPOP) are a set of sales related forecasts for the promotion day, including gross merchandise volume, sales volume, best selling products, etc. SPOP is highly important for e-commerce platforms to efficiently organize merchandise and maximize business values. However, sales patterns during the promotions are varied according to different scenarios, each model of which is designed with different features, static or dynamic, for one task in particular. Therefore, several models are proposed with part of features that are possibly beneficial to other tasks, which indicates the universal representation for the items needs to be learned across different promotion scenarios. To address this problem, this paper proposes a D eep I tem N etwork for O nline P romotions (DINOP). In DINOP, we design a novel T arget U sers C ontrolled G ated R ecurrent U nit (TUC-GRU) structure for dynamic features, and provide a new attention mechanism introducing static users profiles. In contrast to traditional prediction models, the network we proposed can effectively and efficiently learn universal item representation by incorporating users' properties as controllers. Furthermore, it can successfully discover the static and dynamic features guided by the multi-task learning, and is easily extended to other sales related prediction problems without retraining. Empirical results show that performance of DINOP in the real data set of Alibaba's Global Shopping Festival is superior to other state-of-the-arts practical methodologies in terms of the convergence rate and prediction accuracy.

Spread of negative influence (N-Inf) in a networked system seems to be inevitable, e.g., epidemic spread in human networks, rumors in an online social network and computer virus plaguing the Internet etc. The widespread of N-Inf might cause severe damage and hence the Influence Blocking (IB) problem is attracting ample research interest. The IB problem aims at minimizing the N-Inf spread by immunization, i.e. selecting k (budget size) immunization nodes (Imm-nodes) to prevent the N-Inf from spreading. However, existing works for IB problem are all formulated as a one-shot task: selecting all the k Imm-nodes at the very beginning of N-Inf spread. In real world, unforeseen events might occur and one-shot policies will lack reserved measures to handle these situations. A more reasonable policy is to adaptively invest the budget based on the observation of N-Inf spread along as the time goes by. With the adaptive policy, we can both reserve resources for handling unforeseen events and save unnecessary costs if the spread of N-Inf dies out quickly. Motivated by the above considerations, we propose a novel Adaptive Influence Blocking (AIB) problem. Given the intermediate observations of N-Inf spread, the AIB problem aims at selecting Imm-nodes adaptively. We design a k-R (k-nodes-per-Round) policy which selects k Imm-nodes for each round until the budget is exhausted, and an α-T (α-Tolerance) policy which selects a new Imm-node if the expected N-Inf spread exceeds a threshold α. Scalable algorithms with provable approximation guarantees and error bounds are implemented for these policies and significant improvements on time complexity are achieved. Experimental results on real-world datasets demonstrate the effectiveness and scalability of the proposed methods.

Recommendation from implicit feedback is a highly challenging task due to the lack of reliable negative feedback data. Only positive feedback are observed and the unobserved feedback can be attributed to two reasons: unknow or dislike. Existing methods address this challenge by treating all the un-observed data as negative (dislike) but downweight the confidence of these data. However, this treatment causes two problems: (1) Confidence weights of the unobserved data are usually assigned manually, which lack flexible and may create empirical bias in evaluating user's preference. (2) To handle massive volume of the unobserved feedback data, most of the existing methods rely on stochastic inference and data sampling strategies. However, since users are only aware of a very small fraction of items in a large dataset, it is difficult for existing samplers to select informative training instances in which the user really dislikes the item rather than does not know it.

Users' consumption behaviors are affected by both their personal preference and their exposure to items (i.e. whether a user knows the items).Most of the recent works in social recommendation assume that people share similar preference with their socially connected friends. However, this assumption may not hold due to the diversity of social relations, and modeling social influence on users' preference may not be suitable for implicit feedback data (i.e. whether a user has consumed certain items). Since users often share item information with their social relations, it will be less restrictive to model social influence on users' exposure to items. We notice that a user's exposure is affected by the exposure of the other users in his social communities and by the consumption of his connected friends. In this paper, we propose a novel social exposure-based recommendation model SoEXBMF by integrating two kinds of social influence on users' exposure, i.e. social knowledge influence and social consumption influence, into basic EXMF model for better recommendation performance. Furthermore, SoEXBMF uses Bernoulli distribution instead of Gaussian distribution in EXMF to better model the binary implicit feedback data. A variational inference method has been developed for the proposed SoEXBMF model to infer the posterior and make the recommendations. Extensive experiments on three real-world datasets demonstrate the superiority of our method over existing methods in various evaluation metrics.

Attributed network embedding focuses on learning low-dimensional latent representations of nodes which can well preserve the original topological and node attributed proximity at the same time. Existing works usually assume that nodes with similar topology or similar attributes should also be close in the embedding space. This assumption ignores the phenomenon of partial correlation between network topological and node attributed similarities i.e. nodes with similar topology may be dissimilar in their attributes and vice versa. Partial correlation between the two information sources should be considered especially when there exist fraudulent edges (i.e., information from one source is vague) or unbalanced data distributions (i.e, topology structure similarity and node attribute similarity have different distributions). However, it is very challenging to consider the partial correlation between topology and attributes due to the heterogeneity of these two information sources. In this paper, we take partial correlation between topology and attributes into account and propose the Personalized Relation Ranking Embedding (PRRE) method for attributed networks which is capable of exploiting the partial correlation between node topology and attributes. The proposed PRRE model utilizes two thresholds to define different node relations and employs the Expectation-Maximization (EM) algorithm to learn these thresholds as well as other embedding parameters. Extensive experiments results on multiple real-world datasets show that the proposed PRRE model significantly outperforms the state-of-the-art methods in terms of various evaluation metrics.

With the explosive growth of online social networks, many social recommendation methods have been proposed and demonstrated that social information has potential to improve the recommendation performance. However, existing social recommendation methods always assume that the data is missing at random (MAR) but this is rarely the case. In fact, by analysing two real-world social recommendation datasets, we observed the following interesting phenomena: (1) users tend to consume and rate the items that they like and the items that have been consumed by their friends. (2) When the items have been consumed by more friends, the average values of the observed ratings will become smaller, not larger as assumed by the existing models. To model these phenomena, we integrate the missing not at random (MNAR) assumption in social recommendation and propose a new social recommendation method SPMF-MNAR, which models the observation process of rating data based on user's preference and social influence. Extensive experiments conducted on large real-world datasets validate that SPMF-MNAR achieves better performance than existing social recommendation methods and the non-social methods based on MNAR assumption.

The e-commerce era is witnessing a rapid increase of mobile Internet users. Major e-commerce companies nowadays see billions of mobile accesses every day. Hidden in these records are valuable user behavioral characteristics such as their shopping preferences and browsing patterns. And, to extract these knowledge from the huge dataset, we need to first link records to the corresponding mobile devices. This Mobile Access Records Resolution (MARR) problem is confronted with two major challenges: (1) device identifiers and other attributes in access records might be missing or unreliable; (2) the dataset contains billions of access records from millions of devices. To the best of our knowledge, as a novel challenge industrial problem of mobile Internet, no existing method has been developed to resolve entities using mobile device identifiers in such a massive scale. To address these issues, we propose a SParse Identifier-linkage Graph (SPI-Graph) accompanied with the abundant mobile device profiling data to accurately match mobile access records to devices. Furthermore, two versions (unsupervised and semi-supervised) of Parallel Graph-based Record Resolution (PGRR) algorithm are developed to effectively exploit the advantages of the large-scale server clusters comprising of more than 1,000 computing nodes. We empirically show superior performances of PGRR algorithms in a very challenging and sparse real data set containing 5.28 million nodes and 31.06 million edges from 2.15 billion access records compared to other state-of-the-arts methodologies.

Existing works for extracting navigation objects from webpages focus on navigation menus, so as to reveal the information architecture of the site. However, web 2.0 sites such as social networks, e-commerce portals etc. are making the understanding of the content structure in a web site increasingly difficult. Dynamic and personalized elements such as top stories, recommended list in a webpage are vital to the understanding of the dynamic nature of web 2.0 sites. To better understand the content structure in web 2.0 sites, in this paper we propose a new extraction method for navigation objects in a webpage. Our method will extract not only the static navigation menus, but also the dynamic and personalized page-specific navigation lists. Since the navigation objects in a webpage naturally come in blocks, we first cluster hyperlinks into different blocks by exploiting spatial locations of hyperlinks, the hierarchical structure of the DOM-tree and the hyperlink density. Then we identify navigation objects from those blocks using the SVM classifier with novel features such as anchor text lengths etc. Experiments on real-world data sets with webpages from various domains and styles verified the effectiveness of our method.

With the explosive growth of online social networks, it is now well understood that social information is highly helpful to recommender systems. Social recommendation methods are capable of battling the critical cold-start issue, and thus can greatly improve prediction accuracy. The main intuition is that through trust and influence, users are more likely to develop affinity toward items consumed by their social ties. Despite considerable work in social recommendation, little attention has been paid to the important distinctions between strong and weak ties, two well-documented notions in social sciences. In this work, we study the effects of distinguishing strong and weak ties in social recommendation. We use neighbourhood overlap to approximate tie strength and extend the popular Bayesian Personalized Ranking (BPR) model to incorporate the distinction of strong and weak ties. We present an EM-based algorithm that simultaneously classifies strong and weak ties in a social network w.r.t. optimal recommendation accuracy and learns latent feature vectors for all users and all items. We conduct extensive empirical evaluation on four real-world datasets and demonstrate that our proposed method significantly outperforms state-of-the-art pairwise ranking methods in a variety of accuracy metrics.

Graph-based ranking models have been widely applied in information retrieval area. In this paper, we focus on a well known graph-based model - the Ranking on Data Manifold model, or Manifold Ranking (MR). Particularly, it has been successfully applied to content-based image retrieval, because of its outstanding ability to discover underlying geometrical structure of the given image database. However, manifold ranking is computationally very expensive, which significantly limits its applicability to large databases especially for the cases that the queries are out of the database (new samples). We propose a novel scalable graph-based ranking model called Efficient Manifold Ranking (EMR), trying to address the shortcomings of MR from two main perspectives: scalable graph construction and efficient ranking computation. Specifically, we build an anchor graph on the database instead of a traditional$k$-nearest neighbor graph, and design a new form of adjacency matrix utilized to speed up the ranking. An approximate method is adopted for efficient out-of-sample retrieval. Experimental results on some large scale image databases demonstrate that EMR is a promising method for real world retrieval applications.

Nowadays, micro-blogging systems like Twitter have become one of the most important ways for information sharing. In Twitter, a user posts a message (tweet) and the others can forward the message (retweet). Mention is a new feature in micro-blogging systems. By mentioning users in a tweet, they will receive notifications and their possible retweets may help to initiate large cascade diffusion of the tweet. To enhance a tweet's diffusion by finding the right persons to mention, we propose in this paper a novel recommendation scheme named as whom-to-mention. Specifically, we present an in-depth study of mention mechanism and propose a recommendation scheme to solve the essential question of whom to mention in a tweet. In this paper, whom-to-mention is formulated as a ranking problem and we try to address several new challenges which are not well studied in the traditional information retrieval tasks. By adopting features including user interest match, content-dependent user relationship and user influence, a machine learned ranking function is trained based on newly defined information diffusion based relevance. The extensive evaluation using data gathered from real users demonstrates the advantage of our proposed algorithm compared with the traditional recommendation methods.

People are more and more willing to participate in online forums to share their knowledge and experience. However, it may not be easy for them to find their desired threads in online forums due to the information overload problem. Traditional recommendation approaches can not be directly applied to online forums due to two reasons. First, unlike traditional movie or music recommendation problem, there is no rating information in online forums. Second, the sparsity problem is more severe since the users may only read threads but take no actions. To address these limitations, in this paper we propose to make use of the reply relationships among users, as well as thread contents. A learning algorithm is introduced to infer a user-thread alignment manifold in which both users and thread contents can be well represented. Thus, the relatedness between users and threads can be measured on this alignment manifold, and the closest threads which can best meet the corresponding user's information needs are recommended. Experiments on a dataset crawled from digg.com have demonstrated the superiority of our algorithm over traditional recommendation algorithms.

Social tagging services allow users to annotate various on-line resources with freely chosen keywords (tags). They not only facilitate the users in finding and organizing online re-sources, but also provide meaningful collaborative semantic data which can potentially be exploited by recommender systems. Traditional studies on recommender systems fo-cused on user rating data, while recently social tagging data is becoming more and more prevalent. How to perform re-source recommendation based on tagging data is an emerg-ing research topic. In this paper we consider the problem of document (e.g. Web pages, research papers) recommen-dation using purely tagging data. That is, we only have data containing users, tags, documents and the relation-ships among them. We propose a novel graph-based rep-resentation learning algorithm for this purpose. The users, tags and documents are represented in the same semantic space in which two related objects are close to each other. For a given user, we recommend those documents that are sufficiently close to him/her. Experimental results on two data sets crawled from Del.icio.us and CiteULike show that our algorithm can generate promising recommendations and outperforms traditional recommendation algorithms.

Automatic news extraction from news pages is important in many Web applications such as news aggregation. However, the existing news extraction methods based on template-level wrapper induction have three serious limitations. First, the existing methods cannot correctly extract pages belonging to an unseen template. Second, it is costly to maintain up-to-date wrappers for a large amount of news websites, because any change of a template may invalidate the corresponding wrapper. Last, the existing methods can merely extract unformatted plain texts, and thus are not user friendly. In this paper, we tackle the problem of template-independent Web news extraction in a user-friendly way. We formalize Web news extraction as a machine learning problem and learn a template-independent wrapper using a very small number of labeled news pages from a single site. Novel features dedicated to news titles and bodies are developed. Correlations between news titles and news bodies are exploited. Our template-independent wrapper can extract news pages from different sites regardless of templates. Moreover, our approach can extract not only texts, but also images and animates within the news bodies and the extracted news articles are in the same visual style as in the original pages. In our experiments, a wrapper learned from 40 pages from a single news site achieved an accuracy of 98.1% on 3,973 news pages from 12 news sites.

Social tagging is becoming increasingly popular in many Web 2.0 applications where users can annotate resources (e.g. Web pages) with arbitrary keywords (i.e. tags). A tag recommendation module can assist users in tagging process by suggesting relevant tags to them. It can also be directly used to expand the set of tags annotating a resource. The benefits are twofold: improving user experience and enriching the index of resources. However, the former one is not emphasized in previous studies, though a lot of work has reported that different users may describe the same concept in different ways. We address the problem of personalized tag recommendation for text documents. In particular, we model personalized tag recommendation as a "query and ranking" problem and propose a novel graph-based ranking algorithm for interrelated multi-type objects. When a user issues a tagging request, both the document and the user are treated as a part of the query. Tags are then ranked by our graph-based ranking algorithm which takes into consideration both relevance to the document and preference of the user. Finally, the top ranked tags are presented to the user as suggestions. Experiments on a large-scale tagging data set collected from Del.icio.us have demonstrated that our proposed algorithm significantly outperforms algorithms which fail to consider the diversity of different users' interests.

We consider the problem of template-independent news extraction. The state-of-the-art news extraction method is based on template-level wrapper induction, which has two serious limitations. 1) It cannot correctly extract pages belonging to an unseen template until the wrapper for that template has been generated. 2) It is costly to maintain up-to-date wrappers for hundreds of websites, because any change of a template may lead to the invalidation of the corresponding wrapper. In this paper we formalize news extraction as a machine learning problem and learn a template-independent wrapper using a very small number of labeled news pages from a single site. Novel features dedicated to news titles and bodies are developed respectively. Correlations between the news title and the news body are exploited. Our template-independent wrapper can extract news pages from different sites regardless of templates. In experiments, a wrapper is learned from 40 pages from a single news site. It achieved 98.1% accuracy over 3,973 news pages from 12 news sites.

Social annotations on a Web document are highly generalized description of topics contained in that page. Their tagged frequency indicates the user attentions with various degrees. This makes annotations a good resource for summarizing multiple topics in a Web page. In this paper, we present a tag-oriented Web document summarization approach by using both document content and the tags annotated on that document. To improve summarization performance, a new tag ranking algorithm named EigenTag is proposed in this paper to reduce noise in tags. Meanwhile, association mining technique is employed to expand tag set to tackle the sparsity problem. Experimental results show our tag-oriented summarization has a significant improvement over those not using tags.

Focused crawling is a critical technique for topical resource discovery on the Web. We propose a new frontier prioritizing algorithm, namely, the OTIE (On-line Topical Importance Estimation) algorithm, which efficiently and effectively combines link-based and content-based analysis to evaluate the priority of an uncrawled URL in the frontier. We then demonstrate OTIE's advantages over traditional prioritizing algorithms by real crawling experiments.

Blogs have been expanded at an incredible speed in recent years. Plentiful personal information makes blogs a popular way mining user profiles. In this paper, we propose a novel bloggers' interests modeling approach based on forgetting mechanism. A new forgetting function is introduced to track interest drift. Based on that, the Short Term Interest Models (STIM) and Long Term Interest Models (LTIM) are constructed to describe bloggers' short-term and long-term interests. The experiments show that both models can identify bloggers' preferences well respectively.