WWW 2025 Papers

As large language models (LLMs) increasingly power web applications, including social networks, the challenge of moderating hate speech has become a critical concern for the Web. These LLM-powered applications, while offering near-human interaction capabilities, are vulnerable to harmful or biased content due to imperfect training data scraped from the Web. Current hate speech detection methods often struggle with limited annotated data, especially for real-time moderation on these platforms. This paper introduces Sentiment-Aided Hate Speech Detection (SAHSD), a novel approach designed to enhance hate speech detection specifically in LLM-powered web applications. By treating hate speech detection as a few-shot learning task, SAHSD utilizes sentiment analysis to refine pre-trained language models (LM) for improved accuracy in recognizing harmful content. SAHSD first employs publicly available sentiment datasets to train a sentiment analysis model, which is then fine-tuned by merging sentiment prompts with hate speech prompts, enabling efficient and accurate detection even with limited training samples. The effectiveness of SAHSD is demonstrated through experiments on widely used web-sourced datasets like SBIC and HateXplain. SAHSD achieves an exceptional F1-score of 0.99 with only 64 training samples and outperforms advanced techniques such as ToKen, MRP, and HARE, with significant improvements of 33% on SBIC and 95% on HateXplain. SAHSD surpasses GPT-4 in generalization performance across multiple datasets, showing an 8% improvement when trained on equal-sized samples. These results underscore SAHSD's potential to enhance content moderation in LLM-driven web platforms, contributing to a safer, more inclusive and accountable Web ecosystem.

In the realm of Intelligent Tutoring System (ITS), the accurate assessment of students' knowledge states through Knowledge Tracing (KT) is crucial for personalized learning. However, due to data bias, $i.e.$, the unbalanced distribution of question groups ($e.g.$, concepts), conventional KT models are plagued by cognitive bias, which tends to result in cognitive underload for overperformers and cognitive overload for underperformers. More seriously, this bias is amplified with the exercise recommendations by ITS. After delving into the causal relations in the KT models, we identify the main cause as the confounder effect of students' historical correct rate distribution over question groups on the student representation and prediction score. Towards this end, we propose a Disentangled Knowledge Tracing (DisKT) model, which separately models students' familiar and unfamiliar abilities based on causal effects and eliminates the impact of the confounder in student representation within the model. Additionally, to shield the contradictory psychology ($e.g.$, guessing and mistaking) in the students’ biased data, DisKT introduces a contradiction attention mechanism. Furthermore, DisKT enhances the interpretability of the model predictions by integrating a variant of Item Response Theory. Experimental results on 11 benchmarks and 3 synthesized datasets with different bias strengths demonstrate that DisKT significantly alleviates cognitive bias and outperforms 14 baselines in evaluation accuracy. Our code and datasets are available at https://anonymous.4open.science/r/DisKT.

Recent years have witnessed the prevalent integration of Large Language Models (LLMs) in various Web applications, such as search engines and recommender systems. As an emerging technique, instruction tuning aims to align pre-trained LLMs as capable chatbots that excel at following human instructions. Previous research indicates that selecting an appropriate subset of a large instruction dataset can enhance the capabilities of LLMs and reduce training costs. However, existing works tend to overlook external correlations between instruction examples during data selection process, which can introduce potential bias and lead to sub-optimal performance. To bridge this gap, we formalize this problem from graph influence maximization perspective and propose Uncertainty-aware influence Maximization (UniMax), a data selection framework that explicitly incorporates the complex inter-dependencies within instruction data. Specifically, we first define a latent instruction graph, treating each instruction example as a graph node and representing their implicit relations as graph edges. Instead of solely relying on heuristic metrics for graph construction, we develop a self-supervised graph learner to uncover the latent structure beyond surface-level feature correlations. After that, we propose an uncertainty-aware influence function to score each example on the instruction graph, allowing a simple greedy algorithm to select a valuable subset that embodies both high influence and uncertainty with an approximation guarantee. Extensive experiments on public datasets show that the proposed approach can significantly enhance model capabilities, underscoring the importance of exploiting data dependencies in instruction data selection.

A decentralized identity system that can provide users with self-sovereign digital identities to facilitate complete control over their own data is paramount to Web 3.0. The account system on blockchain is an ideal archetype for realizing Web 3.0 decentralized identity. However, a disadvantage of such completely anonymous identity system is that users can create multiple accounts without authentication to obfuscate their activities on the blockchain. In particular, the current anonymous blockchain account system cannot accurately register the social relationships and interactions between real human users, given the amorphous mappings between users and blockchain identities. This work proposes zkBID, a zero-knowledge blockchain-account-based Web 3.0 decentralized identity scheme, to overcome endemic mistrust in blockchain account systems. zkBID links souls (blockchain accounts) to humans (users’ personhood credentials) in a one-to-one manner to truly reflect the social relationships and interactions between humans on the blockchain. zkBID conceals the one-to-one relationships between blockchain accounts and users’ personhood credentials for privacy protection using linkable ring signature. Thus, with zkBID, the users’ blockchain accounts are credible anonymously. Importantly, zkBID is fully decentralized: all user-related data are generated by users and verified by smart contracts on the blockchain. We implemented zkBID and built a blockchain test network for evaluation purposes. Our tests demonstrate the effectiveness of zkBID and suggest proper ways to configure zkBID system parameters.

Due to the growing demand for privacy protection, encrypted tunnels have become increasingly popular among mobile app users, which brings new challenges for app fingerprinting (AF)-based network management. Existing methods primarily transfer traditional AF methods to encrypted tunnels directly, ignoring the core obfuscation and re-encapsulation mechanism of encrypted tunnels, thus resulting in unsatisfactory performance. In this paper, we propose DecETT, a dual decouple-based semantic enhancement method for accurate AF under encrypted tunnels. Specifically, DecETT improves AF under encrypted tunnels from two perspectives: app-specific feature enhancement and irrelevant tunnel feature decoupling. Considering the obfuscated app-specific information in encrypted tunnel traffic, DecETT introduces TLS traffic with stronger app-specific information as a semantic anchor to guide and enhance the fingerprint generation for tunnel traffic. Furthermore, to address the app-irrelevant tunnel feature introduced by the re-encapsulation mechanism, DecETT is designed with a dual decouple-based fingerprint enhancement module, which decouples the tunnel feature and app semantic feature from tunnel traffic separately, thereby minimizing the impact of tunnel features on accurate app fingerprint extraction. Evaluation under five prevalent encrypted tunnels indicates that DecETT outperforms state-of-the-art methods in accurate AF under encrypted tunnels, and further demonstrates its superiority under tunnels with more complicated obfuscation. Project page: https://github.com/DecETT/DecETT

Contemporary datasets sourced from the web often adopt a multi-view format, collecting data from diverse sources, domains, or modules. Existing methodologies employed to analyze such datasets frequently overlook or inaccurately allocate the view-weights, pivotal metrics reflecting each view's significance. This work introduces EVA-MVC, a simple yet effective algorithm designed for Equitable View-weight Allocation (EVA) seamlessly integrated with arbitrary Multi-view Clustering (MVC) methods. Within the EVA phase, we establish theoretical connections between view supplementarity and Multi-view Subspace Learning (MSL), leading to the partiton of views into View Communities (VCs) based on these foundational principles. These VCs exhibit internal supplementarity similarities, facilitating Equitable View-weights Allocation through VC-specific MSL. The proposed EVA process precedes and operates independently of traditional or SOTA MVC approaches, requiring no additional processing or specialized design, making it an ideal preprocessing step for MVC applications. Through comprehensive evaluations across diverse multi-view datasets, our findings reveal that our EVA significantly enhances the effectiveness of mainstream MVC frameworks, resulting in a notable performance improvement.

In conversational recommender systems, interactively presenting queries and leveraging user feedback are crucial for efficiently estimating user preferences and improving recommendation quality. Selecting optimal queries in these systems is a significant challenge that has been extensively studied as a sequential decision problem. The expected value of information (EVOI), which computes the expected reward improvement, provides a principled criterion for query selection. However, it is computationally expensive and lacks theoretical performance guarantees. Conversely, conversational bandits offer provable regret upper bounds, but their query selection strategies yield only marginal regret improvements over non-conversational approaches. To address these limitations, we integrate EVOI within the conversational bandit framework by proposing a new conversational mechanism featuring two key techniques: (1) gradient-based EVOI, which replaces the complex Bayesian updates in conventional EVOI with efficient stochastic gradient descent, significantly reducing computational complexity and facilitating theoretical analysis; and (2) smoothed key term contexts, which enhance exploration by adding random perturbations to uncover more specific user preferences. Our approach applies to both Bayesian (Thompson Sampling) and frequentist (UCB) variants of conversational bandits. We introduce two new algorithms, ConTS-EVOI and ConUCB-EVOI, and rigorously prove that they achieve substantially tighter regret bounds, with both algorithms offering a $\sqrt{d}$ improvement in their dependence on the time horizon $T$, where $d$ is the dimension of the feature space. Extensive evaluations on synthetic and real-world datasets validate the effectiveness of our methods.

WebAssembly (Wasm), a binary instruction format designed for efficient cross-platform execution, has rapidly become a foundational web standard, widely adopted in browsers, client-side, and server-side applications. However, its growing popularity has led to an increase in Wasm-targeted malware, including cryptojackers and obfuscated malicious scripts, which pose significant threats to web security. In spite of progress in deep learning based detection methods for Wasm malware, such as MINOS, these approaches face substantial performance degradation in adversarial environments. In our experiments, MINOS’s detection accuracy dropped to 49.90\% under adversarial attacks, revealing critical vulnerabilities. To address this, we introduce \textbf{WasmGuard}, a robust malware detection framework tailored for Wasm. WasmGuard employs FGSM-based adversarial training with prior-based initialization for perturbation bytes in customized sections, coupled with a novel adversarial contrastive learning objective. Using our large-scale dataset, \textbf{WasmMal-15K} (publicly available), WasmGuard outperforms six competing methods, achieving up to 99.20\% Robust Accuracy and 99.93\% Standard Accuracy under PGD-50 adversarial attacks, while maintaining low training overhead. Additionally, we have released \textbf{WebChecker}, a WasmGuard-powered browser plugin, providing real-time protection against malicious Wasm files.

Predicting answers to queries over knowledge graphs is called a complex reasoning task because answering a query requires subdividing it into subqueries. Existing query embedding methods use this decomposition to compute the embedding of a query as the combination of the embedding of the subqueries. This requirement limits the answerable queries to queries having a single free variable and being decomposable, which are called tree-form queries and correspond to the $SROI^-$ description logic. In this paper, we define a more general set of queries, called DAG queries, formulate a description logic corresponding to them, called DAG-DL, propose a query embedding method for them, called DAGE, and a new benchmark to evaluate query embeddings on them. Given the computational graph of a DAG query, DAGE combines the possibly multiple paths between two nodes into a single path with a trainable operator that represents the intersection of relations and learns DAG-DL tautologies. We show that it is possible to implement DAGE on top of existing query embedding methods, and we empirically measure the outstanding improvement of our method over the results of vanilla methods evaluated in tree-form queries that result in relaxing the DAG queries of our proposed benchmark.

IP addresses with web services are crucial in the Internet ecosystem. Classifying these addresses by industry and organization offers valuable insights into the entities utilizing them, enabling more efficient network management and enhanced security. Previous work in website classification and Internet management struggles to offer an IP-level perspective of the industries of web services due to their limited industry categories or potential industry inconsistencies between IP address owners and AS owners. To this end, we present IPdb, an IP-level industry categorization dataset. To construct the dataset, we developed LLMIC, a Large Language Model-based Industry Categorization framework with a precision of nearly 96\%. IPdb serves as a labeled database for future endeavors in developing IP-level industry classifiers, encompassing over 200 million IP addresses. Furthermore, our study indicates that 30\% $\sim$ 50\% of organizations within critical infrastructure industries deploy web servers across multiple ASes. Our study also validates the problem of mismatched granularity in industry categorization at the AS level with 87.83\% ASes in IPv4 and 72.96\% ASes in IPv6 containing IP addresses from different industries.

In web data, advertising images are crucial for capturing user attention and improving advertising effectiveness. Most existing methods generate background for products primarily focus on the aesthetic quality, which may fail to achieve satisfactory online performance. To address this limitation, we explore the use of Multimodal Large Language Models (MLLMs) for generating advertising images by optimizing for Click-Through Rate (CTR) as the primary objective. Firstly, we build targeted pre-training tasks, and leverage a large-scale e-commerce multimodal dataset to equip MLLMs with initial capabilities for advertising image generation tasks. To further improve the CTR of generated images, we propose a novel reward model to fine-tune pre-trained MLLMs through Reinforcement Learning (RL), which can jointly utilize multimodal features and accurately reflect user click preferences. Meanwhile, a product-centric preference optimization strategy is developed to ensure that the generated background content aligns with the product characteristics after fine-tuning, enhancing the overall relevance and effectiveness of the advertising images. Extensive experiments have demonstrated that our method achieves state-of-the-art performance in both online and offline metrics. We will release our code and weights upon acceptance of the paper.

Community detection is a critical task in graph theory, social network analysis, and bioinformatics, where communities are defined as clusters of densely interconnected nodes. However, detecting communities in large-scale networks with millions of nodes and billions of edges remains challenging due to the inefficiency and unreliability of existing methods. Moreover, many current approaches are limited to specific graph types, such as unweighted or undirected graphs, reducing their broader applicability. To address these limitations, we propose a novel heuristic community detection algorithm inspired by game theory, termed \framework, which identifies communities by minimizing the network's 2-dimensional (2D) structural entropy. In this potential game model, nodes decide whether to stay or transfer to another community based on a strategy that maximizes a 2D structural entropy utility function. Additionally, we introduce a structural entropy-based node overlapping heuristic to detect overlapping communities. The algorithm operates with near-linear time complexity, enabling efficient community detection in large-scale networks. Experimental results on real-world networks demonstrate that CoDeSEG is the fastest method available and achieves state-of-the-art performance in overlapping normalized mutual information (ONMI) and F1 scores.

Taxonomy completion aims to integrate new concepts into existing taxonomies by determining their appropriate hypernym and hyponym. While semantic and structural information are crucial for this task, existing approaches often struggle to balance these aspects effectively. In this paper, we propose **COMI**, an efficient taxonomy completion framework that leverages large language models (LLMs) to capture both semantic and structural information in a unified manner. COMI **co**mpresses node semantics into token representations, enabling LLMs to efficiently process the input structure composed of these tokens. To enhance the model's understanding of the structure, a further fine-tuning process using contrastive learning with **mi**xup data augmentation is applied, where mixup generates diverse and challenging negative samples. Through these innovations, COMI improves the integration of semantic and structural information, leading to more accurate taxonomy completion. The experimental results on three real-world datasets demonstrate that COMI achieves state-of-the-art performance while showing up to 284$\times$ faster inference compared to the previous best method. Our code and compressed tokens will be available for further study upon publication.

In recent years, a large number of on-chain attacks have emerged in the blockchain empowered Web3 ecosystem. In the year of 2023 alone, on-chain attacks have caused losses of over \$585 million. Attackers use blockchain transactions to carry out on-chain attacks, for example, exploiting vulnerabilities or business logic flaws in Web3 applications. A wealth of efforts have been devoted to detecting on-chain attack transactions through expert patterns and machine learning techniques. However, in this ever-evolving ecosystem, the performance of current methods is limited in detecting new on-chain attacks, due to the obsoleting of attack recognition patterns or the reliance on on-chain attack samples. In this paper, we propose a universal approach for detecting on-chain attacks even when there are few or even no new on-chain attack samples. Specifically, an in-depth analysis of the transaction characteristics is conducted, and we propose a new insight to train a generic attack transaction detecting model, i.e., transaction reconstruction. Particularly, to overcome the over-fitting in the transaction reconstruction task, we use the web-scale function comments related to transactions as supervision information, rather than expert-confirmed labels. Experimental results demonstrate that the proposed approach surpasses the supervised state-of-the-art by 13\% in AUC, with just 30 known on-chain attack samples. Moreover, without any known attack samples, our method can still detect new on-chain attacks in the wild (with a precision of 61.83\%). Among attacks detected in the wild, we confirm 1,692 address poisoning attacks, a new type of on-chain attack targeting token holders. Our code is available at: https://anonymous.4open.science/r/6F40.

The resistance diameter of a graph is the maximum resistance distance among all pairs of nodes in the graph, which has found various applications in many scenarios. However, direct computation of resistance diameter involves the pseudoinverse of graph Laplacian, which takes cubic time and is thus infeasible for huge networks with millions of nodes. In this paper, we consider the computation and optimization problems for resistance diameter of a graph. First, we develop a nearly linear time algorithm to approximate the resistance diameter, which has a theoretically guaranteed error. Then, we propose and study an optimization problem of adding a fixed number of edges to a graph, such that the resistance diameter of the resulting graph is minimized. We show that this optimization problem is NP-hard, and that the objective function is non-supermodular but monotone. Moreover, we propose two fast heuristic algorithms to approximately solve this problem. Finally, we conduct extensive experiments on different networks with sizes up to one million nodes, demonstrating the superiority of our algorithms in terms of efficiency and effectiveness.

Code summarization aims to generate concise natural language descriptions of source code, helping developers to acquaint with software systems and reduce maintenance costs. Existing code summarization approaches widely employ attention mechanisms to assess the relevance between nodes in the Abstract Syntax Tree (AST), which generates context vectors that reflect the semantics of the source code. However, these approaches with AST fail to extract other granular features, such as token sequences and Control Flow Graphs (CFG), which suffer from severe semantic gaps when capturing data and control dependencies. To address this issue, we design an enhanced semantic extractor with multi-granularity feature fusion (MGF-ESE) to improve the model capability in comprehending and processing the overall semantics of the code. Specifically, to process the AST more effectively, we present a novel AST generation method with compresses the scale of nodes to enhance the semantic information of each node. Then, we present a disentangled attention mechanism based on relative positional embeddings for further encoding. Moreover, we extract the token sequences and CFG of source code to supplement the syntactic and structural information, and further fuse them with the AST separately through cross-attention modules. Finally, extensive experiments on two public datasets show that MGF-ESE outperforms the state-of-the-arts with higher-quality code summaries on key metrics, including BLEU, METEOR, and ROUGE.

With the widespread use of social media, organizations, and individuals use these platforms to raise funds and support causes. Unfortunately, this has led to the rise of scammers in soliciting fraudulent donations. In this study, we conduct a large-scale analysis of donation-based scams on social media platforms. More specifically, we studied profile creation and scam operation fraudulent donation solicitation on X, Instagram, Facebook, YouTube, and Telegram. By collecting data from 151,966 accounts and their 3,053,333 posts related to donations between March 2024 and May 2024, we identified 832 scammers using various techniques to deceive users into making fraudulent donations. Analyzing the fraud communication channels such as phone number, email, and external URL linked, we show that these scamming accounts perform various fraudulent donation schemes, including classic abuse such as fake fundraising website setup, crowdsourcing fundraising, and asking users to communicate via email, phone, and pay via various payment methods. Through collaboration with industry partners PayPal and cryptocurrency abuse database Chainabuse, we further validated the scams and measured the financial losses on these platforms. Our study highlights significant weaknesses in social media platforms' ability to protect users from fraudulent donations. Additionally, we recommended social media platforms, and financial services for taking proactive steps to block these fraudulent activities. Our study provides a foundation for the security community and researchers to automate detecting and mitigating fraudulent donation solicitation on social media platforms.

In the graph domain, SSL has garnered significant attention, particularly in employing Graph Neural Networks (GNNs) with pretext tasks originally designed for other domains, such as contrastive learning and feature reconstruction. However, it remains uncertain whether these methods effectively reflect essential graph properties, such as representation similarity with its neighbors. We observe that existing methods position opposite ends of a spectrum driven by the graph embedding smoothness, with each end corresponding to outperformance on specific downstream tasks. Further insights suggest that balancing between the extremes can lead to improved performance across a wider range of downstream tasks. To find the balance respective to the graph embedding smoothness, we decompose the SSL objective into three terms, which are derived by incorporating the neighbor representation variable through the lens of information theory. A framework, \textbf{\mname{}} (\textbf{B}alancing \textbf{S}moothness in \textbf{G}raph SSL), introduces novel loss functions designed to supplement the representation quality in graph-based SSL by optimizing the derived three terms: neighbor loss, minimal loss, and divergence loss. We present a rigorous theoretical analysis of the effects of these loss functions, highlighting their significance from both the SSL and graph smoothness perspectives. Extensive experiments on multiple real-world datasets across node classification and link prediction consistently demonstrate that \mname{} achieves state-of-the-art performance, outperforming existing methods. Our implementation code is available at \url{https://anonymous.4open.science/r/BSG-2025/}.

Next Point-of-Interest (POI) recommendation has become a crucial task in Location-Based Social Networks (LBSNs), which provide personalized recommendations by predicting the user's next check-in locations. Commonly used models including Recurrent Neural Networks (RNNs) and Graph Convolutional Networks (GCNs) have been widely explored. However, these models face significant challenges, including the difficulty of capturing the hierarchical and tree-like structure of POIs in Euclidean space and the sparsity problem inherent in POI recommendations. To address these challenges, we propose a Hyperbolic Variational Graph Auto-Encoder (HVGAE) for next POI recommendation. Specifically, we utilize a Hyperbolic Graph Convolutional Network (Hyperbolic GCN) to model hierarchical structures and tree-like relationships by converting node embeddings from euclidean space to hyperbolic space. Then we use Variational Graph Auto-Encoder (VGAE) to convert node embeddings to probabilistic distributions, enhancing the capture of deeper latent features and providing a more robust model structure. Furthermore, we combine the Mamba4Rec recommender and Rotary Position Embedding (RoPE) and propose Rotary Position Mamba (RPMamba) to effectively utilize POI embeddings rich in sequential information, which improves the accuracy of the next POI recommendation. Extensive experiments on three public datasets demonstrate the superior performance of the HVGAE model.