With the widespread adoption of large vision-language models, the capacity for color vision in these models is crucial. However, the color vision abilities of large visual-language models have not yet been thoroughly explored. To address this gap, we define a color vision testing task for large vision-language models and construct a dataset that covers multiple categories of test questions and tasks of varying difficulty levels. Furthermore, we analyze the types of errors made by large vision-language models and propose a chain-of-thought prompting strategy to enhance their performance in color vision tests.

While deep neural networks (DNNs) excel in computer vision tasks, their real-world deployment is hindered by robustness limitations compared to human perception. Adversarial attacks and data distribution shifts remain critical vulnerabilities, degrading model performance under practical conditions. To address these challenges and advance robustness research, we introduce a comprehensive, user-friendly toolkit for training, evaluating, and analyzing robust vision models. It targets two key dimensions of robustness: 1) Adversarial robustness-defending against malicious worst-case perturbations (adversarial examples); 2) Natural robustness-maintaining performance under real-world corruptions and distribution shifts. Through extensive image classification benchmarks, our toolkit enables precise model assessment. We envision this toolkit accelerating the development of practically robust models and bridging the gap between machine and human vision capabilities.

Denoising Diffusion Probabilistic Models (DDPMs) have recently demonstrated exceptional performance in generating high-quality data. In this work, we propose CommunityDF, a novel framework that applies DDPMs to the community search problem, which involves identifying subgraphs containing nodes closely related to a given query node. However, three key challenges arise in this context: (I) learning effective node representations from limited examples, (II) discretizing continuous node representations into community members, and (III) reducing the number of diffusion steps without sacrificing performance. To tackle these, CommunityDF introduces several innovations. First, we focus on subgraphs around the query node to reduce interference from unrelated nodes, improve scalability. We then employ a contrastive learning approach, treating node states at different diffusion steps as positive examples and designing various negative sampling strategies to learn high-quality node representations from limited examples. Second, we propose a dynamic thresholding mechanism that effectively converts continuous representations into community members. Finally, we reduce the number of diffusion steps by leveraging the rough communities to initialize the process with rough community structures, which accelerates convergence while maintaining high accuracy. Extensive experiments on seven real-world datasets demonstrate that CommunityDF outperforms existing methods by 16%-47%, establishing it as a state-of-the-art solution for community search. The source code is available at https://github.com/JiazunChen/CommunityDF.

In China, the number of riders in the on-demand delivery industry has surpassed ten million. Ensuring that these riders earn a decent income can enhance their financial security, reduce poverty, and promote social equity and stability. Due to ease of use, lower-cost maintenance and environmental friendliness, electric bicycles (e-bikes) are the primary mode of transportation for delivery riders. However, these riders frequently encounter depleted batteries due to limited capacity and prolonged charging times, necessitating inconvenient swaps or recharges during deliveries. To address this issue, we propose the e-bike Battery Swap-as-a-Service (eBaaS), an innovative battery-swapping system that leverages an intelligent AIoT network for seamless battery swapping at distributed locations across urban areas. eBaaS integrates edge-cloud collaboration, battery resource allocation, battery anomaly detection, and battery range prediction to minimize downtime and reduce unnecessary mileage. While eBaaS's potential benefits are evident, there has been a lack of robust methods to quantify its impact. Thus, we further developed the eBaaS Impact Evaluation Method (EIEM), the first comprehensive model to address this gap. EIEM analyzes data from approximately 260,000 delivery riders and 5 million riding trajectories. Findings indicate that eBaaS reduces average invalid mileage by 6 km and increases the order volume by an average of over 20% daily per e-bike rider. Meanwhile, the annual electricity savings result in a reduction of 2.74 million kilograms of carbon emissions for 260,000 riders. The eBaaS system is therefore significantly beneficial for environmental conservation and sustainable urban development.

Along with the rapid technological and commercial innovation on e-commerce platforms, an increasing number of frauds cause great harm to these platforms. Many frauds are conducted by organized groups of fraudsters for higher efficiency and lower costs, also known as group-based frauds. Despite the high concealment and strong destructiveness of group-based fraud, no existing research can thoroughly exploit the information within the transaction networks of e-commerce platforms for group-based fraud detection. In this work, we analyze and summarize the characteristics of group-based frauds. Based on this, we propose a novel end-to-end semi-supervised Group-based Fraud Detection Network (GFDN) to support such fraud detection in real-world applications. In addition, we introduce a module namedTemporal Group Dynamics Analyzer(TGDA) that strengthens the ability to analyze temporal information on group fraudulent activity. Based on this, we built an enhanced model named TGFDN. Experimental results on large-scale e-commerce datasets from Taobao and Bitcoin trading datasets show our proposed model's superior effectiveness and efficiency for group-based fraud detection on bipartite graphs.

Dynamic graph data mining has gained popularity in recent years due to the rich information contained in dynamic graphs and their widespread use in the real world. Despite the advances in dynamic graph neural networks (DGNNs), the rich information and diverse downstream tasks have posed significant difficulties for the practical application of DGNNs in industrial scenarios. To this end, in this paper, we propose to address them by pre-training and present the Contrastive Pre-Training Method for Dynamic Graph Neural Networks (CPDG). CPDG tackles the challenges of pre-training for DGNNs, including generalization capability and long-short term modeling capability, through a flexible structural-temporal subgraph sampler along with structural-temporal contrastive pre-training schemes. Extensive experiments conducted on both large-scale research and industrial dynamic graph datasets show that CPDG outperforms existing methods in dynamic graph pre-training for various downstream tasks under three transfer settings.

What is the widest community in which a person exercises a strong impact? Although extensive attention has been devoted to searching communities containing given individuals, the problem of finding their unique communities of influence has barely been examined. In this paper, we study the novel problem of Characteristic cOmmunity Discovery (COD) in attributed graphs. Our goal is to identify the largest community, taking into account the query attribute, in which the query node has a significant impact. The key challenge of the COD problem is that it requires evaluating the influence of the query node over a large number of hierarchically structured communities. We first propose a novel compressed COD evaluation approach to accelerate the influence estimation by eliminating redundant computations for overlapping communities. Then, we further devise a local hierarchical reclustering method to alleviate the skewness of hierarchical communities generated by global clustering for a specific query attribute. Extensive experiments confirm the effectiveness and efficiency of our solutions to COD: they find characteristic communities better than existing community search methods by several quality measures and achieve up to 25 x speedups against well-crafted baselines.

Methods based on graph neural networks for solving combinatorial optimization (CO) problems have exhibited promising results in tackling a range of NP-hard problems, eliminating the necessity for reliance on manually created domain knowledge. Existing models including reinforcement learning (RL) framework assume that combinatorial instances in the training set contribute equally during training. Nevertheless, there is considerable variation in the quality of training instances, and the performance of models may suffer from the inclusion of low-quality training instances. This paper expands the current scope of neural solvers for CO problems through the incorporation of curriculum learning (CL). To alleviate the adverse impact of low-quality training instances, we propose CL4CO which utilizes CL strategy, a selective training method, to train models based on the rank of instances' quality in neural Combinatorial Optimization framework. Also, we introduce several candidate topology-aware metrics based on heterophily ratio and evaluation of clustering for the training scheduler. Furthermore, it is noteworthy to emphasize that it has potential to enhance the generalization capacity of RL-based baselines and we give a experimental validation. This enhancement plugin from the fact that CL empowers the acquired RL-based solver to effectively leverage commonly shared features within the same class of CO. Empirically, we conduct a case study on MaxCut, a classical discrete Oil-vector CO, to verify our findings and our results demonstrate that CL4CO is efficient and superiority with good generalization ability.

Electric bicycles (EBs) have gained immense popularity as an environmentally friendly and convenient transportation mode. However, range anxiety remains a major concern for EB users. This paper presents a real-time route planning model focused on predicting the remaining range of EBs. First, we represent the user's interaction data and the real-time battery state as a dynamic graph. Then we propose a novel approach called the Real-Time Electric Bicycle Remaining Range (RtRR) prediction model, which leverages the graph structure and jointly optimizes temporal edge convolution, LSTM, and Transformer models to estimate the remaining EB battery range. Based on the prediction, we can update the optimal cycling routes for users in real-time, considering charging station locations. Extensive evaluations demonstrate that our proposed RtRR model outperforms 9 baseline methods on real-world datasets. The route planning based on RtRR prediction effectively alleviates range anxiety and enhances the user experience. It can be accessed at https://github.com/gu-yongchun/Real-time-E-bike-Route-Planning-with-Battery-Range-Prediction.

Unsupervised Graph Domain Adaptation (UGDA) has emerged as a practical solution to transfer knowledge from a label-rich source graph to a completely unlabelled target graph. However, most methods require a labelled source graph to provide supervision signals, which might not be accessible in the real-world settings due to regulations and privacy concerns. In this paper, we explore the scenario of source-free unsupervised graph domain adaptation, which tries to address the domain adaptation problem without accessing the labelled source graph. Specifically, we present a novel paradigm called GraphCTA, which performs model adaptation and graph adaptation collaboratively through a series of procedures: (1) conduct model adaptation based on node's neighborhood predictions in target graph considering both local and global information; (2) perform graph adaptation by updating graph structure and node attributes via neighborhood contrastive learning; and (3) the updated graph serves as an input to facilitate the subsequent iteration of model adaptation, thereby establishing a collaborative loop between model adaptation and graph adaptation. Comprehensive experiments are conducted on various public datasets. The experimental results demonstrate that our proposed model outperforms recent source-free baselines by large margins.

Abstract Personalized review generation is significant for e-commerce applications, such as providing explainable recommendation and assisting the composition of reviews. With the success of pre-trained language models (PLMs), prompt learning-based approaches have been employed to handle this task. However, the existing approach neglects the historical user-item interactions as well as the diverse semantics of the reviews (including semantically relevant reviews and semantically irrelevant reviews). In this paper, we propose GRAPA, a graph-enhanced prompt learning approach for personalized review generation. Specifically, GRAPA extracts topic-level information for each review to address the semantic diversity of reviews. Moreover, GRAPA employs a heterogeneous graph neural network (GNN) to explore the collaborative information hidden in historical user-item interactions. User and item representations generated by the GNN module as well as their ID embeddings are used as prompts and fed into a PLM to guide the generation process. To alleviate the interference of semantically irrelevant reviews, GRAPA further proposes a contrastive learning module to distinguish them. Experimental results on public datasets show that GRAPA outperforms existing methods by up to 4.3% in BLEU-4 and 5.4% in ROUGE2-F.

Graph pooling technique as the essential component of graph neural networks has gotten increasing attention recently and it aims to learn graph-level representations for the whole graph. Besides, graph pooling is important in graph classification and graph generation tasks. However, current graph pooling methods mainly coarsen a sequence of small-sized graphs to capture hierarchical structures, potentially resulting in the deterioration of the global structure of the original graph and influencing the quality of graph representations. Furthermore, these methods artificially select the number of graph pooling layers for different graph datasets rather than considering each graph individually. In reality, the structure and size differences among graphs necessitate a specific number of graph pooling layers for each graph. In this work, we propose reinforced pooling graph neural networks via adaptive hybrid graph coarsening networks. Specifically, we design a hybrid graph coarsening strategy to coarsen redundant structures of the original graph while retaining the global structure. In addition, we introduce multi-agent reinforcement learning to adaptively perform the graph coarsening process to extract the most representative coarsened graph for each graph, enhancing the quality of graph-level representations. Finally, we design graph-level contrast to improve the preservation of global information in graph-level representations. Extensive experiments with rich baselines on six benchmark datasets show the effectiveness of ReiPool1.

Graph Neural Networks (GNNs) have achieved impressive performance in many tasks on graph data. Recent studies show that they are vulnerable to adversarial attacks. Deliberate and unnoticeable perturbations on topology structure could render them near-useless in applications. How to design effective methods to improve the robustness of GNNs is a crucial problem. To solve this problem, some works attempt to design more robust GNN models, while others attempt to remove perturbations from the poisoned graph. Different from the previous works, this paper proposes a general framework termed asGraphReshapeto enhance the robustness of GNNs via directly correcting the shifted classification boundary of GNN models in the presence of adversarial attacks.GraphReshapeconsists of two modules:locating tractive nodesthat could correct GNNs andreshaping local structureto improve their representations in the latent space. Extensive experiments on four real-world datasets show thatGraphReshapeachieves significant performance gain compared with state-of-the-art baselines against different adversarial attacks.

Recently, ChatGPT, a representative large language model (LLM), has gained considerable attention. Due to their powerful emergent abilities, recent LLMs are considered as a possible alternative to structured knowledge bases like knowledge graphs (KGs). However, while LLMs are proficient at learning probabilistic language patterns and engaging in conversations with humans, they, like previous smaller pre-trained language models (PLMs), still have difficulty in recalling facts while generating knowledge-grounded contents. To overcome these limitations, researchers have proposed enhancing data-driven PLMs with knowledge-based KGs to incorporate explicit factual knowledge into PLMs, thus improving their performance in generating texts requiring factual knowledge and providing more informed responses to user queries. This paper reviews the studies on enhancing PLMs with KGs, detailing existing knowledge graph enhanced pre-trained language models (KGPLMs) as well as their applications. Inspired by existing studies on KGPLM, this paper proposes enhancing LLMs with KGs by developing knowledge graph-enhanced large language models (KGLLMs). KGLLM provides a solution to enhance LLMs’ factual reasoning ability, opening up new avenues for LLM research.

Graph label propagation (LP) is a core component in many downstream applications such as fraud detection, recommendation and image segmentation. In this paper, we proposeGLP, a GPU-based framework to enable efficientLPprocessing on large-scale graphs. By investigating the data processing pipeline in a large e-commerce platform, we have identified two key challenges on integrating GPU-acceleratedLPprocessing to the pipeline: (1) programmability for evolving application logics; (2) demand for real-time performance. Motivated by these challenges, we offer a set of expressive APIs that data engineers can customize and deploy efficientLPalgorithms on GPUs with ease. To achieve better performance, we propose novel GPU-centric optimizations by leveraging the community as well as power-law properties of large graphs. Further, we significantly reduce the expensive data transfer cost between CPUs and GPUs by enablingLPprocessing on compressed graphs. Extensive experiments have confirmed the effectiveness of our proposed approaches over the state-of-the-art GPU methods. Furthermore, our proposed solution supports a real billion-scale graph workload for fraud detection and achieves 13.2× speedup to the current in-house solution running on a high-end multicore machine with compressed graphs.

Incremental maximal clique enumeration (IMCE), which maintains maximal cliques in dynamic graphs, is a fundamental problem in graph analysis. A maximal clique has a solid descriptive power of dense structures in graphs. Real-world graph data is often large and dynamic. Studies on IMCE face significant challenges in the efficiency of incremental batch computation and hybrid edge changes. Moreover, with growing graph sizes, new requirements occur on indexing global maximal cliques and obtaining maximal cliques under specific vertex scope constraints. This work presents a new data structure SOMEi to maintain intermediate maximal cliques during construction. SOMEi serves as a space-efficient index to retrieve scope-constrained maximal cliques on the fly. Based on SOMEi, we design a procedure-oriented IMCE algorithm to deal with hybrid edge changes within a unified algorithm framework. In particular, the algorithm is able to process a large batch of edge changes and significantly improve the average processing time of a single edge change through an efficient pruning strategy. Experimental results on real and synthetic graph data demonstrate that the proposed algorithm outperforms all the baselines and achieves good efficiency through pruning.

Graph neural networks (GNNs) are prone to catastrophic forgetting of past experience in continuous learning scenarios. In this work, we propose a novel method for class-incremental graph learning (CGL) by class prototype construction and augmentation, which can effectively overcome catastrophic forgetting and requires no storage of exemplars (i.e., data-free). Concretely, on the one hand, we construct class prototypes in the embedding space that contain rich topological information of nodes or graphs to represent past data, which are then used for future learning. On the other hand, to boost the adaptability of the model to new classes, we employ class prototype augmentation (PA) to create virtual classes by combining current prototypes. Theoretically, we show that PA can promote the model's adaptation to new data and reduce the inconsistency of old prototypes in the embedding space, therefore further mitigate catastrophic forgetting. Extensive experiments on both node and graph classification datasets show that our method significantly outperforms the existing methods in reducing catastrophic forgetting, and beats the existing methods in most cases in terms of classification accuracy.

Large-scale e-commercial platforms in the real-world usually contain various recommendation scenarios (domains) to meet demands of diverse customer groups. Multi-Domain Recommendation (MDR), which aims to jointly improve recommendations on all domains and easily scales to thousands of domains, has attracted increasing attention from practitioners and researchers. Existing MDR methods usually employ a shared structure and several specific components to respectively leverage reusable features and domain-specific information. However, data distribution differs across domains, making it challenging to develop a general model that can be applied to all circumstances. Additionally, during training, shared parameters often suffer from domain conflict while specific parameters are inclined to overfitting on data sparsity domains. In this paper, we first present a scalable MDR platform served in Taobao that enables to provide services for thousands of domains without specialists involved. To address the problems of MDR methods, we propose a novel model agnostic learning framework, namely MAMDR, for the multi-domain recommendation. Specifically, we first propose a Domain Negotiation (DN) strategy to alleviate the conflict between domains. Then, we develop a Domain Regularization (DR) to improve the generalizability of specific parameters by learning from other domains. We integrate these components into a unified framework and present MAMDR, which can be applied to any model structure to perform multi-domain recommendation. Finally, we present a large-scale implementation of MAMDR in the Taobao application and construct various public MDR benchmark datasets which can be used for following studies. Extensive experiments on both benchmark datasets and industry datasets demonstrate the effectiveness and generalizability of MAMDR.

Unsupervised graph anomaly detection is crucial for various practical applications as it aims to identify anomalies in a graph that exhibit rare patterns deviating significantly from the majority of nodes. Recent advancements have utilized Graph Neural Networks (GNNs) to learn high-quality node representations for anomaly detection by aggregating information from neighborhoods. However, the presence of anomalies may render the observed neighborhood unreliable and result in misleading information aggregation for node representation learning. Selecting the proper neighborhood is critical for graph anomaly detection but also challenging due to the absence of anomaly-oriented guidance and the interdependence with representation learning. To address these issues, we utilize the advantages of reinforcement learning in adaptively learning in complex environments and propose a novel method that incorporates Reinforcement neighborhood selection for unsupervised graph ANomaly Detection (RAND). RAND begins by enriching the candidate neighbor pool of the given central node with multiple types of indirect neighbors. Next, RAND designs a tailored reinforcement anomaly evaluation module to assess the reliability and reward of considering the given neighbor. Finally, RAND selects the most reliable subset of neighbors based on these rewards and introduces an anomaly-aware aggregator to amplify messages from reliable neighbors while diminishing messages from unreliable ones. Extensive experiments on both three synthetic and two real-world datasets demonstrate that RAND outperforms the state-of-the-art methods.

Decision-focused learning (DFL) combines both machine learning and combinatorial optimization so as to enhance the quality of decision-making. In general, DFL adds an optimization layer after the neural network and solves a focused combinatorial optimization problem. The optimization layer is usually based on KKT conditions or surrogate functions. However, the optimizer associated with the DFL is inflexible and requires a large amount of expert knowledge. Furthermore, it has been shown that the optimizer exhibits a lack of robustness and differentiability when confronted with complex tasks, especially in the graph domain. To solve this problem, we study a more generic situation by taking the optimization layer as a black-box operator on graphs. Then, we design a mixed zeroth-order optimization to differentiate the layer. Specifically, we combine DFL with graph neural networks and present a decision-focused graph neural network named ZO4Graph based on a mixed zeroth-order optimizer. Empirically, we conduct extensive experiments compared with two-stage models on community detection. Numerical results demonstrate that our proposed framework outperforms its peers.

Anomaly detection in multi-type event sequences is a crucial and challenging problem with important applications in various domains, including cybersecurity, finance and healthcare. Temporal point process has emerged as a powerful technique for modeling event sequences and has gained considerable attention in the field of anomaly detection. However, existing temporal point process approaches are either inapplicable to multi-type event sequence data or incur the loss of valuable information in subsequences associated with specific event types. To this end, we propose a novel Multiple Hypothesis Testing based Anomaly Detection method (MultiAD) to detect anomalous multi-type event sequences. The basic idea of MultiAD is to capture the underlying distribution of normal sequences using a neural multivariate point process, based on which the original hypothesis testing problem can be converted into a multiple hypothesis testing using the multivariate time rescaling theorem. By conducting multiple hypothesis tests on the time-rescaled subsequences, MultiAD makes full use of the valuable information contained within individual subsequences. Moreover, we claim that the existing test statistic ignores the sequential information of inter-event time intervals and propose new statistics to address this shortcoming. Finally, we employ the kernel method to obtain a smooth estimator of the distribution of the proposed statistics under the null hypothesis. This ensures a more accurate and reliable computation of the p-value, providing robust statistical inference. Extensive experimental results demonstrate that MultiAD significantly outperforms the state-of-the-art methods on both synthetic and real-world data.

Along with the rapid technological and commercial innovation on the e-commerce platforms, there are an increasing number of frauds that bring great harm to these platforms. Many frauds are conducted by organized groups of fraudsters for higher efficiency and lower costs, which are also known as group-based frauds. Despite the high concealment and strong destructiveness of group-based fraud, there is no existing research work that can thoroughly exploit the information within the transaction networks of e-commerce platforms for group-based fraud detection. In this work, we analyze and summarize the characteristics of group-based frauds, based on which we propose a novel end-to-end semi-supervised Group-based Fraud Detection Network (GFDN) to support such fraud detection in real-world applications. Experimental results on large-scale e-commerce datasets from Taobao and Bitcoin trading datasets show the superior effectiveness and efficiency of our proposed model for group-based fraud detection on bipartite graphs.

Graph neural networks (GNNs) are powerful models to handle graph data and can achieve state-of-the-art in many critical tasks including node classification and link prediction. However, existing graph neural networks still face both challenges of over-smoothing and over-squashing based on previous literature. To this end, we propose a new Curvature-based topology-aware Dropout sampling technique named CurvDrop, in which we integrate the Discrete Ricci Curvature into graph neural networks to enable more expressive graph models. Also, this work can improve graph neural networks by quantifying connections in graphs and using structural information such as community structures in graphs. As a result, our method can tackle the both challenges of over-smoothing and over-squashing with theoretical justification. Also, numerous experiments on public datasets show the effectiveness and robustness of our proposed method. The code and data are released in https://github.com/liu-yang-maker/Curvature-based-Dropout.

With the proliferation of pump-and-dump schemes (P&Ds) in the cryptocurrency market, it becomes imperative to detect such fraudulent activities in advance to alert potentially susceptible investors. In this paper, we focus on predicting the pump probability of all coins listed in the target exchange before a scheduled pump time, which we refer to as the target coin prediction task. Firstly, we conduct a comprehensive study of the latest 709 P&D events organized in Telegram from Jan. 2019 to Jan. 2022. Our empirical analysis reveals some interesting patterns of P&Ds, such as that pumped coins exhibit intra-channel homogeneity and inter-channel heterogeneity. Here channel refers a form of group in Telegram that is frequently used to coordinate P&D events. This observation inspires us to develop a novel sequence-based neural network, dubbed SNN, which encodes a channel's P&D event history into a sequence representation via the positional attention mechanism to enhance the prediction accuracy. Positional attention helps to extract useful information and alleviates noise, especially when the sequence length is long. Extensive experiments verify the effectiveness and generalizability of proposed methods. Additionally, we release the code and P&D dataset on GitHub https://github.com/Bayi-Hu/Pump-and-Dump-Detection-on-Cryptocurrency, and regularly update the dataset.

Negative sequential pattern mining (SPM) is an important SPM research topic. Unlike positive SPM, negative SPM can discover events that should have occurred but have not occurred, and it can be used for financial risk management and fraud detection. However, existing methods generally ignore the repetitions of the pattern and do not consider gap constraints, which can lead to mining results containing a large number of patterns that users are not interested in. To solve this problem, this article discovers frequent one-off negative sequential patterns (ONPs). This problem has the following two characteristics. First, the support is calculated under the one-off condition, which means that any character in the sequence can only be used once at most. Second, the gap constraint can be given by the user. To efficiently mine patterns, this article proposes the ONP-Miner algorithm, which employs depth-first and backtracking strategies to calculate the support. Therefore, ONP-Miner can effectively avoid creating redundant nodes and parent-child relationships. Moreover, to effectively reduce the number of candidate patterns, ONP-Miner uses pattern join and pruning strategies to generate and further prune the candidate patterns, respectively. Experimental results show that ONP-Miner not only improves the mining efficiency but also has better mining performance than the state-of-the-art algorithms. More importantly, ONP mining can find more interesting patterns in traffic volume data to predict future traffic.

Graph neural networks (GNNs) are popularly used to analyze non-Euclidean graph data. Despite their successes, the design of graph neural networks requires heavy manual work and rich domain knowledge. Recently, neural architecture search algorithms are widely used to automatically design neural architectures for CNNs and RNNs. Inspired by the success of neural architecture search algorithms, we present a graph neural architecture search algorithm GraphNAS that enables automatic design of the best graph neural architecture based on reinforcement learning. Specifically, GraphNAS uses a recurrent network as the controller to generate variable-length strings that describe the architectures of graph neural networks, and trains the recurrent network with policy gradient to maximize the expected accuracy of the generated architectures on a validation data set. Moreover, based on GraphNAS, we design a new GraphNAS++ model using distributed neural architecture search. Compared with GraphNAS that generates and evaluates only one candidate architecture at each iteration, GraphNAS++ generates a mini-batch of candidate architectures and evaluates them in a distributed computing environment until convergence. Experiments on real-world datasets demonstrate that GraphNAS can design a novel network architecture that rivals the best human-invented architecture. Moreover, GraphNAS++ can speed up the design process at least five times by using the distributed training framework with GPUs.

Heterogeneous graphs are commonly used to describe networked data with multiple types of nodes and edges. Heterogeneous Graph Neural Networks (HGNNs) are powerful tools for analyzing heterogeneous graphs. However, designing neural architectures of HGNNs requires extensive domain knowledge and time-consuming manual work. Recently, neural architecture search algorithms have become popular in automatically designing neural architectures for homogeneous graph neural networks. In this paper, we present a Heterogeneous Graph Neural Architecture Search algorithm (HGNAS for short) which allows the automatic design of heterogeneous graph neural architectures. Specifically, HGNAS first designs a new search space based on existing popular HGNNs. Then, HGNAS uses a policy network as the controller to sample and find the best neural architecture from the designed search space by maximizing the expected accuracy of the selected architectures on a given validation dataset. Moreover, we design a new method HGNAS++ to improve the efficiency of HGNAS by training the RNN controller within a generative adversarial learning framework. The basic idea of HGNAS++ is to embed a pairwise ranker into the reinforcement learning based architecture search algorithm. The pairwise ranker can be taken as a discriminator which selects more accurate architectures between pairs of candidate architectures. Then, the RNN controller can be updated more efficiently by only using a relatively small number of candidate architectures selected by the pairwise ranker. Experiments on real-world heterogeneous graph datasets show that HGNAS is capable of designing novel HGNNs that beat the best human-invented HGNNs. On the benchmark datasets, HGNAS++ improves HGNAS in terms of evaluation cost, with a reduction of 50% of the evaluated candidate architectures and a decrease of 24% in search time on average. As a byproduct, HGNAS++ can find sparse yet powerful neural architectures for HGNNs.

This paper concentrates on group anomalies in general large-scale networks. Existing algorithms on group anomalies mainly focus on homogeneous or bipartite networks, and thus are difficult to apply to heterogeneous networks directly. Moreover, these algorithms follow the non-overlapping hypothesis of groups implicitly, which is improper in many scenarios. For example, fraud users in Alibaba E-commerce platform may join more than one organization at the same time. In this paper, we introduce a novel algorithm calledAdaptive Label Propagation (ALP)to solve these problems. ALP is designed based on label propagation (LP) frameworks, for the reason that LP-based frameworks are simple in thought and easy to scale. ALP is able to find overlapping groups by label propagation with belonging coefficients, and can be applied to heterogeneous networks for its design of adaptive neighbor weighting. Assigning different weights to neighbors in label propagation is a challenging task. Inspired by the combinatorial multi-armed bandit mechanism, ALP views the neighbors of each node as arms to be selected, and iteratively updates their weights by evaluating their expected rewards in following iterations. Experiments are conducted on four real-world networks (including two bipartite ones and two heterogeneous ones). The results show that LP-based methods are effective for detecting group anomalies, and the comparison results with several state-of-the-art label propagation based community detection methods show the effectiveness of the proposed method.

Unsupervised graph embedding method generates node embeddings to preserve structural and content features in a graph without human labeling. However, most unsupervised graph representation learning methods suffer issues like poor scalability or limited utilization of content/structural relationships, especially on attributed graphs. In this paper, we propose SAGES, a graph sampling based autoencoder framework, which can alleviate these issues. Specifically, we propose a graph sampler considering both structural and content features, in which nodes with greater influence on each other have more chances to be sampled in the same subgraph. In addition, an unbiased Graph Autoencoder (GAE) with structure-level, content-level, and community-level reconstruction loss is built from the properly sampled subgraph each iteration. The time and space complexity analysis is carried out to show the scalability of SAGES. We conducted experiments on three medium-size attributed graphs and three large attributed graphs. Experimental results illustrate that SAGES achieves the competitive performance in unsupervised attributed graph learning on various downstream tasks including node classification, link prediction, and node clustering.

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.

Short video has witnessed rapid growth in the past few years in e-commerce platforms like Taobao. To ensure the freshness of the content, platforms need to release a large number of new videos every day, making conventional click-through rate (CTR) prediction methods suffer from the item cold-start problem. In this paper, we propose GIFT, an efficient Graph-guIded Feature Transfer system, to fully take advantages of the rich information of warmed-up videos to compensate for the cold-start ones. Specifically, we establish a heterogeneous graph that contains physical and semantic linkages to guide the feature transfer process from warmed-up video to cold-start videos.Specifically, we establish a heterogeneous graph that contains physical and semantic linkages to guide the feature transfer process. The physical linkages consist of the explicit relationships (e.g., produced by the same author, or showcasing the same product etc.), and the semantic linkages measure the proximity of multi-modal representations of two videos. We elaborately design the feature transfer function to make aware of different parts of transferred features (e.g., id representations and historical statistics) from different types of nodes and edges along the metapath on the graph. We conduct extensive experiments on a large real-world dataset, and the results show that our GIFT system outperforms SOTA methods significantly and brings a 6.82% lift on CTR in the homepage of Taobao App.

Graph Neural Networks (GNNs) have drawn attention due to their excellent performance in fraud detection tasks, which reveal fraudsters by aggregating the features of their neighbors. However, some fraudsters typically tend to alleviate their suspiciousness by connecting with many benign ones. Besides, label-imbalanced neighborhood also deteriorates fraud detection accuracy. Such behaviors violate the homophily assumption and worsen the performance of GNN-based fraud detectors. In this paper, we propose a Dual-Augment Graph Neural Network (DAGNN) for fraud detection tasks. In DAGNN, we design a two-pathway framework including disparity augment (DA) pathway and similarity augment (SA) pathway. Accordingly, we devise two novel information aggregation strategies. One is to augment the disparity between target node and its heterogenous neighbors in original topology. The other is to augment its similarity to homogenous neighbors in a relatively label-balanced neighborhood. The experimental results compared with the state-of-the-art models on two real-world datasets demonstrate the superiority of the proposed DAGNN.

Multi-task learning has been established as an important machine learning framework for leveraging shared knowledge among multiple different but related tasks, with the generalization performance of models enhanced. As a promising learning paradigm, multi-task learning has been widely adopted by various real-world applications, such as recommendation systems. Multi-gate Mixture-of-Experts (MMoE), a well-received multi-task learning method in industry, based on the classic and inspiring Mixture-of-Experts (MoE) structure, explicitly models task relationships and learns task-specific functionalities, generating significant improvements. However, in our applications, negative transfer, which confuses considerable existing multi-task learning methods, is still observed to happen to MMoE. In this paper, an in-depth empirical investigation into negative transfer is launched. And it reveals that, incompetent experts, which play fundamental roles under the learning framework of MoE, are the key technique bottleneck. To tackle this dilemma, we propose the Calibrated Mixture of Insightful Experts (CMoIE), with three novel modules (Conflict Resolution, Expert Communication, and Mixture Calibration), customed for multi-task learning. Hence a group of insightful experts are constructed with enhanced diversity, communication and specialization. To validate the proposed method CMoIE, experiments are conducted on three public datasets and one real-world click-through-rate prediction dataset we construct based on traffic logs collected from a large-scale online product recommendation system. Our approach yields best performance across all of these benchmarks, demonstrating the superiority of it.

Recommender Systems (RS), as an efficient tool to discover users' interested items from a very large corpus, has attracted more and more attention from academia and industry. As the initial stage of RS, large-scale matching is fundamental yet challenging. A typical recipe is to learn user and item representations with a two-tower architecture and then calculate the similarity score between both representation vectors, which however still struggles in how to properly deal with negative samples. In this paper, we find that the common practice that randomly sampling negative samples from the entire space and treating them equally is not an optimal choice, since the negative samples from different sub-spaces at different stages have different importance to a matching model. To address this issue, we propose a novel method named Unbiased Model-Agnostic Matching Approach (UMA2). It consists of two basic modules including 1) General Matching Model (GMM), which is model-agnostic and can be implemented as any embedding-based two-tower models; and 2) Negative Samples Debias Network (NSDN), which discriminates negative samples by borrowing the idea of Inverse Propensity Weighting (IPW) and re-weighs the loss in GMM. UMA$^2$ seamlessly integrates these two modules in an end-to-end multi-task learning framework. Extensive experiments on both real-world offline dataset and online A/B test demonstrate its superiority over state-of-the-art methods.

In online shopping, ever-changing fashion trends make merchants need to prepare more differentiated products to meet the diversified demands, and e-commerce platforms need to capture the market trend with a prophetic vision. For the trend prediction, the attribute tags, as the essential description of items, can genuinely reflect the decision basis of consumers. However, few existing works explore the attribute trend in the specific community for e-commerce. In this paper, we focus on the community trend prediction on the item attribute and propose a unified framework that combines the dynamic evolution of two graph patterns to predict the attribute trend in a specific community. Specifically, we first design a community-attribute bipartite graph at each time step to learn the collaboration of different communities. Next, we transform the bipartite graph into a hypergraph to exploit the associations of different attribute tags in one community. Lastly, we introduce a dynamic evolution component based on the recurrent neural networks to capture the fashion trend of attribute tags. Extensive experiments on three real-world datasets in a large e-commerce platform show the superiority of the proposed approach over several strong alternatives and demonstrate the ability to discover the community trend in advance.

In many classical e-commerce platforms, personalized recommendation has been proven to be of great business value, which can improve user satisfaction and increase the revenue of platforms. In this paper, we present a new recommendation problem, Trigger-Induced Recommendation (TIR), where users’ instant interest can be explicitly induced with a trigger item and follow-up related target items are recommended accordingly. TIR has become ubiquitous and popular in e-commerce platforms. In this paper, we figure out that although existing recommendation models are effective in traditional recommendation scenarios by mining users’ interests based on their massive historical behaviors, they are struggling in discovering users’ instant interests in the TIR scenario due to the discrepancy between these scenarios, resulting in inferior performance. To tackle the problem, we propose a novel recommendation method named Deep Interest Highlight Network (DIHN) for Click-Through Rate (CTR) prediction in TIR scenarios. It has three main components including 1) User Intent Network (UIN), which responds to generate a precise probability score to predict user’s intent on the trigger item; 2) Fusion Embedding Module (FEM), which adaptively fuses trigger item and target item embeddings based on the prediction from UIN; and (3) Hybrid Interest Extracting Module (HIEM), which can effectively highlight users’ instant interest from their behaviors based on the result of FEM. Extensive offline and online evaluations on a real-world e-commerce platform demonstrate the superiority of DIHN over state-of-the-art methods. Our code is available 1.

Outcome prediction aims to predict the future health condition of patients from Electronic Health Record (EHR) data. Because of the sequential characteristic of EHR data, recurrent neural network (RNN)-based outcome prediction methods have achieved state-of-the-art results. However, the major drawback of RNN-based outcome prediction methods is lack of interpretability, which would lead to trust issues. Aiming at this problem, this paper proposes interpretable outcome prediction model with hierarchical attention (IoHAN), an interpretable outcome prediction model by leveraging attention mechanism. The main novelty of IoHAN is that it can pinpoint the fine-grained influence on the final prediction result of each medical component by decomposing the attention weights hierarchically into hospital visits, medical variables, and interactions between medical variables. We evaluated IoHAN on MIMIC-III, a large real-world EHR data set. The experiment results demonstrate that IoHAN can achieve higher prediction accuracy than state-of-the-art outcome prediction models. In addition, the hierarchical decomposed attention weights can interpret the prediction results in a more natural and understandable way.

The stock market investors aim at maximizing their investment returns. Stock recommendation task is to recommend stocks with higher return ratios for the investors. Most stock prediction methods study the historical sequence patterns to predict stock trend or price in the near future. In fact, the future price of a stock is correlated not only with its historical price, but also with other stocks. In this article, we take into account the relationships between stocks (corporations) by stock relation graph. Furthermore, we propose a Time-aware Relational Attention Network (TRAN) for graph-based stock recommendation according to return ratio ranking. In TRAN, the time-aware relational attention mechanism is designed to capture time-varying correlation strengths between stocks by the interaction of historical sequences and stock description documents. With the dynamic strengths, the nodes of the stock relation graph aggregate the features of neighbor stock nodes by graph convolution operation. For a given group of stocks, the proposed TRAN model can output the ranking results of stocks according to their return ratios. The experimental results on several real-world datasets demonstrate the effectiveness of our TRAN for stock recommendation.

Impression regulation plays an important role in various online ranking systems, e.g. , e-commerce ranking systems always need to achieve local commercial demands on some pre-labeled target items like fresh item cultivation and fraudulent item counteracting while maximizing its global revenue. However, local impression regulation may cause “butterfly effects” on the global scale, e.g. , in e-commerce, the price preference fluctuation in initial conditions (overpriced or underpriced items) may create a significantly different outcome, thus affecting shopping experience and bringing economic losses to platforms. To prevent “butterfly effects”, some researchers define their regulation objectives with global constraints, by using contextual bandit at the page-level that requires all items on one page sharing the same regulation action, which fails to conduct impression regulation on individual items. To address this problem, in this article, we propose a personalized impression regulation method that can directly makes regulation decisions for each user-item pair. Specifically, we model the regulation problem as a C onstrained D ual-level B andit (CDB) problem, where the local regulation action and reward signals are at the item-level while the global effect constraint on the platform impression can be calculated at the page-level only. To handle the asynchronous signals, we first expand the page-level constraint to the item-level and then derive the policy updating as a second-order cone optimization problem. Our CDB approaches the optimal policy by iteratively solving the optimization problem. Experiments are performed on both offline and online datasets, and the results, theoretically and empirically, demonstrate CDB outperforms state-of-the-art algorithms.

Multi-label active learning (MAL) aims to learn an accurate multi-label classifier by selecting which examples (or example-label pairs) will be annotated and reducing query effort. MAL is a more complicated and expensive process than single-label active learning, due to one example can be associated with a set of non-exclusive labels and the annotator has to scrutinize the whole example and label space to provide correct annotations. Instead of scrutinizing the whole example for annotation, we may just examine some of its subexamples with respect to a label for annotation. In this way, we can not only save the annotation cost but also speedup the annotation process. Given this observation, we introduce CMAL, a two-stage Cost-effective MAL strategy (CMAL) by querying subexamples. CMAL first selects the most informative example-label pairs by leveraging uncertainty, label correlation and label space sparsity. Specifically, the uncertainty of a label to an example can be reduced if its correlated labels already annotated to the example, and its uncertainty can be reduced also if more examples annotated to this label. Next, CMAL greedily queries the most probable positive subexample-label pairs of the selected example-label pair. In addition, we propose rCMAL to account for the representative of examples to more reliably select example-label pairs in the first stage. Extensive experiments on multi-label datasets from diverse domains show that our proposed CMAL and rCMAL can better save the query cost than state-of-the-art MAL methods. The contribution of leveraging label correlation, label sparsity, and representative for saving cost is also confirmed.

With the development of e-commerce, fraud behaviors have been becoming one of the biggest threats to the e-commerce business. Fraud behaviors seriously damage the ranking system of e-commerce platforms and adversely influence the shopping experience of users. It is of great practical value to detect fraud behaviors on e-commerce platforms. However, the task is non-trivial, since the adversarial action taken by fraudsters. Existing fraud detection systems used in the e-commerce industry easily suffer from performance decay and can not adapt to the upgrade of fraud patterns, as they take already known fraud behaviors as supervision information to detect other suspicious behaviors. In this article, we propose a competitive graph neural networks (CGNN)-based fraud detection system (eFraudCom) to detect fraud behaviors at one of the largest e-commerce platforms, “Taobao” 1 . In the eFraudCom system, (1) the competitive graph neural networks (CGNN) as the core part of eFraudCom can classify behaviors of users directly by modeling the distributions of normal and fraud behaviors separately; (2) some normal behaviors will be utilized as weak supervision information to guide the CGNN to build the profile for normal behaviors that are more stable than fraud behaviors. The algorithm dependency on fraud behaviors will be eliminated, which enables eFraudCom to detect fraud behaviors in presence of the new fraud patterns; (3) the mutual information regularization term can maximize the separability between normal and fraud behaviors to further improve CGNN. eFraudCom is implemented into a prototype system and the performance of the system is evaluated by extensive experiments. The experiments on two Taobao and two public datasets demonstrate that the proposed deep framework CGNN is superior to other baselines in detecting fraud behaviors. A case study on Taobao datasets verifies that CGNN is still robust when the fraud patterns have been upgraded.

In this demo, we present an online multi-modal community query engine (MQE) on Alibaba's billion-scale heterogeneous network. MQE has two distinct features in comparison with existing community query engines. Firstly, MQE supports multimodal community search on heterogeneous graphs with keyword and image queries. Secondly, to facilitate community understanding in real business scenarios, MQE generates natural language descriptions for the retrieved community in combination with other useful demographic information. The distinct features of MQE benefit many downstream applications in Alibaba's e-commerce platform like recommendation. Our experiments confirm the effectiveness and efficiency of MQE on graphs with billions of edges.

On-demand food delivery (OFD) platforms have greatly impacted the food service industry, where OFD recommendation systems play a central role in enhancing user experience and raising revenues. OFD recommendation, compared with existing online e-commerce recommendation systems, needs to put more emphasis on fulfillment time related variables, because the order fulfillment cycle time (OFCT) which refers to the time elapsed between a user placing a food order and receiving the food significantly influences a user's choice from the recommended items. In this paper, we investigate the OFCT related information and propose a Fulfillment-Time-Aware Personalized Ranking (FTAPR) method for recommendation. FTAPR mainly consists of three components. First, Transformers are used to estimate OFCT based on a large amount of user order sequences. Then, the predicted OFCT and other OFCT related features are fused and encoded by a deep & cross network to learn fulfillment time related feature representation. At the last step, the time bias representation from the deep & cross network is integrated into the ranking system to deliver final search results. Extensive offline and online experiments on real-world datasets collected from one of China's largest OFD platforms Ele.me show the superiority of our model, e.g., an online A/B testing shows that FTAPR brings 1.3% and 2.5% gains in CTR and CVR compared with baselines.

Irregularly, asynchronously and sparsely sampled multivariate time series (IASS-MTS) are characterized by sparse non-uniform time intervals between successive observations and different sampling rates amongst series. Those properties pose substantial challenges to mainstream machine learning models for learning complicated relations within and across IASS-MTS. This is because that most of the models assume that the time series in question are even, complete (fixed-dimensional features) and synchronous. To address these challenges, we present a novel time-aware Dual-Attention and Memory-Augmented Network (DAMA-Net). The proposed model can leverage both time irregularity, multi-sampling rates and global temporal patterns information inherent in IASS-MTS so as to learn more effective representations for improving prediction performance. Comprehensive experiments on real datasets show that the DAMA-Net outperforms the state-of-the-art methods in multivariate time series classification task.

Online deposit-free leasing service has witnessed rapid growth in China and shows a promising market in the future. While eliminating the requirement of a deposit does attract more users to the service, it also lowers the cost for fraudsters. Since the emergence of this service is relatively new, there are few works in literature focusing on detecting fraud transactions in it. Existing efforts mainly fall into hard-coded solutions such as block-listing or scorecard methods, which can be impotent in the face of the diverse fraud tactics, e.g., identity theft, or even suffering concept drift problem as the tactics evolve. In this paper, we contribute Turbo, an efficient graph-based anti-fraud system, to fully exploit the abundant user behavior logs in a real-time manner. Turbo is able to additionally make use of the implicit user relationships beyond the user features in the logs. To capture the user relationships, we first propose a novel algorithm to construct a time-evolving user behavior network called BN. Empirical analysis demonstrates that fraudsters in BN exhibit unique temporal aggregation and homophilic patterns, which inspires us to develop a novel heterogeneous adaptive graph neural network algorithm called HAG. Specifically, in HAG two graph operators are presented to mitigate the over-smoothing problem and make better use of the heterogeneous behavior relations in BN. Extensive experiments on a real-world dataset show that our method outperforms state-of-the-art methods significantly and can give a response in seconds for each detection request.

In this paper, we study the problem of efficient motif-based graph partitioning (MGP). We observe that existing methods require to enumerate all motif instances to compute the exact edge weights for partitioning. However, the enumeration is prohibitively expensive against large graphs. We thus propose a sampling-based MGP (SMGP) framework that employs an unbiased sampling mechanism to efficiently estimate the edge weights while trying to preserve the partitioning quality. To further improve the effectiveness, we propose a novel adaptive sampling framework called SMGP+. SMGP+ iteratively partitions the input graph based on up-to-date estimated edge weights, and adaptively adjusts the sampling distribution so that edges that are more likely to affect the partitioning outcome will be prioritized for weight estimation. To our best knowledge, this is the first attempt to solve the MGP problem without employing exact edge weight computations, which gives hope for existing MGP methods to perform on complicated motifs in a scalable yet effective manner. Extensive experiments on seven real-world datasets have validated that our framework delivers competitive partitioning quality compared to existing workflows based on exact edge weights, while achieving orders of magnitude speedup.

With the development of e-commerce platforms, e-commerce recommendation systems are playing an increasingly important role for the purpose of product recommendation. As a new attack model against e-commerce recommendation systems, the "Ride Item's Coattails" attack creates fake click information to establish the deceptive correlation between popular products and low-quality products in order to mislead the recommendation system of e-commerce platform to boost the sales of low-quality products. This attack is characterized by high concealment and strong destructiveness, which can cause great damage to e-commerce recommendation systems, and adversely affect the usability of the e-commerce platform and users' shopping experience. It is therefore of great practical significance to study how to quickly and effectively identify the false click information and the corresponding "Ride Item's Coattails" attack to better safeguard e-commerce recommendation systems. At present, there is no previously reported relevant research work conducted specifically for addressing the detection of the "Ride Item's Coattails" attack. In this work, we carried out pioneering work in analyzing and summarizing the characteristics of the false click information produced by attackers on the target products in the "Ride Item's Coattails" attack and designed a set of attack detection techniques suitable for e-commerce recommendation systems. Experimental results on real e-commerce datasets show that our proposed techniques can quickly and effectively detect the large-scale fake click information as well as the associated "Ride Item's Coattails" attack in e-commerce recommendation systems.

The e-commerce era is witnessing rising new arrivals of items on e-commerce platforms every day. Identifying potential popular items accurately is of great importance in creating commercial value. Click-Through Rate (CTR) is a general indicator to evaluate item popularity. However, existing methods fail in new arrivals prediction because of sparse item features, missing item statistics and high time complexity of computing for all pairs of users and items. To tackle these challenges, we propose a novel Adversarial Two-tower Neural Network (ATNN) model for new arrivals CTR predictions by introducing an adversarial network to a two-tower network. We design a generator and a discriminator to better learn an item vector based on item profiles without item statistics. We also develop a strategy with an O(1) time complexity for a new item's popularity prediction by constructing a user group and utilizing its mean user vector in a time-efficient manner. We implement ATNN on a largescale real-world dataset from one of the world's largest ecommerce platforms, “Tmall.com”. Empirical results show that ATNN is strongly capable of learning item vectors from item profiles for e-commerce. Furthermore, by introducing multi-task learning technology, we extend ATNN to food delivery service. Experimental results on one popular food delivery platform, “Ele.me”, demonstrate that ATNN can recognize attractive and welcoming new restaurants that have higher Value per Page View (VpPV) and generate more Gross Merchandise Volume (GMV).

Heterogeneous Graph Neural networks (HGNNs) have been popularly used in processing complicated networks such as academic networks, social networks, and knowledge graphs. Despite their success, the design of the neural architectures of HGNNs still requires rich domain knowledge and heavy manual work. In this paper, we propose a Heterogeneous Graph Neural Architecture Search algorithm (HGNAS for short) which enables automatic design of the best neural architectures with minimal human effort. Specifically, HGNAS first defines a general HGNN framework on top of existing popular HGNNs. A search space of HGNAS is designed based on the general framework that includes multiple groups of message encoding and aggregation functions. Then, HGNAS uses a policy network as the controller to sample and find the best neural architecture from the designed search space by maximizing the expected accuracy of the selected architectures on a validation dataset. Moreover, we introduce effective methods to improve HGNAS from three aspects, i.e., the optimization of hyper-parameters, the improvement of search space, and the selection of message receptive fields. Experiments on public datasets show that HGNAS is capable of designing novel HGNNs that rival the best human-invented HGNNs. More interestingly, HGNAS finds some sparse yet powerful neural architectures for HGNNs on the benchmark datasets.

Multivariate Hawkes processes (MHPs) are classic methods to learn temporal patterns in event sequences of different entities. Traditional MHPs with explicit parametric intensity functions are friendly to model interpretability. However, recent Deep MHPs which employ various variants of recurrent neural networks are hardly to understand, albeit more expressive towards event sequences. The lack of model interpretability of Deep MHPs leads to a limited comprehension of complicated dynamics between events. To this end, we present a new Disentangled Deep Multivariate Hawkes Process $(\mathrm{D}^{2}$ MHP) to enhance model expressiveness and meanwhile maintain model interpretability. $\mathrm{D}^{2}$ MHP achieves state disentanglement by disentangling the latent representation of an event sequence into static and dynamic latent variables, and matches these latent variables to interpretable factors in the intensity function. Moreover, considering that an entity typically has multiple identities, $\mathrm{D}^{2}$ MHP further splits these latent variables into factorized representations, each of which is associated with a corresponding identity. Experiments on real-world datasets show that $\mathrm{D}^{2}$ MHP yields significant and consistent improvements over state-of-the-art baselines. We also demonstrate model interpretability via the detailed analysis.

Graph Neural Networks (GNNs), combining node features and structure information flexibly, have been widely studied and applied in many fields. The growth of graph size and rich features generates a considerable demand for achieving scalability while maintaining good classification performance in the research of GNNs. Graph partition technique, as used in a recent work ClusterGCN, which divides the graph into several sub-graphs, has become an important strategy to achieve the scalability, but the loss of information still affects the results. In this paper, AdClusterGCN is proposed to establish the interaction between graph partition and node classification, in which they can promote each other, and the effectiveness and efficiency of the model can be ensured at the same time. AdClusterGCN combines GNN models trained on a sequence of graph partitions to capture different features, where the current partition is affected using adjusted node/edge weights computed from the results of GNN models on previous partitions. The PageRank and resampling techniques are adopted to keep sufficient attention on important nodes in different models. We implement our method with TensorFlow and experimental studies show that AdClusterGCN achieves state-of-the-art performance on several public benchmarks.

Live-streaming platforms have recently gained significant popularity by attracting an increasing number of young users and have become a very promising form of online shopping. Similar to the traditional online shopping platforms such as Taobao, live-streaming platforms also suffer from online malicious fraudulent behaviors where many transactions are not genuine. The existing anti-fraud models proposed to recognize fraudulent transactions on traditional online shopping platforms are inapplicable on live-streaming platforms. This is mainly because live-streaming platforms are characterized by a unique type of heterogeneous live-streaming networks where multiple heterogeneous types of nodes such as users, live-streamers, and products are connected with multiple different types of edges associated with edge features. In this paper, we propose a new approach based on a heterogeneous graph neural network for LIve-streaming Fraud dEtection (called LIFE). LIFE designs an innovative heterogeneous graph learning model that fully utilizes various heterogeneous information of shopping transactions, users, streamers, and items from a given live-streaming platform. Moreover, a label propagation algorithm is employed within our LIFE framework to handle the limited number of labeled fraudulent transactions for model training. Extensive experimental results on a large-scale Taobao live-streaming platform demonstrate that the proposed method is superior to the baseline models in terms of fraud detection effectiveness on live-streaming platforms. Furthermore, we conduct a case study to show that the proposed method is able to effectively detect fraud communities for live-streaming e-commerce platforms.

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.

Graph neural architecture search has received a lot of attention as Graph Neural Networks (GNNs) has been successfully applied on the non-Euclidean data recently. However, exploring all possible GNNs architectures in the huge search space is too time-consuming or impossible for big graph data. In this paper, we propose a parallel graph architecture search (GraphPAS) framework for graph neural networks. In GraphPAS, we explore the search space in parallel by designing a sharing-based evolution learning, which can improve the search efficiency without losing the accuracy. Additionally, architecture information entropy is adopted dynamically for mutation selection probability, which can reduce space exploration. The experimental result shows that GraphPAS outperforms state-of-art models with efficiency and accuracy simultaneously.

Fraud detection is a pressing challenge for most financial and commercial platforms. In this paper, we study the processing pipeline of fraud detection in a large e-commerce platform of TaoBao. Graph label propagation (LP) is a core component in this pipeline to detect suspicious clusters from the user-interaction graph.Furthermore, the run-time of the LP component occupies 75% overhead of TaoBao's automated detection pipeline. To enable real-time fraud detection, we propose a GPU-based framework, called GLP, to support large-scale LP workloads in enterprises.We have identified two key challenges when integrating GPU acceleration into TaoBao's data processing pipeline: (1)programmability for evolving fraud detection logics; (2)demand for real-time performance. Motivated by these challenges, we offer a set of expressive APIs that data engineers can customize and deploy efficient LP algorithms on GPUs with ease. We propose novel GPU-centric optimizations by leveraging the community as well as power-law properties of large graphs. Extensive experiments have confirmed the effectiveness of our proposed optimizations. With a single GPU, GLP supports a real billion-scale graph workload from the fraud detection pipeline of TaoBao and achieves 8.2x speedup to the current in-house distributed solution running on high-end multicore machines.

A relational database, consisting of multiple tables, provides heterogeneous information across various entities, widely used in real-world services. This paper studies the supervised learning task on multiple tables, aiming to predict one label column with the help of multiple-tabular data. However, classical ML techniques mainly focus on single-tabular data. Multiple-tabular data refers to many-to-many mapping among joinable attributes and n-ary relations, which cannot be utilized directly by classical ML techniques. Besides, current graph techniques, like heterogeneous information network (HIN) and graph neural networks (GNN), are infeasible to be deployed directly and automatically in a multi-table environment, which limits the learning on databases.

GraphScope is a system and a set of language extensions that enable a new programming interface for large-scale distributed graph computing. It generalizes previous graph processing frameworks (e.g. , Pregel, GraphX) and distributed graph databases ( e.g ., Janus-Graph, Neptune) in two important ways: by exposing a unified programming interface to a wide variety of graph computations such as graph traversal, pattern matching, iterative algorithms and graph neural networks within a high-level programming language; and by supporting the seamless integration of a highly optimized graph engine in a general purpose data-parallel computing system. A GraphScope program is a sequential program composed of declarative data-parallel operators, and can be written using standard Python development tools. The system automatically handles the parallelization and distributed execution of programs on a cluster of machines. It outperforms current state-of-the-art systems by enabling a separate optimization (or family of optimizations) for each graph operation in one carefully designed coherent framework. We describe the design and implementation of GraphScope and evaluate system performance using several real-world applications.

Searching a community containing a given query vertex in an online social network enjoys wide applications like recommendation, team organization, etc. When applied to real-life networks, the existing approaches face two major limitations. First, they usually take two steps, i.e. , crawling a large part of the network first and then finding the community next, but the entire network is usually too big and most of the data are not interesting to end users. Second, the existing methods utilize hand-crafted rules to measure community membership, while it is very difficult to define effective rules as the communities are flexible for different query vertices. In this paper, we propose an Interactive Community Search method based on Graph Neural Network (shortened by ICS-GNN) to locate the target community over a subgraph collected on the fly from an online network. Specifically, we recast the community membership problem as a vertex classification problem using GNN, which captures similarities between the graph vertices and the query vertex by combining content and structural features seamlessly and flexibly under the guide of users' labeling. We then introduce a k -sized Maximum-GNN-scores (shortened by kMG ) community to describe the target community. We next discover the target community iteratively and interactively. In each iteration, we build a candidate subgraph using the crawled pages with the guide of the query vertex and labeled vertices, infer the vertex scores with a GNN model trained on the subgraph, and discover the kMG community which will be evaluated by end users to acquire more feedback. Besides, two optimization strategies are proposed to combine ranking loss into the GNN model and search more space in the target community location. We conduct the experiments in both offline and online real-life data sets, and demonstrate that ICS-GNN can produce effective communities with low overhead in communication, computation, and user labeling.

Due to diverse graph data and algorithms, programming and orchestration of complex computation pipelines have become the major challenges to making use of graph applications for Web-scale data analysis. GraphScope aims to provide a one-stop and efficient solution for a wide range of graph computations at scale. It extends previous systems by offering a unified and high-level programming interface and allowing the seamless integration of specialized graph engines in a general data-parallel computing environment. As we will show in this demo, GraphScope enables developers to write sequential graph programs in Python and provides automatic parallel execution on a cluster. This further allows GraphScope to seamlessly integrate with existing data processing systems in PyData ecosystem. To validate GraphScope's efficiency, we will compare a complex, multi-staged processing pipeline for a real-life fraud detection task with a manually assembled implementation comprising multiple systems. GraphScope achieves a 2.86× speedup on a trillion-scale graph in real production at Alibaba.

In the emerging business of food delivery, rider traffic accidents raise financial cost and social traffic burden. Although there has been much effort on traffic accident forecasting using temporal-spatial prediction models, none of the existing work studies the problem of detecting the takeaway rider accidents based on food delivery trajectory data. In this article, we aim to detect whether a takeaway rider meets an accident on a certain time period based on trajectories of food delivery and riders’ contextual information. The food delivery data has a heterogeneous information structure and carries contextual information such as weather and delivery history, and trajectory data are collected as a spatial-temporal sequence. In this article, we propose a TakeAway Rider Accident detection fusion network TARA-Net to jointly model these heterogeneous and spatial-temporal sequence data. We utilize the residual network to extract basic contextual information features and take advantage of a transformer encoder to capture trajectory features. These embedding features are concatenated into a pyramidal feed-forward neural network. We jointly train the above three components to combine the benefits of spatial-temporal trajectory data and sparse basic contextual data for early detecting traffic accidents. Furthermore, although traffic accidents rarely happen in food delivery, we propose a sampling mechanism to alleviate the imbalance of samples when training the model. We evaluate the model on a transportation mode classification dataset Geolife and a real-world Ele.me dataset with over 3 million riders. The experimental results show that the proposed model is superior to the state-of-the-art.

Helpful reviews in e-commerce sites can help customers acquire detailed information about a certain item, thus affecting customers' buying decisions. Predicting review helpfulness automatically in Taobao is an essential but challenging task for two reasons: (1) whether a review is helpful not only relies on its text, but also is related with the corresponding item and the user who posts the review, (2) the criteria of classifying review helpfulness under different items are not the same. To handle these two challenges, we propose CA-GNN (Category Aware Graph Neural Networks), which uses graph neural networks (GNNs) to identify helpful reviews in a multi-task manner --- we employ GNNs with one shared and many item-specific graph convolutions to learn the common features and each item's specific criterion for classifying reviews simultaneously. To reduce the number of parameters in CA-GNN and further boost its performance, we partition the items into several clusters according to their category information, such that items in one cluster share a common graph convolution.We conduct solid experiments on two public datasets and demonstrate that CA-GNN outperforms existing methods by up to 10.9% in AUC. We also deployed our system in Taobao with online A/B Test and verify that CA-GNN still outperforms the baseline system in most cases.

Large-scale online promotions, such as Double 11 and Black Friday, are of great value to e-commerce platforms nowadays. Traditional methods are not successful when we aim to maximize global Gross Merchandise Volume (GMV) in the promotion scenarios due to three limitations. The first is that the GMV of sellers varies significantly from daily scenarios to promotions. Second, these methods do not consider explosive demands in promotions, so that a consumer may fail to purchase some popular items due to sellers' limited capacities. Third, the traffic distribution over sellers presents divergence in different channels, thus rendering the performance of the traditional single-channel methods far from optimal in creating commercial values. To address these problems, we design a Multi-Channel Sellers Traffic Allocation (MCSTA) optimization model to obtain optimal page view (PV) distribution concerning global GMV. Then we propose a general constrained non-smooth convex optimization solution with a Multi-Objective Shortest Distance (MOSD) hyperparameter tuning method to solve MCSTA. This is the first work to systematically address this issue in the scenario of large-scale online promotions. The empirical results show that MCSTA achieves significant improvement of GMV by 1.1% based on A/B test during Alibaba's "Global Shopping Festival", one of the world's largest online sales events. Furthermore, we deploy MCSTA in other popular scenarios, including everyday promotion and video live stream service, to showcase that MCSTA can be widely applied in e-commerce and online entertainment services.

Recommending stock with the highest return ratio is always a challenging problem in the field of financial technology. In this paper, we propose a time-aware graph relational attention network (TRAN) for stock recommendation based on return ratio ranking. In TRAN, time-aware relational attention mechanism is the key unit to capture time-varying correlation strength between stocks by the interaction of historical sequences and stock description documents. With the dynamic strength, the nodes of the stock relation graph aggregate the features of neighbor stock nodes by graph convolution operation. For a given group of stocks, our model can output the ranking results of stocks according to their return ratios. The experimental results on several real-world datasets demonstrate the effectiveness of our TRAN for stock recommendation.

In this paper, we study the privacy of online health data. We present a novel online health data De-Anonymization (DA) framework, named De-Health. Leveraging two real world online health datasets WebMD and HealthBoards, we validate the DA efficacy of De-Health. We also present a linkage attack framework which can link online health/medical information to real world people. Through a proof-of-concept attack, we link 347 out of 2805 WebMD users to real world people, and find the full names, medical/health information, birthdates, phone numbers, and other sensitive information for most of the re-identified users. This clearly illustrates the fragility of the privacy of those who use online health forums.

The e-commerce appeals to a multitude of online shoppers by providing personalized experiences and becomes indispensable in our daily life. Accurately predicting user preference and making a recommendation of favorable items plays a crucial role in improving several key tasks such as Click Through Rate (CTR) and Conversion Rate (CVR) in order to increase commercial value. Some state-of-the-art collaborative filtering methods exploiting non-linear interactions on a user-item bipartite graph are able to learn better user and item representations with Graph Neural Networks (GNNs), which do not learn hierarchical representations of graphs because they are inherently flat. Hierarchical representation is reportedly favorable in making more personalized item recommendations in terms of behaviorally similar users in the same community and a context of topic-driven taxonomy. However, some advanced approaches, in this regard, are either only considering linear interactions, or adopting single-level community, or computationally expensive. To address these problems, we propose a novel method with Hierarchical bipartite Graph Neural Network (HiGNN) to handle large-scale e-commerce tasks. By stacking multiple GNN modules and using a deterministic clustering algorithm alternately, HiGNN is able to efficiently obtain hierarchical user and item embeddings simultaneously, and effectively predict user preferences on a larger scale. Extensive experiments on some real-world e-commerce datasets demonstrate that HiGNN achieves a significant improvement compared to several popular methods. Moreover, we deploy HiGNN in Taobao, one of the largest e-commerces with hundreds of million users and items, for a series of large-scale prediction tasks of item recommendations. The results also illustrate that HiGNN is arguably promising and scalable in real-world applications.

Data-driven recommender systems that can help to predict users' preferences are deployed in many real online service platforms. Several studies show that they are vulnerable to data poisoning attacks, and attackers have the ability to mislead the system to perform as their desires. Considering the realistic scenario, where the recommender system is usually a black-box for attackers and complex algorithms may be deployed in them, how to learn effective attack strategies on such recommender systems is still an under-explored problem. In this paper, we propose an adaptive data poisoning framework, PoisonRec, which can automatically learn effective attack strategies on various recommender systems with very limited knowledge. PoisonRec leverages the reinforcement learning architecture, in which an attack agent actively injects fake data (user behaviors) into the recommender system, and then can improve its attack strategies through reward signals that are available under the strict black-box setting. Specifically, we model the attack behavior trajectory as the Markov Decision Process (MDP) in reinforcement learning. We also design a Biased Complete Binary Tree (BCBT) to reformulate the action space for better attack performance. We adopt 8 widely-used representative recommendation algorithms as our testbeds, and make extensive experiments on 4 different real-world datasets. The results show that PoisonRec has the ability to achieve good attack performance on various recommender systems with limited knowledge.

Predicting the survival of cancer patients holds significant meaning for public health, and has attracted increasing attention in medical information communities. In this study, we propose a novel framework for cancer survival prediction named Multimodal Graph Neural Network (MGNN), which explores the features of real-world multimodual data such as gene expression, copy number alteration and clinical data in a unified framework. In order to explore the inherent relation, we first construct the bipartite graphs between patients and multimodal data. Subsequently, graph neural network is adopted to obtain the embedding of each patient on different bipartite graphs. Finally, a multimodal fusion neural layer is designed to fuse the features from different modal data. The output of our method is the classification of short term survival or long term survival for each patient. Experimental results on one breast cancer dataset demonstrate that MGNN outperforms all baselines. Furthermore, we test the trained model on lung cancer dataset, and the experimental results verify the strong robust by comparing with state-of-the-art methods.

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.

Recently, recurrent neural networks (RNNs) based methods have achieved profitable performance on mining temporal characteristics in user behavior. However, user preferences are changing over time and have not been fully exploited in e-commerce scenarios. To fill in the gap, we propose an approach, called quantum inspired preference interactive networks (QPIN), which leverages the mathematical formalism of quantum theory (QT) and the long short term memory (LSTM) network, to interactively learn user preferences. Specifically, the tensor product operation is used to model the interaction among a single user's own preferences, i.e. individual preferences. A quantum many-body wave function (QMWF) is employed to model interaction among all users' preferences, i.e. group preferences. Further, we bridge them by deriving a rigorous projection, and thus take the interplay between them into account. Experiments on an Amazon dataset as well as a real-world e-commerce dataset demonstrate the effectiveness of QPIN, which achieves superior performances compared with the state-of-the-art methods in terms of AUC and F1-score.

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.

Fraud sellers in e-commerce usually promote their products via fake transactions. Such behaviors damage the reputation of the e-commerce platform and jeopardize the business environment in the platform. The search engine of existing e-commerce platforms mainly focuses on generating transactions by matching users' queries and sellers' products. The most common method to defense fraud sellers is to set up a blacklist based on fraud detection and manual investigation, and then punish those sellers in the list, which is inefficient and can only cover a small fraction of potential fraud sellers. In this paper, we propose the first fraud aware impression regulation system (FAIR) which is data-driven and can work in large-scale e-commerce platforms. Its main function is to actively regulate the impressions received by all potential fraud sellers in a real-time fashion. It utilizes the reinforcement learning architecture to dynamically adjust the impression regulation strategy under different reward settings, which can not only promote the impression regulation effects, but also improve the revenue of the platform simultaneously. We deploy FAIR on the Taobao platform of Alibaba, one of the world's largest e-commerce search platform, and perform an A/B test for two weeks. The results show that FAIR can effectively reduce the fraud impressions and improve the overall platform revenue at the same time.

Nowadays, the popularity of e-commerce has brought huge economic benefits to factories, third-party merchants, and e-commerce service providers. Driven by such huge economic benefits, malicious merchants attempt to promote items through inserting fraudulent purchases, fake review scores, and/or feedback, into them. Mitigating this threat is challenging due to the difficulty of obtaining internal e-commerce data, the variance of e-commerce services used by malicious merchants, and the reluctance of service providers in cooperation. In this paper, we present an efficient, platform-independent, and robust e-commerce fraud detection system, CATS, to detect frauds for different large-scale e-commerce platforms. We implement the design of CATS into a prototype system and evaluate this prototype on the world's popular e-commerce platform Taobao. The evaluation result on Taobao shows that CATS can achieve a high accuracy of 91% in detecting frauds. Based on this success, we then apply CATS on another large-scale e-commerce platforms, and again CATS achieves an accuracy of 96%, suggesting that CATS is very effective on real e-commerce platforms. Based on the cross-platform evaluation results, we conduct a comprehensive analysis on the reported frauds and reveal several abnormal yet interesting behaviors of those reported frauds. Our study in this paper is expected to shed light on defending against frauds for various e-commerce platforms.

Hashing has been widely studied for big data retrieval due to its low storage cost and fast query speed. Zero-shot hashing (ZSH) aims to learn a hashing model that is trained using only samples from seen categories, but can generalize well to samples of unseen categories. ZSH generally uses category attributes to seek a semantic embedding space to transfer knowledge from seen categories to unseen ones. As a result, it may perform poorly when labeled data are insufficient. ZSH methods are mainly designed for single-modality data, which prevents their application to the widely spread multi-modal data. On the other hand, existing cross-modal hashing solutions assume that all the modalities share the same category labels, while in practice the labels of different data modalities may be different. To address these issues, we propose a general Cross-modal Zero-shot Hashing (CZHash) solution to effectively leverage unlabeled and labeled multi-modality data with different label spaces. CZHash first quantifies the composite similarity between instances using label and feature information. It then defines an objective function to achieve deep feature learning compatible with the composite similarity preserving, category attribute space learning, and hashing coding function learning. CZHash further introduces an alternative optimization procedure to jointly optimize these learning objectives. Experiments on benchmark multi-modal datasets show that CZHash significantly outperforms related representative hashing approaches both on effectiveness and adaptability.

Product bundling, offering a combination of items to customers, is one of the marketing strategies commonly used in online e-commerce and offline retailers. A high-quality bundle generalizes frequent items of interest, and diversity across bundles boosts the user-experience and eventually increases transaction volume. In this paper, we formalize the personalized bundle list recommendation as a structured prediction problem and propose a bundle generation network (BGN), which decomposes the problem into quality/diversity parts by the determinantal point processes (DPPs). BGN uses a typical encoder-decoder framework with a proposed feature-aware softmax to alleviate the inadequate representation of traditional softmax, and integrates the masked beam search and DPP selection to produce high-quality and diversified bundle list with an appropriate bundle size. We conduct extensive experiments on three public datasets and one industrial dataset, including two generated from co-purchase records and the other two extracted from real-world online bundle services. BGN significantly outperforms the state-of-the-art methods in terms of quality, diversity and response time over all datasets. In particular, BGN improves the precision of the best competitors by 16% on average while maintaining the highest diversity on four datasets, and yields a 3.85x improvement of response time over the best competitors in the bundle list recommendation problem.

Fraud transactions are one of the major threats faced by online e-commerce platforms. Recently, deep learning based classifiers have been deployed to detect fraud transactions. Inspired by findings on adversarial examples, this paper is the first to analyze the vulnerability of deep fraud detector to slight perturbations on input transactions, which is very challenging since the sparsity and discretization of transaction data result in a non-convex discrete optimization. Inspired by the iterative Fast Gradient Sign Method (FGSM) for the L8 attack, we first propose the Iterative Fast Coordinate Method (IFCM) for discrete L1 and L2 attacks which is efficient to generate large amounts of instances with satisfactory effectiveness. We then provide two novel attack algorithms to solve the discrete optimization. The first one is the Augmented Iterative Search (AIS) algorithm, which repeatedly searches for effective “simple” perturbation. The second one is called the Rounded Relaxation with Reparameterization (R3), which rounds the solution obtained by solving a relaxed and unconstrained optimization problem with reparameterization tricks. Finally, we conduct extensive experimental evaluation on the deployed fraud detector in TaoBao, one of the largest e-commerce platforms in the world, with millions of real-world transactions. Results show that (i) The deployed detector is highly vulnerable to attacks as the average precision is decreased from nearly 90% to as low as 20% with little perturbations; (ii) Our proposed attacks significantly outperform the adaptions of the state-of-the-art attacks. (iii) The model trained with an adversarial training process is significantly robust against attacks and performs well on the unperturbed data.

Modeling user behaviors as sequences provides critical advantages in predicting future user actions, such as predicting the next product to purchase or the next song to listen to, for personalized search and recommendation. Recently, recurrent neural networks (RNNs) have been adopted to leverage their power in modeling sequences. However, most of the previous RNN-based work suffers from the complex dependency problem, which may lose the integrity of highly correlated behaviors and may introduce noises derived from unrelated behaviors. In this paper, we propose to integrate a novel Time Slice Self-Attention (TiSSA) mechanism into RNNs for better modeling sequential user behaviors, which utilizes the time-interval-based gated recurrent units to exploit the temporal dimension when encoding user actions, and has a specially designed time slice hierarchical self-attention function to characterize both local and global dependency of user actions, while the final context-aware user representations can be used for downstream applications. We have performed experiments on a huge dataset collected from one of the largest e-commerce platforms in the world. Experimental results show that the proposed TiSSA achieves significant improvement over the state-of-the-art. TiSSA is also adopted in this large e-commerce platform, and the results of online A/B test further indicate its practical value.

E-commerce taxonomy plays an essential role in online retail business. Existing taxonomy of e-commerce platforms organizes items into an ontology structure. However, the ontology-driven approach is subject to costly manual maintenance and often does not capture user's search intention, particularly when user searches by her personalized needs rather than a universal definition of the items. Observing that search queries can effectively express user's intention, we present a novel large-Scale Hierarchical taxOnomy via grAph based query coaLition ( SHOAL ) to bridge the gap between item taxonomy and user search intention. SHOAL organizes hundreds of millions of items into a hierarchical topic structure . Each topic that consists of a cluster of items denotes a conceptual shopping scenario, and is tagged with easy-to-interpret descriptions extracted from search queries. Furthermore, SHOAL establishes correlation between categories of ontology-driven taxonomy, and offers opportunities for explainable recommendation. The feedback from domain experts shows that SHOAL achieves a precision of 98% in terms of placing items into the right topics, and the result of an online A/B test demonstrates that SHOAL boosts the Click Through Rate (CTR) by 5%. SHOAL has been deployed in Alibaba and supports millions of searches for online shopping per day.

Matching buyers with most suitable sellers providing relevant items (e.g., products) is essential for e-commerce platforms to guarantee customer experience. This matching process is usually achieved through modeling inter-group (buyer-seller) proximity by e-commerce ranking systems. However, current ranking systems often match buyers with sellers of various qualities, and the mismatch is detrimental to not only buyers' level of satisfaction but also the platforms' return on investment (ROI). In this paper, we address this problem by incorporating intra-group structural information (e.g., buyer-buyer proximity implied by buyer attributes) into the ranking systems. Specifically, we propose De ep Gr aph E mbe dding (DEGREE), a deep learning based method, to exploit both inter-group and intra-group proximities jointly for structural learning. With a sparse filtering technique, DEGREE can significantly improve the matching performance with computation resources less than that of alternative deep learning based methods. Experimental results demonstrate that DEGREE outperforms state-of-the-art graph embedding methods on real-world e-commence datasets. In particular, our solution boosts the average unit price in purchases during an online A/B test by up to 11.93%, leading to better operational efficiency and shopping experience.

Nowadays, e-commerce has become prevalent world-wide. With the big success of e-commerce, many malicious promotion services also rise: with the goal of increasing sales, malicious merchants attempt to promote their target items by illegally optimizing the search results using fake visits, purchases, etc. In this paper, we study the fraud detection problem on large-scale e-commerce platforms. First, we develop an efficient and scalable AnTi-Fraud system (ATF) to detect e-commerce frauds for large-scale e-commerce platforms, and implement it in parallel on a large-scale computing platform, called Open Data Processing Service (ODPS). Then, we evaluate ATF using two real large-scale e-commerce datasets (with tens of millions users and items). The results demonstrate that both the precision and the recall of ATF can achieve 0.97+, which suggests that ATF is very effective. More importantly, we deploy ATF on the Taobao platform of Alibaba, which is one of the world's largest e-commerce platforms. The evaluation results show that ATF can also achieve an accuracy of 98.16% on Taobao, which again suggests that ATF is very effective and deployable in practice. Our study in this paper is expected to shed light on defending against online frauds for practical e-commerce platforms.

Multi-label active learning addresses the scarce labeled example problem by querying the most valuable unlabeled examples, or example-label pairs, to achieve a better performance with limited query cost. Current multi-label active learning methods require the scrutiny of the whole example in order to obtain its annotation. In contrast, one can find positive evidence with respect to a label by examining specific patterns (i.e., subexample), rather than the whole example, thus making the annotation process more efficient. Based on this observation, we propose a novel two-stage cost effective multi-label active learning framework, called CMAL. In the first stage, a novel example-label pair selection strategy is introduced. Our strategy leverages label correlation and label space sparsity of multi-label examples to select the most uncertain example-label pairs. Specifically, the unknown relevant label of an example can be inferred from the correlated labels that are already assigned to the example, thus reducing the uncertainty of the unknown label. In addition, the larger the number of relevant examples of a particular label, the smaller the uncertainty of the label is. In the second stage, CMAL queries the most plausible positive subexample-label pairs of the selected example-label pairs. Comprehensive experiments on multi-label datasets collected from different domains demonstrate the effectiveness of our proposed approach on cost effective queries. We also show that leveraging label correlation and label sparsity contribute to saving costs.

Current efforts on multi-label learning generally assume that the given labels of training instances are noise-free. However, obtaining noise-free labels is quite difficult and often impractical, and the presence of noisy labels may compromise the performance of multi-label learning. Partial multi-label learning (PML) addresses the scenario in which each instance is annotated with a set of candidate labels, of which only a subset corresponds to the ground-truth. The PML problem is more challenging than partial-label learning, since the latter assumes that only one label is valid and may ignore the correlation among candidate labels. To tackle the PML challenge, we introduce a feature induced PML approach called fPML, which simultaneously estimates noisy labels and trains multi-label classifiers. In particular, fPML simultaneously factorizes the observed instance-label association matrix and the instance-feature matrix into low-rank matrices to achieve coherent low-rank matrices from the label and the feature spaces, and a low-rank label correlation matrix as well. The low-rank approximation of the instance-label association matrix is leveraged to estimate the association confidence. To predict the labels of unlabeled instances, fPML learns a matrix that maps the instances to labels based on the estimated association confidence. An empirical study on public multi-label datasets with injected noisy labels, and on archived proteomic datasets, shows that fPML can more accurately identify noisy labels than related solutions, and consequently can achieve better performance on predicting labels of instances than competitive methods.

With the development of cloud computing and big data, the reliability of data storage systems becomes increasingly important. Previous researchers have shown that machine learning algorithms based on SMART attributes are effective methods to predict hard drive failures. In this paper, we use SMART attributes to predict hard drive health degrees which are helpful for taking different fault tolerant actions in advance. Given the highly imbalanced SMART datasets, it is a nontrivial work to predict the health degree precisely. The proposed model would encounter overfitting and biased fitting problems if it is trained by the traditional methods. In order to resolve this problem, we propose two strategies to better utilize imbalanced data and improve performance. Firstly, we design a layerwise perturbation-based adversarial training method which can add perturbations to any layers of a neural network to improve the generalization of the network. Secondly, we extend the training method to the semi-supervised settings. Then, it is possible to utilize unlabeled data that have a potential of failure to further improve the performance of the model. Our extensive experiments on two real-world hard drive datasets demonstrate the superiority of the proposed schemes for both supervised and semi-supervised classification. The model trained by the proposed method can correctly predict the hard drive health status 5 and 15 days in advance.

Semantic proximity search on heterogeneous graph is an important task, and is useful for many applications. It aims to measure the proximity between two nodes on a heterogeneous graph w.r.t. some given semantic relation. Prior work often tries to measure the semantic proximity by paths connecting a query object and a target object. Despite the success of such path-based approaches, they often modeled the paths in a weakly coupled manner, which overlooked the rich interactions among paths. In this paper, we introduce a novel concept of interactive paths to model the inter-dependency among multiple paths between a query object and a target object. We then propose an Interactive Paths Embedding (IPE) model, which learns low-dimensional representations for the resulting interactive-paths structures for proximity estimation. We conduct experiments on seven relations with four different types of heterogeneous graphs, and show that our model outperforms the state-of-the-art baselines.

"Add to Favorites" is a popular function in online shopping sites which helps users to make a record of potentially interesting items for future purchases. It is usually regarded as a type of explicit feedback signal for item popularity and therefore also adopted as a ranking signal by many shopping search engines. With the increasing usage of crowdsourcing platforms, some malicious online sellers also organize crowdturfing activities to increase the numbers of "Add to Favorites" for their items. By this means, they expect the items to gain higher positions in search ranking lists and therefore boost sales. This kind of newly-appeared malicious activity proposes challenges to traditional search spam detection efforts because it involves the participation of many crowd workers who are normal online shopping users in most of the times, and these activities are composed of a series of behaviors including search, browse, click and add to favorites. To shed light on this research question, we are among the first to investigate this particular spamming activity by looking into both the task organization information in crowdsourcing platforms and the user behavior information from online shopping sites. With a comprehensive analysis of some ground truth spamming activities from the perspective of behavior, user and item, we propose a factor graph based model to identify this kind of spamming activity. Experimental results based on data collected in practical shopping search environment show that our model helps detect malicious "Add to Favorites" activities effectively.

In this paper, we provide some insights into analysis of fraud transaction recognition on Alipay's Money Flow Network. We first show that the Money Flow Network follows a power-law distribution on daily, monthly or yearly basis, based on which we propose a new approach of fraud transaction recognition on the Money Flow Network from two perspectives. First, the Collapse Network is identified by the discovery that fraud transaction requires a huge amount of active controlled 'zombie' accounts, which are always intentionally manipulated by fraudulent online sellers, and the collapse of the Money Flow Network emerges due to their economic inactivity; Second, we define the Activation Forest that leads to the recognition of the controlled 'zombies' even no sooner than they enter into Alipay's ecosphere. These two networks are fully explored from the perspective of detecting 'zombies', and several key features have been adopted into anti-fraud recognition. Experimental results show that our strategy is capable of effectively identifying fraud transactions on the Money Flow Network with the accuracy as high as 99.88%.

The problem of modeling topics on user behavior data in social networks has been widely studied in social marketing and social emotion analysis, where latent topic models are commonly used as the solutions. The user behavior data are highly related in time and space, which demands new latent topic models that consider both temporal and spatial distances. However, existing topic models either fail to model these two factors simultaneously, or cannot handle the high order dependence among user behaviors. In this paper we present a new nonparametric Bayesian model Time and Space Dependent Chinese Restaurant Processes (TSD-CRP for short). TSD-CRP can auto-select the number of topics and model high-order temporal and spatial dependence behind user behavior data. Empirical results on real-world data sets demonstrate the effectiveness of the proposed method.

Group feature selection makes use of structural information among features to discover a meaningful subset of features. Existing group feature selection algorithms only deal with pre-given candidate feature sets and they are incapable of handling streaming features. On the other hand, feature selection algorithms targeted for streaming features can only perform at the individual feature level without considering intrinsic group structures of the features. In this paper, we perform group feature selection with streaming features. We propose to perform feature selection at the group and individual feature levels simultaneously in a manner of a feature stream rather than a pre-given candidate feature set. In our approach, the group structures are fully utilized to reduce the cost of evaluating streaming features. We have extensively evaluated the proposed method. Experimental results have demonstrated that our proposed algorithms statistically outperform state-of-the-art methods of feature selection in terms of classification accuracy.