Live streaming has become a cornerstone of today's internet, enabling massive real-time social interactions. However, it faces severe risks arising from sparse, coordinated malicious behaviors among multiple participants, which are often concealed within normal activities and challenging to detect timely and accurately. In this work, we provide a pioneering study on risk assessment in live streaming rooms, characterized by weak supervision where only room-level labels are available. We formulate the task as a Multiple Instance Learning (MIL) problem, treating each room as a bag and defining structured user–timeslot capsules as instances. These capsules represent subsequences of user actions within specific time windows, encapsulating localized behavioral patterns.

Delayed feedback poses a core challenge for online CVR prediction, forcing a trade-off between label accuracy and data freshness. Existing methods address this through delay modeling or sample reweighting, yet neglect how post-click behaviors evolve over the observation period. To overcome this limitation, we formalize this evolution as feedback trajectory and propose TRACE. Instead of forcing hard labels on unrevealed samples, our method evaluates how well the accumulated feedback status aligns with conversion versus non-conversion, dynamically refining posteriors without waiting for final outcomes. To counteract early-stage trajectory sparsity, we further design a reliability-gated retrospective completer that leverages full-lifecycle data to provide adaptive posterior guidance for unrevealed samples. Extensive experiments validate TRACE's superiority over state-of-the-art baselines and confirm the retrospective completion module as a model-agnostic enhancer for existing systems. Our code is available at https://github.com/LunaZhangxy/TRACE.

Text-attributed graphs (TAGs) enhance graph learning by integrating rich textual semantics and topological context for each node. While boosting expressiveness, they also expose new vulnerabilities in graph learning through text-based adversarial surfaces. Recent advances leverage diverse backbones, such as graph neural networks (GNNs) and pre-trained language models (PLMs), to capture both structural and textual information in TAGs. This diversity raises a key question: How can we design universal adversarial attacks that generalize across architectures to assess the security of TAG models? The challenge arises from the stark contrast in how different backbones—GNNs and PLMs—perceive and encode graph patterns, coupled with the fact that many PLMs are only accessible via APIs, limiting attacks to black-box settings. To address this, we propose BadGraph, a novel attack framework that deeply elicits large language models' (LLMs) understanding of general graph knowledge to jointly perturb both node topology and textual semantics. Specifically, we design a target influencer retrieval module that leverages graph priors to construct cross-modally aligned attack shortcuts, thereby enabling efficient LLM-based perturbation reasoning. Experiments show that BadGraph achieves universal and effective attacks across GNN- and LLM-based reasoners, with up to a 76.3% performance drop, while theoretical and empirical analyses confirm its stealthy yet interpretable nature.

A critical limitation of conventional sequential recommendation models (SRMs) is their reliance on observed user-item interaction sequences within a closed-world setting, which hinders their ability to generalize to unseen or infrequent items. Recently, Large Language Models (LLMs) have shown remarkable promise in recommendation systems due to their vast world knowledge and advanced reasoning capabilities. Current research has predominantly explored two approaches: using LLMs to directly generate recommendations and distilling knowledge from LLMs to enhance conventional SRMs. However, these approaches face two major challenges: (1) high inference costs, as they require LLM responses during inference, either for generating predictions or as supplementary input; (2) inadequate distillation of the reasoning process, as existing methods focus mainly on improving embeddings or aligning outputs, without fully integrating LLMs' inherent reasoning capabilities. To address these issues, we propose LCKD-SR, anLLM-drivenCascadedKnowledgeDistillation framework forSequentialRecommendation. In this framework, an LLM, a Teacher SRM, and a Student SRM form a hierarchical distillation structure, enabling an LLM-free inference by using only the Student model. Beyond traditional embedding and ranking distillation, our framework abstracts the LLM's sequential reasoning abilities by identifying key interactions that subsequently guide the Teacher's attention using learnable markers. The Student model, which mirrors the architecture of the Teacher, achieves seamless knowledge alignment from the Teacher across all three aspects. Extensive experiments demonstrate the effectiveness and efficiency of the proposed LCKD-SR, showcasing its scalability to perform multi-level knowledge transfer while enabling LLM-independent inference, thereby overcoming the inference cost and reasoning limitations of existing methods.

Accurate personalized headline generation hinges on precisely capturing user interests from historical behaviors. However, existing methods neglect personalized-irrelevant click noise in entire historical clickstreams, which may lead to hallucinated headlines that deviate from genuine user preferences. In this paper, we reveal the detrimental impact of click noise on personalized generation quality through rigorous analysis in both user and news dimensions. Based on these insights, we propose a novel Personalized Headline Generation framework via Denoising Fake Interests from Implicit Feedback (PHG-DIF). PHG-DIF first employs dual-stage filtering to effectively remove clickstream noise, identified by short dwell times and abnormal click bursts, and then leverages multi-level temporal fusion to dynamically model users' evolving and multi-faceted interests for precise profiling. Moreover, we release DT-PENS, a new benchmark dataset comprising the click behavior of 1,000 carefully curated users and nearly 10,000 annotated personalized headlines with historical dwell time annotations. Extensive experiments demonstrate that PHG-DIF substantially mitigates the adverse effects of click noise and significantly improves headline quality, achieving state-of-the-art (SOTA) results on DT-PENS. Our framework implementation and dataset are available at https://github.com/liukejin-up/PHG-DIF.

Graph Neural Networks (GNNs) are vulnerable to perturbations in both edges and attributes by fraudsters attempting to evade detection. A low-cost and effective perturbation strategy involves establishing connections with benign users and providing as little information as possible, leading to a graph with noisy structure and absent attributes. We formulate a novel problem as learning in Graphs with Noisy structures and Absent node attributes (LGNA), for which no existing methods are specifically designed. To mitigate this gap, we propose a reliable graph learning framework called RENA, which implements a “Dilution of Unreliable Information” approach for the LGNA task. The core principle of RENA is to utilize more reliable information to decrease the proportion of unreliable information, thus diluting its impact. Specifically, only the observed node attributes and unconnected node pairs are considered reliable, while imputed attributes and connected node pairs are deemed unreliable. We first randomly sample a large number of unconnected node pairs and fewer connected pairs to create different structural views to supervise structure learning and dilute the impact of noisy edges. Next, we apply a graph autoencoder framework, assigning higher weights to the observed attributes and lower weights to the imputed attributes during the reconstruction process, thereby diluting the impact of imputation noise. Experiments show that our method outperforms state-of-the-art baselines on LGNA scenarios and conventional incomplete graph learning tasks. Code is available at https://github.com/lxx01110/RENA.

Typical financial risk management involves distinct phases for pre-service risk assessment and in-service default detection, often modeled separately. This paper proposes a novel framework, Multi-Granularity Knowledge Distillation (abbreviated as MGKD), aimed at improving pre-service risk prediction through the integration of in-service user behavior data. MGKD follows the idea of knowledge distillation, where the teacher model, trained on historical in-service data, guides the student model, which is trained on pre-service data. By using soft labels derived from in-service data, the teacher model helps the student model improve its risk prediction prior to service activation. Meanwhile, a multi-granularity distillation strategy is introduced, including coarse-grained, fine-grained, and self-distillation, to align the representations and predictions of the teacher and student models. This approach not only reinforces the representation of default cases but also enables the transfer of key behavioral patterns associated with defaulters from the teacher to the student model, thereby improving the overall performance of pre-service risk assessment. Moreover, we adopt a re-weighting strategy to mitigate the model's bias towards the minority class. Experimental results on large-scale real-world datasets from Tencent Mobile Payment demonstrate the effectiveness of our proposed approach in both offline and online scenarios.

In this paper, we reveal that existing Differentially Private Graph Neural Networks (DP-GNNs) are not effective against Graph Reconstruction Attack (GRA). We further attribute the ineffectiveness of existing DP-GNNs against GRA to their unstructured perturbation mechanism, which only induces unidirectional shift in the embedding similarity distribution. Specifically, this perturbation mechanism tends to decrease the embedding similarity of all node pairs without significantly disrupting the relative ranking, thus allowing GRA to still reconstruct the original graph structure by leveraging the relative ranking of similarities. To address this, we propose a novel Differentially Private Graph Neural Network based on Structured Perturbation (GRASP). Specifically, we observe that independent noise tends to decrease the embedding similarity, while identical noise tends to increase it. By integrating these two types of noise using a Bernoulli technique, we introduce a simple yet effective structured perturbation mechanism, which promotes bidirectional shift in the embedding similarity distribution, thereby effectively disrupting the relative ranking and defending against GRA. Extensive experiments on eight benchmark datasets demonstrate that GRASP effectively defends against GRA. Furthermore, GRASP achieves a superior privacy-utility trade-off compared to existing graph structure protection methods. The implementation of GRASP is available at https://github.com/ZhiyuZone/GRASP/.

Recent prevailing works on graph machine learning typically follow a similar methodology that involves designing advanced variants of graph neural networks (GNNs) to maintain the superior performance of GNNs on different graphs. In this paper, we aim to streamline the GNN design process and leverage the advantages of Large Language Models (LLMs) to improve the performance of GNNs on downstream tasks. We formulate a new paradigm, coined "LLMs-as-Consultants", which integrates LLMs with GNNs in an interactive manner. A framework named LOGIN (LLM cOnsulted GNN traINing) is instantiated, empowering the interactive utilization of LLMs within the GNN training process. First, we attentively craft concise prompts for spotted nodes, carrying comprehensive semantic and topological information, and serving as input to LLMs. Second, we refine GNNs by devising a complementary coping mechanism that utilizes the responses from LLMs, depending on their correctness. We empirically evaluate the effectiveness of Lalebox1 [0.8]O Galebox1 [0.8]IN on node classification tasks across both homophilic and heterophilic graphs. The results illustrate that even basic GNN architectures, when employed within the proposed LLMs-as-Consultants paradigm, can achieve comparable performance to advanced GNNs with intricate designs. Our code is available at https://github.com/QiaoYRan/LOGIN.

Graph Neural Networks (GNNs) are vulnerable to backdoor attacks, where adversaries implant malicious triggers to manipulate model predictions. Existing graph backdoor attacks are susceptible to defense mechanisms or robust classifiers because they rely on subgraph injection or structural perturbations, e.g., creating additional edges to attach backdoor triggers to the original graph. To enhance the stealthiness of graph backdoors, we propose SPEAR, a novel structure-preserving graph backdoor attack that avoids modifying the graph's topology. SPEAR operates within a limited attack budget by selectively perturbing node attributes while ensuring the triggers exert significant influence through a global importance-driven feature selection strategy. Additionally, a neighborhood-aware trigger generator is employed to underpin a high attack success rate by utilizing semantic information from the neighborhood. SPEAR amplifies effectiveness and stealthiness by combining subtle yet impactful attribute manipulation with a refined trigger generation mechanism. Extensive experiments demonstrate that SPEAR achieves state-of-the-art effectiveness in bypassing defenses on real-world datasets, establishing it as a potent and stealthy backdoor attack for graph-based tasks. Code is available at https://github.com/yhDing/SPEAR.

Online inclusive financial services encounter significant financial risks due to their expansive user base and low default costs. By real-world practice, we reveal that utilizing longer-term user payment behaviors can enhance models' ability to forecast financial risks. However, learning long behavior sequences is non-trivial for deep sequential models. Additionally, the diverse fields of payment behaviors carry rich information, requiring thorough exploitation. These factors collectively complicate the task of long-term user behavior modeling. To tackle these challenges, we propose a Long-term Payment Behavior Sequence Folding method, referred to as LBSF. In LBSF, payment behavior sequences are folded based on merchants, using the merchant field as an intrinsic grouping criterion, which enables informative parallelism without reliance on external knowledge. Meanwhile, we maximize the utility of payment details through a multi-field behavior encoding mechanism. Subsequently, behavior aggregation at the merchant level followed by relational learning across merchants facilitates comprehensive user financial representation. We evaluate LBSF on the financial risk assessment task using a large-scale real-world dataset. The results demonstrate that folding long behavior sequences based on internal behavioral cues effectively models long-term patterns and changes, thereby generating more accurate user financial profiles for practical applications.

Graph Neural Networks (GNNs) conventionally operate under the assumption that node attributes are entirely observable. Their performance notably deteriorates when confronted with incomplete graphs due to the inherent message-passing mechanisms. Current solutions either employ classic imputation techniques or adapt GNNs to tolerate missed attributes. However, their ability to generalize is impeded especially when dealing with high rates of missing attributes. To address this, we harness the representations of the essential views on graphs, attributes and structures, into a common shared latent space, ensuring robust tolerance even at high missing rates. Our proposed neural model, named ASD-VAE, parameterizes such space via a coupled-and-decoupled learning procedure, reminiscent of brain cognitive processes and multimodal fusion. Initially, ASD-VAE separately encodes attributes and structures, generating representations for each view. A shared latent space is then learned by maximizing the likelihood of the joint distribution of different view representations through coupling. Then, the shared latent space is decoupled into separate views, and the reconstruction loss of each view is calculated. Finally, the missing values of attributes are imputed from this learned latent space. In this way, the model offers enhanced resilience against skewed and biased distributions typified by missing information and subsequently brings benefits to downstream graph machine-learning tasks. Extensive experiments conducted on four typical real-world incomplete graph datasets demonstrate the superior performance of ASD-VAE against the state-of-the-art

In this article, we study the problem of personalized news headline generation, which aims to produce not only concise and fact-consistent titles for news articles but also decorate these titles as personalized irresistible reading invitations by incorporating readers’ preferences. We propose an approach named PNG ( P ersonalized N ews headline G enerator) by utilizing distant supervision in readers’ past click behaviors to resolve. First, user preference representations are learned through a knowledge-aware user encoder that comprehensively captures the genuine, sequential, and flash interests of users reflected in their historical clicked news. Then, a user-perturbed pointer-generator network is devised to accomplish the headline generation in which the learned user representations implicitly affect the word prediction. The proposed model is optimized by reinforcement learning solvers where indicators on factual, personalized, and linguistic aspects of the generated headline are regarded as rewards. Extensive experiments are conducted on the real-world dataset PENS, 1 which is a large-scale benchmark collected from Microsoft News. Both the quantitative and qualitative results validate the effectiveness of our approach.

Obtaining accurately calibrated conversion rate predictions is essential for the bidding and ranking process in online advertising systems. Nevertheless, the inherent latency between clicks and conversions leads to delayed feedback, which may introduce bias into the prediction models. Compared to indefinitely long conversion delays, post-click user behaviors manifest within a relatively brief time and have been empirically validated to exert a favorable influence on the precision of conversion rate estimates. In light of this, we propose a novel approach that leverages post-click user behaviors to calibrate conversion rate predictions. Specifically, we treat user behaviors as predictable targets to improve accuracy and enhance timeliness. An adaptive loss function based on task uncertainty is employed for multi-task learning. To further reduce calibration error, we integrate the modified prediction model with a parameterized scaling technique. Experiments conducted on two real-world datasets demonstrate that our proposed method outperforms existing models in providing more calibrated predictions.

With the ever-increasing urbanization process, modeling people’s spatiotemporal activities from their online traces has become a crucial task. State-of-the-art methods for this task rely on cross-modal embedding, which maps items from different modalities (e.g., location, time, text) into the same latent space. Despite their inspiring results, existing cross-modal embedding methods merely capture co-occurrences between items without modeling their high-order interactions. In this paper, we first construct the user interaction graph and the activity graph from raw data records and propose a hierarchical cross-modal embedding method that takes the high-order relationships into consideration. We introduce both inter-record and intra-record meta-graph structures, which enable learning distributed representations that preserve high-order proximities across graphs from different layers. Our empirical experiments on three real-world datasets demonstrate that our method not only outperforms state-of-the-art methods for spatiotemporal activity prediction but also captures cross-modal proximity at a finer granularity.

Personalized news headline generation, aiming at generating user-specific headlines based on readers’ preferences, burgeons a recent flourishing research direction. Existing studies generally inject a user interest embedding into an encoder-decoder headline generator to make the output personalized, while the factual consistency of headlines is inadequate to be verified. In this paper, we propose a framework Fact-Preserved Personalized News Headline Generation (short for FPG), to prompt a tradeoff between personalization and consistency. In FPG, the similarity between the candidate news to be exposed and the historical clicked news is used to give different levels of attention to key facts in the candidate news, and the similarity scores help to learn a fact-aware global user embedding. Besides, an additional training procedure based on contrastive learning is devised to further enhance the factual consistency of generated headlines. Extensive experiments conducted on a real-world benchmark PENS1validate the superiority of FPG, especially on the tradeoff between personalization and factual consistency.1https://msnews.github.io/pens.html

Graph Neural Networks (GNNs) have achieved remarkable success in various domains but most of them are developed under the in-distribution assumption. Under out-of-distribution (OOD) settings, they suffer from the distribution shift between the training set and the test set and may not generalize well to the test distribution. Several methods have tried the invariance principle to improve the generalization of GNNs in OOD settings. However, in previous solutions, the graph encoder is immutable after the invariant learning and cannot be adapted to the target distribution flexibly. Confronting the distribution shift, a flexible encoder with refinement to the target distribution can generalize better on the test set than the stable invariant encoder. To remedy these weaknesses, we propose a Flexible invariant Learning framework for Out-Of-Distribution generalization on graphs (FLOOD), which comprises two key components, invariant learning and bootstrapped learning. The invariant learning component constructs multiple environments from graph data augmentation and learns invariant representation under risk extrapolation. Besides, the bootstrapped learning component is devised to be trained in a self-supervised way with a shared graph encoder with the invariant learning part. During the test phase, the shared encoder is flexible to be refined with the bootstrapped learning on the test set. Extensive experiments are conducted for both transductive and inductive node classification tasks. The results demonstrate that FLOOD consistently outperforms other graph OOD generalization methods and effectively improves the generalization ability.

In the field of quantitative trading, it is common practice to transform raw historical stock data into indicative signals for the market trend. Such signals are called alpha factors. Alphas in formula forms are more interpretable and thus favored by practitioners concerned with risk. In practice, a set of formulaic alphas is often used together for better modeling precision, so we need to find synergistic formulaic alpha sets that work well together. However, most traditional alpha generators mine alphas one by one separately, overlooking the fact that the alphas would be combined later. In this paper, we propose a new alpha-mining framework that prioritizes mining a synergistic set of alphas, i.e., it directly uses the performance of the downstream combination model to optimize the alpha generator. Our framework also leverages the strong exploratory capabilities of reinforcement learning (RL) to better explore the vast search space of formulaic alphas. The contribution to the combination models' performance is assigned to be the return used in the RL process. This return drives the alpha generator to find better alphas that improve upon the current set. Experimental evaluations on real-world stock market data demonstrate both the effectiveness and the efficiency of our framework for stock trend forecasting. The investment simulation results show that our framework is able to achieve higher returns compared to previous approaches.

The delayed feedback is becoming one of the main obstacles in online advertising due to the pervasive deployment of the cost-per-conversion display strategy requesting a real-time conversion rate (CVR) prediction. It makes the observed data contain a large number of fake negatives that temporarily have no feedback but will convert later. Training on such biased data distribution would severely harm the performance of models. Prevailing approaches wait for a set period of time to see if samples convert before training on them, but solutions to guaranteeing data freshness remain under-explored by current research. In this work, we propose Delayed Feed-back modeling via neural Satellite Networks (DFSN for short) for online CVR prediction. It tackles the issue of data freshness to permit adaptive waiting windows. We first assign a long waiting window for our main model to cover most of conversions and greatly reduce fake negatives. Meanwhile, two kinds of satellite models are devised to learn from the latest data, and online transfer learning techniques are utilized to sufficiently exploit their knowledge. With information from satellites, our main model can deal with the issue of data freshness, achieving better performance than previous methods. Extensive experiments on two real-world advertising datasets demonstrate the superiority of our model.

Prevailing calibration methods may fail to generalize well due to the pervasively delayed feedback issue in online advertising. That is, the labels of recent samples are more likely to be inaccurate because of the delayed feedback by users, while the old samples with complete feedback may suffer from the data shift compared to the recent ones. In this paper, we propose to calibrate conversion rate prediction models considering delayed feedback via the knowledge distillation technique. Specifically, we deploy a teacher model modeling by the samples with complete feedback to learn long-term conversion patterns and a student model modeling by the recent data to reduce the impact of data shift. We also devise a distillation loss to buoy the student model to learn from the teacher. Experimental results on two real-world advertising conversion rate prediction datasets demonstrate that our method can provide more calibrated predictions compared with the existing ones. We also exhibit that our method can be extended to different base models.

Though graph neural networks (GNNs)-based fraud detectors have received remarkable success in identifying fraudulent activities, few of them pay equal attention to models' performance and explainability. In this paper, we attempt to achieve high performance for graph-based fraud detection while considering model explainability. We propose NGS (Neural meta-Graph Search), in which the message passing process of a GNN is formalized as a meta-graph, and a differentiable neural architecture search is devised to determine the optimized message passing graph structure. We further enhance the model by aggregating multiple searched meta-graphs to make the final prediction. Experimental results on two real-world datasets demonstrate that NGS outperforms state-of-the-art baselines. In addition, the searched meta-graphs concisely describe the information used for prediction and produce reasonable explanations.

Recent years have witnessed remarkable progress on knowledge graph embedding (KGE) methods to learn the representations of entities and relations in static knowledge graphs (SKGs). However, knowledge changes over time. In order to represent the facts happening in a specific time, temporal knowledge graph (TKG) embedding approaches are put forward. While most existing models ignore the independence of semantic and temporal information. We empirically find that current models have difficulty distinguishing representations of the same entity or relation at different timestamps. In this regard, we propose a TimeLine-Traced Knowledge Graph Embedding method (TLT-KGE) for temporal knowledge graph completion. TLT-KGE aims to embed the entities and relations with timestamps as a complex vector or a quaternion vector. Specifically, TLT-KGE models semantic information and temporal information as different axes of complex number space or quaternion space. Meanwhile, two specific components carving the relationship between semantic and temporal information are devised to buoy the modeling. In this way, the proposed method can not only distinguish the independence of the semantic and temporal information, but also establish a connection between them. Experimental results on the link prediction task demonstrate that TLT-KGE achieves substantial improvements over state-of-the-art competitors. The source code will be available on https://github.com/zhangfw123/TLT-KGE.

Benefiting from the message passing mechanism, Graph Neural Networks (GNNs) have been successful on flourish tasks over graph data. However, recent studies have shown that attackers can catastrophically degrade the performance of GNNs by maliciously modifying the graph structure. A straightforward solution to remedy this issue is to model the edge weights by learning a metric function between pairwise representations of two end nodes, which attempts to assign low weights to adversarial edges. The existing methods use either raw features or representations learned by supervised GNNs to model the edge weights. However, both strategies are faced with some immediate problems: raw features cannot represent various properties of nodes (e.g., structure information), and representations learned by supervised GNN may suffer from the poor performance of the classifier on the poisoned graph. We need representations that carry both feature information and as mush correct structure information as possible and are insensitive to structural perturbations. To this end, we propose an unsupervised pipeline, named STABLE, to optimize the graph structure. Finally, we input the well-refined graph into a downstream classifier. For this part, we design an advanced GCN that significantly enhances the robustness of vanilla GCN [24] without increasing the time complexity. Extensive experiments on four real-world graph benchmarks demonstrate that STABLE outperforms the state-of-the-art methods and successfully defends against various attacks.

Recent studies on Graph Neural Networks (GNNs) point out that most GNNs depend on the homophily assumption but fail to generalize to graphs with heterophily where dissimilar nodes connect. The concept of homophily or heterophily defined previously is a global measurement of the whole graph and cannot describe the local connectivity of a node. From the node-level perspective, we find that real-world graph structures exhibit a mixture of homophily and heterophily, which refers to the co-existence of both homophilous and heterophilous nodes. Under such a mixture, we reveal that GNNs are severely biased towards homophilous nodes, suffering a sharp performance drop on heterophilous nodes. To mitigate the bias issue, we explore an Uncertainty-aware Debiasing (UD) framework, which retains the knowledge of the biased model on certain nodes and compensates for the nodes with high uncertainty. In particular, UD estimates the uncertainty of the GNN output to recognize heterophilous nodes. UD then trains a debiased GNN by pruning the biased parameters with certain nodes and retraining the pruned parameters on nodes with high uncertainty. We apply UD on both homophilous GNNs (GCN and GAT) and heterophilous GNNs (Mixhop and GPR-GNN) and conduct extensive experiments on synthetic and benchmark datasets, where the debiased model consistently performs better and narrows the performance gap between homophilous and heterophilous nodes.

The outbreak of COVID-19 burgeons newborn services on online platforms and simultaneously buoys multifarious online fraud activities. Due to the rapid technological and commercial innovation that opens up an ever-expanding set of products, the insufficient labeling data renders existing supervised or semi-supervised fraud detection models ineffective in these emerging services. However, the ever accumulated user behavioral data on online platforms might be helpful in improving the performance of fraud detection on newborn services. To this end, in this paper, we propose to pre-train user behavior sequences, which consist of orderly arranged actions, from the large-scale unlabeled data sources for online fraud detection. Recent studies illustrate accurate extraction of user intentions~(formed by consecutive actions) in behavioral sequences can propel improvements in the performance of online fraud detection. By anatomizing the characteristic of online fraud activities, we devise a model named UB-PTM that learns knowledge of fraud activities by three agent tasks at different granularities, i.e., action, intention, and sequence levels, from large-scale unlabeled data. Extensive experiments on three downstream transaction and user-level online fraud detection tasks demonstrate that our UB-PTM is able to outperform the state-of-the-art designing for specific tasks.

Conversational recommender systems (CRS) aim to provide highquality recommendations in conversations. However, most conventional CRS models mainly focus on the dialogue understanding of the current session, ignoring other rich multi-aspect information of the central subjects (i.e., users) in recommendation. In this work, we highlight that the user's historical dialogue sessions and look-alike users are essential sources of user preferences besides the current dialogue session in CRS. To systematically model the multi-aspect information, we propose a User-Centric Conversational Recommendation (UCCR) model, which returns to the essence of user preference learning in CRS tasks. Specifically, we propose a historical session learner to capture users' multi-view preferences from knowledge, semantic, and consuming views as supplements to the current preference signals. A multi-view preference mapper is conducted to learn the intrinsic correlations among different views in current and historical sessions via self-supervised objectives. We also design a temporal look-alike user selector to understand users via their similar users. The learned multi-aspect multi-view user preferences are then used for the recommendation and dialogue generation. In experiments, we conduct comprehensive evaluations on both Chinese and English CRS datasets. The significant improvements over competitive models in both recommendation and dialogue generation verify the superiority of UCCR.

Recommendation fairness has attracted great attention recently. In real-world systems, users usually have multiple sensitive attributes (e.g. age, gender, and occupation), and users may not want their recommendation results influenced by those attributes. Moreover, which of and when these user attributes should be considered in fairness-aware modeling should depend on users' specific demands. In this work, we define the selective fairness task, where users can flexibly choose which sensitive attributes should the recommendation model be bias-free. We propose a novel parameter-efficient prompt-based fairness-aware recommendation (PFRec) framework, which relies on attribute-specific prompt-based bias eliminators with adversarial training, enabling selective fairness with different attribute combinations on sequential recommendation. Both task-specific and user-specific prompts are considered. We conduct extensive evaluations to verify PFRec's superiority in selective fairness. The source codes are released in \urlhttps://github.com/wyqing20/PFRec.

Though Graph Neural Networks (GNNs) have been successful for fraud detection tasks, they suffer from imbalanced labels due to limited fraud compared to the overall userbase. This paper attempts to resolve this label-imbalance problem for GNNs by maximizing the AUC (Area Under ROC Curve) metric since it is unbiased with label distribution. However, maximizing AUC on GNN for fraud detection tasks is intractable due to the potential polluted topological structure caused by intentional noisy edges generated by fraudsters. To alleviate this problem, we propose to decouple the AUC maximization process on GNN into a classifier parameter searching and an edge pruning policy searching, respectively. We propose a model named AO-GNN (Short for AUC-oriented GNN), to achieve AUC maximization on GNN under the aforementioned framework. In the proposed model, an AUC-oriented stochastic gradient is applied for classifier parameter searching, and an AUC-oriented reinforcement learning module supervised by a surrogate reward of AUC is devised for edge pruning policy searching. Experiments on three real-world datasets demonstrate that the proposed AO-GNN patently outperforms state-of-the-art baselines in not only AUC but also other general metrics, e.g. F1-macro, G-means.

With the ever-increasing urbanization process, modeling people's spatiotemporal activities from their online traces has become a crucial task. State-of-the-art methods for this task rely on cross-modal embedding, which maps items from different modalities (e.g., location, time, text) into the same latent space. Despite their inspiring results, existing cross-modal embedding methods merely capture co-occurrences between items without modeling their high-order interactions. In this paper, we first construct two graphs from raw data records to represent the user interaction graph layer and activity graph layer and propose a hierarchical cross-modal embedding method that takes the high-order relationships into consideration. The key notion behind our method is a novel hierarchical embedding framework with meta-graphs connecting different layers. We introduce bothinter-recordandintra-recordmeta-graph structures, which enable learning distributed representations that preserve high-order proximities across graphs from different layers. Our empirical experiments on three real-world datasets demonstrate that our method not only outperforms state-of-the-art methods for spatiotemporal activity prediction, but also captures cross-modal proximity at a finer granularity.

Fraud transactions have been the major threats to the healthy development of e-commerce platforms, which not only damage the user experience but also disrupt the orderly operation of the market. User behavioral data is widely used to detect fraud transactions, and recent works show that accurate modeling of user intentions in behavioral sequences can propel further improvements on the performances. However, most existing methods treat each transaction as an independent data instance without considering the transaction-level interactions accessed by transaction attributes, e.g., information on remark, logistics, payment, device and etc., which may fail to achieve satisfactory results in more complex scenarios. In this paper, a novel heterogeneous transaction-intention network is devised to leverage the cross-interaction information over transactions and intentions, which consists of two types of nodes, namely transaction and intention nodes, and two types of edges, i.e., transaction-intention and transaction-transaction edges. Then we propose a graph neural method coined IHGAT(Intention-aware Heterogeneous Graph ATtention networks) that not only perceives sequence-like intentions, but also encodes the relationship among transactions. Extensive experiments on a real-world dataset of Alibaba platform show that our proposed algorithm outperforms state-of-the-art methods in both offline and online modes.

Lifelong learning capabilities are crucial for sentiment classifiers to process continuous streams of opinioned information on the Web. However, performing lifelong learning is non-trivial for deep neural networks as continually training of incrementally available information inevitably results in catastrophic forgetting or interference. In this paper, we propose a novel i terative network p runing with uncertainty r egularization method for l ifelong s entiment classification (IPRLS), which leverages the principles of network pruning and weight regularization. By performing network pruning with uncertainty regularization in an iterative manner, IPRLS can adapt a single BERT model to work with continuously arriving data from multiple domains while avoiding catastrophic forgetting and interference. Specifically, we leverage an iterative pruning method to remove redundant parameters in large deep networks so that the freed-up space can then be employed to learn new tasks, tackling the catastrophic forgetting problem. Instead of keeping the old-tasks fixed when learning new tasks, we also use an uncertainty regularization based on the Bayesian online learning framework to constrain the update of old tasks weights in BERT, which enables positive backward transfer, i.e. learning new tasks improves performance on past tasks while protecting old knowledge from being lost. In addition, we propose a task-specific low-dimensional residual function in parallel to each layer of BERT, which makes IPRLS less prone to losing the knowledge saved in the base BERT network when learning a new task. Extensive experiments on 16 popular review corpora demonstrate that the proposed IPRLS method significantly outperforms the strong baselines for lifelong sentiment classification. For reproducibility, we submit the code and data at: \urlhttps://github.com/siat-nlp/IPRLS .

The delayed feedback problem is one of the imperative challenges in online advertising, which is caused by the highly diversified feedback delay of a conversion varying from a few minutes to several days. It is hard to design an appropriate online learning system under these non-identical delay for different types of ads and users. In this paper, we propose to tackle the delayed feedback problem in online advertising by "Following the Prophet" (FTP for short). The key insight is that, if the feedback came instantly for all the logged samples, we could get a model without delayed feedback, namely the "prophet". Although the prophet cannot be obtained during online learning, we show that we could predict the prophet's predictions by an aggregation policy on top of a set of multi-task predictions, where each task captures the feedback patterns of different periods. We propose the objective and optimization approach for the policy, and use the logged data to imitate the prophet. Extensive experiments on three real-world advertising datasets show that our method outperforms the previous state-of-the-art baselines.

Consumer lending service is escalating in E-Commerce platforms due to its capability in enhancing buyers' purchasing power, improving average order value, and increasing revenue of the platforms. Credit risk forecasting and credit limits setting are two fundamental problems in E-Commerce/online consumer lending services. Currently, the majority of institutes rely on two-separate-step methods to resolve. First, build a rating model to evaluate credit risk, and then design heuristic strategies to set credit limits, which requires a large amount of prior knowledge and lacks theoretical justifications. In this paper, we propose an end-to-end multi-view and multi-task learning based approach named MvMoE (Multi-view-aware Mixture-of-Experts network) to solve these two problems simultaneously. First, a multi-view network with a hierarchical attention mechanism is constructed to distill users' heterogeneous financial information into shared hidden representations. Then, we jointly train these two tasks with a view-aware multi-gate mixture-of-experts network and a subsequent progressive network to improve their performances. With the real-world dataset contained 5.44 million users, we investigate the effectiveness of MvMoE. Experimental results exhibit that the proposed model is able to improve AP over 5.60% on credit risk forecasting and MAE over 9.52% on credit limits setting compared with conventional methods. Meanwhile, MvMoE has good interpretability, which better underpins the imperative demands in financial industries.

Graph-based fraud detection approaches have escalated lots of attention recently due to the abundant relational information of graph-structured data, which may be beneficial for the detection of fraudsters. However, the GNN-based algorithms could fare poorly when the label distribution of nodes is heavily skewed, and it is common in sensitive areas such as financial fraud, etc. To remedy the class imbalance problem of graph-based fraud detection, we propose a Pick and Choose Graph Neural Network (PC-GNN for short) for imbalanced supervised learning on graphs. First, nodes and edges are picked with a devised label-balanced sampler to construct sub-graphs for mini-batch training. Next, for each node in the sub-graph, the neighbor candidates are chosen by a proposed neighborhood sampler. Finally, information from the selected neighbors and different relations are aggregated to obtain the final representation of a target node. Experiments on both benchmark and real-world graph-based fraud detection tasks demonstrate that PC-GNN apparently outperforms state-of-the-art baselines.

The exploration/exploitation (E&E) dilemma lies at the core of interactive systems such as online advertising, for which contextual bandit algorithms have been proposed. Bayesian approaches provide guided exploration via uncertainty estimation, but the applicability is often limited due to over-simplified assumptions. Non-Bayesian bootstrap methods, on the other hand, can apply to complex problems by using deep reward models, but lack a clear guidance to the exploration behavior. It still remains largely unsolved to develop a practical method for complex deep contextual bandits.

Financial credit risk assessment serves as the impetus to evaluate the credit admission or potential business failure of customers in order to make early actions prior to the actual financial crisis. It aims to predict the probability that a customer may belong to a high-risk group, which is usually formulated as a binary classification problem. However, due to the lack of high-risk samples, the prevailing models suffer from the severe class-imbalance problem. Oversampling those high-risk users could alleviate this problem but the effect of noise examples is also amplified. In this paper, we propose a novel adversarial data augmentation method to solve the class imbalance problem in financial credit risk assessment. We train a generator for synthetic sample generation with a discriminator to identify real or fake instances. Besides, an auxiliary risk discriminator is trained cooperatively with the generator to assess the credit risk. Experimental results on three real-world datasets demonstrate the effectiveness of the proposed

Credit risk forecasting generally aims to evaluate the default probability of users in financial service. It is usually regarded as a binary classification problem, which suffers from the severe class-imbalance problem due to the extremely limited throngs and the concept drift problem brought by the delayed verification. In this paper, we investigate these problems in credit risk forecasting and propose a semi-supervised meta-learning based approach called TRUST (TRainable Undersampling with Self Training) to resolve. First, it decides whether to sample the data through meta-learning based reinforcement learning. Secondly, it learns the distribution of the data that have not yet shown financial performance via self-training and updates the model trained in the first step. Finally, the updated model is evaluated on the validation dataset, the result of which will be fed back through the evaluator. These three steps will be iterated until the model converges. With the real-world industrial dataset containing 1.75 million users, we investigate the effectiveness of our method. Experimental results exhibit that the proposed method is able to improve AP over 5.94% on credit risk forecasting task compared with the recent methods.

Fraud transactions obtain the rights and interests of e-commerce platforms by illegal ways, and have been the emerging threats to the healthy development of these platforms. Recently, user behavioral data is extensively exploited to detect fraud transactions, and it is usually processed as a sequence consisting of individual actions. However, such sequence-like user behaviors have logical patterns associated with user intentions, which motivates a fine-grained management strategy that binds and cuts off these actions into intention-related segments. In this paper, we devise a tree-like structure named behavior tree to reorganize the user behavioral data, in which a group of successive sequential actions denoting a specific user intention are represented as a branch on the tree. We then propose a novel neural method coined LIC Tree-LSTM(Local Intention Calibrated Tree-LSTM) to utilize the behavior tree for fraud transactions detection. In our LIC Tree-LSTM, the global user intention is captured by an attentional method applied on different branches. Then, we calibrate the entire tree by attentions within tree branches to pinpoint the balance between global and local user intentions. We investigate the effectiveness of LIC Tree-LSTM on a real-world dataset of Alibaba platform, and the experimental results show that our proposed algorithm outperforms state-of-the-art methods in both offline and online modes. Furthermore, our model provides good interpretability which helps us better understand user behaviors.

It is often observed that the probabilistic predictions given by a machine learning model can disagree with averaged actual outcomes on specific subsets of data, which is also known as the issue of miscalibration. It is responsible for the unreliability of practical machine learning systems. For example, in online advertising, an ad can receive a click-through rate prediction of 0.1 over some population of users where its actual click rate is 0.15. In such cases, the probabilistic predictions have to be fixed before the system can be deployed.

Default user detection plays one of the backbones in credit risk forecasting and management. It aims at, given a set of corresponding features, e.g., patterns extracted from trading behaviors, predicting the polarity indicating whether a user will fail to make required payments in the future. Recent efforts attempted to incorporate attributed heterogeneous information network (AHIN) for extracting complex interactive features of users and achieved remarkable success on discovering specific default users such as fraud, cash-out users, etc. In this paper, we consider default users, a more general concept in credit risk, and propose a multi-view attributed heterogeneous information network based approach coined MAHINDER to remedy the special challenges. First, multiple views of user behaviors are adopted to learn personal profile due to the endogenous aspect of financial default. Second, local behavioral patterns are specifically modeled since financial default is adversarial and accumulated. With the real datasets contained 1.38 million users on Alibaba platform, we investigate the effectiveness of MAHINDER, and the experimental results exhibit the proposed approach is able to improve AUC over 2.8% and [email protected]=0.1 over 13.1% compared with the state-of-the-art methods. Meanwhile, MAHINDER has as good interpretability as tree-based methods like GBDT, which buoys the deployment in online platforms.

In this paper, we propose a multi-task learning approach for cross-modal image-text retrieval. First, a correlation network is proposed for relation recognition task, which helps learn the complicated relations and common information of different modalities. Then, we propose a correspondence cross-modal autoencoder for cross-modal input reconstruction task, which helps correlate the hidden representations of two uni-modal autoencoders. In addition, to further improve the performance of cross-modal retrieval, two regularization terms (variance and consistency constraints) are introduced to the cross-modal embeddings such that the learned common information has large variance and is modality invariant. Finally, to enable large-scale cross-modal similarity search, a flexible binary transform network is designed to convert the text and image embeddings into binary codes. Extensive experiments on two benchmark datasets demonstrate that our model has robust superiority over the compared strong baseline methods. Source code is available at \urlhttps://github.com/daerv/DAEVR.

Recent image captioning approaches are typically trained on generation-based or retrieval-based approaches. Both methods have their advantages but limited by the disadvantages. In this paper, we propose a Unified Generation-Retrieval framework for Image Captioning (UGRIC) by using adversarial learning. Different from previous methods, the proposed UGRIC model leverages the informative contents of N-best response candidates provided by the retrieval-based model to enhance the generation-based method. In addition, to further improve the informativeness of the generated caption, we employ copying mechanism to choose words from the retrieved candidate captions and put them into proper positions of the output sequence. Experiments on MSCOCO dataset demonstrate the effectiveness of the UGRIC model through various evaluation metrics.\footnoteCode and data are available at: \urlhttp://tinyurl.com/y6z2x6ho.

Click-through rate (CTR) prediction has been one of the most central problems in computational advertising. Lately, embedding techniques that produce low-dimensional representations of ad IDs drastically improve CTR prediction accuracies. However, such learning techniques are data demanding and work poorly on new ads with little logging data, which is known as the cold-start problem.

Frequent Episode Mining (FEM), which aims at mining frequent sub-sequences from a single long event sequence, is one of the essential building blocks for the sequence mining research field. Existing studies about FEM suffer from unsatisfied scalability when faced with complex sequences as it is an NP-complete problem for testing whether an episode occurs in a sequence. In this article, we propose a scalable, distributed framework to support FEM on “big” event sequences. As a rule of thumb, “big” illustrates an event sequence is either very long or with masses of simultaneous events. Meanwhile, the events in this article are arranged in a predefined hierarchy. It derives some abstractive events that can form episodes that may not directly appear in the input sequence. Specifically, we devise an event-centered and hierarchy-aware partitioning strategy to allocate events from different levels of the hierarchy into local processes. We then present an efficient special-purpose algorithm to improve the local mining performance. We also extend our framework to support maximal and closed episode mining in the context of event hierarchy, and to the best of our knowledge, we are the first attempt to define and discover hierarchy-aware maximal and closed episodes. We implement the proposed framework on Apache Spark and conduct experiments on both synthetic and real-world datasets. Experimental results demonstrate the efficiency and scalability of the proposed approach and show that we can find practical patterns when taking event hierarchies into account.

Latent Factor Models (LFMs) based on Collaborative Filtering (CF) have been widely applied in many recommendation systems, due to their good performance of prediction accuracy. In addition to users' ratings, auxiliary information such as item features is often used to improve performance, especially when ratings are very sparse. To the best of our knowledge, most existing LFMs integrate different item features in the same way for all users. Nevertheless, the attention on different item attributes varies a lot from user to user. For personalized recommendation, it is valuable to know what feature of an item a user cares most about. Besides, the latent vectors used to represent users or items in LFMs have few explicit meanings, which makes it difficult to explain why an item is recommended to a specific user. In this work, we propose the Attention-driven Factor Model (AFM), which can not only integrate item features driven by users' attention but also help answer this "why". To estimate users' attention distributions on different item features, we propose the Gated Attention Units (GAUs) for AFM. The GAUs make it possible to let the latent factors "talk", by generating user attention distributions from user latent vectors. With users' attention distributions, we can tune the weights of item features for different users. Moreover, users' attention distributions can also serve as explanations for our recommendations. Experiments on several real-world datasets demonstrate the advantages of AFM (using GAUs) over competitive baseline algorithms on rating prediction.

Episode Rule Mining is a popular framework for discovering sequential rules from event sequential data. However, traditional episode rule mining methods only tell that the consequent event is likely to happen within a given time interval after the occurrence of the antecedent events. As a result, they cannot satisfy the requirement of many time sensitive applications, such as program security trading and intelligent transportation management due to the lack of fine-grained response time. In this study, we come up with the concept offixed-gap episodeto address this problem. A fixed-gap episode consists of an ordered set of events where the elapsed time between any two consecutive events is a constant. Based on this concept, we formulate the problem of miningprecise-positioning episode rulesin which the occurrence time of each event in the consequent is clearly specified. In addition, we develop a trie-based data structure to mine such precise-positioning episode rules with several pruning strategies incorporated for improving the performance as well as reducing memory consumption. Experimental results on real datasets show the superiority of our proposed algorithms.

Episode Rule Mining is a popular framework for discovering sequential rules from event sequential data. However, traditional episode rule mining methods only tell that the consequent event is likely to happen within a given time intervals after the occurrence of the antecedent events. As a result, they cannot satisfy the requirement of many time sensitive applications, such as program security trading due to the lack of fine-grained response time. In this study, we come up with the concept of fixed-gap episode to address this problem. A fixed-gap episode consists of an ordered set of events where the elapsed time between any two consecutive events is a constant. Based on this concept, we formulate the problem of mining precise-positioning episode rules in which the occurrence time of each event in the consequent is clearly specified. In addition, we develop a triebased data structure to mine such precise-positioning episode rules with several pruning strategies incorporated for improving the performance as well as reducing memory consumption. Experimental results on real datasets show the superiority of our proposed algorithms.

Frequent episode mining is a popular framework for discovering sequential patterns from sequence data. Previous studies on this topic usually process data offline in a batch mode. However, for fast-growing sequence data, old episodes may become obsolete while new useful episodes keep emerging. More importantly, in time-critical applications we need a fast solution to discovering the latest frequent episodes from growing data. To this end, we formulate the problem of Online Frequent Episode Mining (OFEM). By introducing the concept of last episode occurrence within a time window, our solution can detect new minimal episode occurrences efficiently, based on which all recent frequent episodes can be discovered directly. Additionally, a trie-based data structure, episode trie, is developed to store minimal episode occurrences in a compact way. We also formally prove the soundness and completeness of our solution and analyze its time as well as space complexity. Experiment results of both online and offline FEM on real data sets show the superiority of our solution.