arXiv subjects

Shuyu Jiang

Publications and source records attributed to Shuyu Jiang.

12 recordsLinked to original sources

arXiv2026

ASCon: A Direction-Aware Reciprocal Agent--Step Contextualization Model for Failure Attribution in Multi-Agent Systems

Failure attribution in LLM-based multi-agent systems (MAS) aims to answer who caused failures, when they occurred, and why by identifying responsible targets including faulty agents, erroneous steps, and failure modes. Existing methods have primarily focused on developing dedicated models for specific attribution targets, with limited attention to the evidential dependencies among them. Despite these attribution targets are different, they rely on common diagnostic evidence from MAS trajectories, including task constraints, agent roles, behavioral histories and inter-agent interactions. This commonality motivates us to develop a unified representation model that aggregates the trajectory evidence into individual agent and step representations, which can subsequently be adapted to different attribution targets. Accordingly, we propose ASCon, a direction-aware reciprocal \textbf{A}gent--\textbf{S}tep \textbf{Con}textualization model for multiple failure attribution targets. ASCon introduces direction-aware graph attention to model execution context, masked step-to-agent attention to construct behavior-aware agent representations, and agent-conditioned step contextualization to incorporate agent context back into step representations. The resulting contextualized representations enable different attribution targets through lightweight target-specific heads. Experiments show that ASCon can improve faulty-agent detection by 5.83\%+ in micro-accuracy, faulty-step detection by 10.63\%+ in micro-accuracy, and failure-mode detection by 14.73\%+ in Macro-F1. Meanwhile, it can also substantially enhance the LLM-based methods' attribution capabilities in out-of-domain scenarios.

cs.MA↗

arXiv2026

EvoJail: Evolutionary Diverse Jailbreak Prompt Generation for Large Language Models

As LLMs continue to shape real-world applications, automated jailbreak generation becomes essential to reveal safety weaknesses and guide model improvement. Existing automatic jailbreak generation methods have not yet fully considered two important aspects: adaptability to evolving safety-finetuned models, which affects their effectiveness on newer model versions, and diversity in generated prompts, which can cause narrow or repetitive attack patterns. To address these issues, we propose EvoJail, an instruction-fusion-driven evolutionary jailbreak generation framework that formalizes jailbreak prompt generation as a multi-objective black-box optimization problem and leverages the principles of evolutionary algorithms to search for jailbreak prompts that can adapt across different model versions and exhibit diverse attack patterns. Specifically, EvoJail integrates jailbreak prompt generation into an iterative evolutionary loop, where at each iteration candidate prompts are evaluated directly against the target model and then selected and varied based on the target model's responses, enabling the generation process to continuously adapt to model updates. To enhance diversity, EvoJail introduces field-aware instruction fusion to construct diverse starting points and incorporates diversity-aware objectives into the evolutionary fitness function, guiding the search toward prompts with richer semantic variation, while further designing multi-level LLM-based mutation operators that modify prompt structures at different granularities to promote structural diversity throughout the evolutionary process. Results demonstrate that EvoJail has stronger adaptability and can achieve over $93\%$ attack success rate and more than $5.6\%$ improvement in diversity metrics over state-of-the-art methods.

cs.NE↗

arXiv2026

One Jailbreak, Many Tongues: Learning Language-Insensitive Intention Representations for Multilingual Jailbreak Detection

Large language models (LLMs) are increasingly deployed in applications for global multilingual users, yet safety training remains concentrated in dominant languages and has not progressed in parallel with multilingual capability, creating exploitable gaps for jailbreak attacks. Current jailbreak defenses are largely developed and evaluated in dominant languages, and their effectiveness is limited by the scarcity of aligned multilingual supervision and representations dispersion caused by language variation. To address this issue, we propose MLJailDe, a multilingual jailbreak detection framework designed to improve both multilingual robustness and cross-lingual generalization. MLJailDe first introduces a multilingual back-translation data augmentation algorithm to construct a semantically consistent and functionally effective dataset spanning 11 languages, consisting of 2,232 benign and 1,239 jailbreak samples. On this basis, MLJailDe employs relative-distance constraints to reduce cross-lingual representation dispersion and encourage jailbreak prompts with similar intent to form consistent clusters across languages, while an imbalance-aware classification objective is further used to alleviate class imbalance and learn more reliable multilingual decision boundaries. Experimental results show that MLJailDe outperforms state-of-the-art baselines across multiple languages, achieving an F1 score of 98.5\%, and obtains an average F1 score of 97.1\% on unseen languages, demonstrating strong effectiveness and cross-lingual generalization.

cs.CL↗

arXiv2026

Large language models for spreading dynamics in complex systems

Spreading dynamics is a central topic in the physics of complex systems and network science, providing a unified framework for understanding how information, behaviors, and diseases propagate through interactions among system units. In many propagation contexts, spreading processes are influenced by multiple interacting factors, such as information expression patterns, cultural contexts, living environments, cognitive preferences, and public policies, which are difficult to incorporate directly into classical modeling frameworks. Recently, large language models (LLMs) have exhibited strong capabilities in natural language understanding, reasoning, and generation, enabling explicit perception of semantic content and contextual cues in spreading processes, thereby supporting the analysis of the different influencing factors. Beyond serving as external analytical tools, LLMs can also act as interactive agents embedded in propagation systems, potentially influencing spreading pathways and feedback structures. Consequently, the roles and impacts of LLMs on spreading dynamics have become an active and rapidly growing research area across multiple research disciplines. This review provides a comprehensive overview of recent advances in applying LLMs to the study of spreading dynamics across two representative domains: digital epidemics, such as misinformation and rumors, and biological epidemics, including infectious disease outbreaks. We first examine the foundations of epidemic modeling from a complex-systems perspective and discuss how LLM-based approaches relate to traditional frameworks. We then systematically review recent studies from three key perspectives, which are epidemic modeling, epidemic detection and surveillance, and epidemic prediction and management, to clarify how LLMs enhance these areas. Finally, open challenges and potential research directions are discussed.

physics.soc-ph↗

arXiv2025

Network Alignment

Complex networks are frequently employed to model physical or virtual complex systems. When certain entities exist across multiple systems simultaneously, unveiling their corresponding relationships across the networks becomes crucial. This problem, known as network alignment, holds significant importance. It enhances our understanding of complex system structures and behaviours, facilitates the validation and extension of theoretical physics research about studying complex systems, and fosters diverse practical applications across various fields. However, due to variations in the structure, characteristics, and properties of complex networks across different fields, the study of network alignment is often isolated within each domain, with even the terminologies and concepts lacking uniformity. This review comprehensively summarizes the latest advancements in network alignment research, focusing on analyzing network alignment characteristics and progress in various domains such as social network analysis, bioinformatics, computational linguistics and privacy protection. It provides a detailed analysis of various methods' implementation principles, processes, and performance differences, including structure consistency-based methods, network embedding-based methods, and graph neural network-based (GNN-based) methods. Additionally, the methods for network alignment under different conditions, such as in attributed networks, heterogeneous networks, directed networks, and dynamic networks, are presented. Furthermore, the challenges and the open issues for future studies are also discussed.

physics.soc-ph↗

arXiv2024

ReZG: Retrieval-Augmented Zero-Shot Counter Narrative Generation for Hate Speech

The proliferation of hate speech (HS) on social media poses a serious threat to societal security. Automatic counter narrative (CN) generation, as an active strategy for HS intervention, has garnered increasing attention in recent years. Existing methods for automatically generating CNs mainly rely on re-training or fine-tuning pre-trained language models (PLMs) on human-curated CN corpora. Unfortunately, the annotation speed of CN corpora cannot keep up with the growth of HS targets, while generating specific and effective CNs for unseen targets remains a significant challenge for the model. To tackle this issue, we propose Retrieval-Augmented Zero-shot Generation (ReZG) to generate CNs with high-specificity for unseen targets. Specifically, we propose a multi-dimensional hierarchical retrieval method that integrates stance, semantics, and fitness, extending the retrieval metric from single dimension to multiple dimensions suitable for the knowledge that refutes HS. Then, we implement an energy-based constrained decoding mechanism that enables PLMs to use differentiable knowledge preservation, countering, and fluency constraint functions instead of in-target CNs as control signals for generation, thereby achieving zero-shot CN generation. With the above techniques, ReZG can integrate external knowledge flexibly and improve the specificity of CNs. Experimental results show that ReZG exhibits stronger generalization capabilities and outperforms strong baselines with significant improvements of 2.0%+ in the relevance and 4.5%+ in the countering success rate metrics.

cs.CL↗

arXiv2023

A Low-cost, High-impact Node Injection Approach for Attacking Social Network Alignment

Social network alignment (SNA) holds significant importance for various downstream applications, prompting numerous professionals to develop and share SNA tools. Unfortunately, these tools can be exploited by malicious actors to integrate sensitive user information, posing cybersecurity risks. While many researchers have explored attacking SNA (ASNA) through a network modification attack way, practical feasibility remains a challenge. This paper introduces a novel approach, the node injection attack. To overcome the problem of modeling and solving within a limited time and balancing costs and benefits, we propose a low-cost, high-impact node injection attack via dynamic programming (DPNIA) framework. DPNIA models ASNA as a problem of maximizing the number of confirmed incorrect correspondent node pairs who have a greater similarity scores than the pairs between existing nodes, making ASNA solvable. Meanwhile, it employs a cross-network evaluation method to identify node vulnerability, facilitating a progressive attack from easy to difficult. Additionally, it utilizes an optimal injection strategy searching method, based on dynamic programming, to determine which links should be added between injected nodes and existing nodes, thereby achieving a high impact for attack effectiveness at a low cost. Experiments on four real-world datasets consistently demonstrate that DPNIA consistently and significantly outperforms various attack baselines.

cs.SI↗

arXiv2023

Prompt Packer: Deceiving LLMs through Compositional Instruction with Hidden Attacks

Recently, Large language models (LLMs) with powerful general capabilities have been increasingly integrated into various Web applications, while undergoing alignment training to ensure that the generated content aligns with user intent and ethics. Unfortunately, they remain the risk of generating harmful content like hate speech and criminal activities in practical applications. Current approaches primarily rely on detecting, collecting, and training against harmful prompts to prevent such risks. However, they typically focused on the "superficial" harmful prompts with a solitary intent, ignoring composite attack instructions with multiple intentions that can easily elicit harmful content in real-world scenarios. In this paper, we introduce an innovative technique for obfuscating harmful instructions: Compositional Instruction Attacks (CIA), which refers to attacking by combination and encapsulation of multiple instructions. CIA hides harmful prompts within instructions of harmless intentions, making it impossible for the model to identify underlying malicious intentions. Furthermore, we implement two transformation methods, known as T-CIA and W-CIA, to automatically disguise harmful instructions as talking or writing tasks, making them appear harmless to LLMs. We evaluated CIA on GPT-4, ChatGPT, and ChatGLM2 with two safety assessment datasets and two harmful prompt datasets. It achieves an attack success rate of 95%+ on safety assessment datasets, and 83%+ for GPT-4, 91%+ for ChatGPT (gpt-3.5-turbo backed) and ChatGLM2-6B on harmful prompt datasets. Our approach reveals the vulnerability of LLMs to such compositional instruction attacks that harbor underlying harmful intentions, contributing significantly to LLM security development. Warning: this paper may contain offensive or upsetting content!

cs.CL↗

arXiv2022

Network structural perturbation against interlayer link prediction

Interlayer link prediction aims at matching the same entities across different layers of the multiplex network. Existing studies attempt to predict more accurately, efficiently, or generically from the aspects of network structure, attribute characteristics, and their combination. Few of them analyze the effects of intralayer links. Namely, few works study the backbone structures which can effectively preserve the predictive accuracy while dealing with a smaller number of intralayer links. It can be used to investigate what types of intralayer links are most important for correct prediction. Are there any intralayer links whose presence leads to worse predictive performance than their absence, and how to attack the prediction algorithms at the minimum cost? To this end, two kinds of network structural perturbation methods are proposed. For the scenario where the structural information of the whole network is completely known, we offer a global perturbation strategy that gives different perturbation weights to different types of intralayer links and then selects a predetermined proportion of intralayer links to remove according to the weights. In contrast, if these information cannot be obtained at one time, we design a biased random walk procedure, local perturbation strategy, to execute perturbation. Four kinds of interlayer link prediction algorithms are carried out on different real-world and artificial perturbed multiplex networks. We find out that the intralayer links connected with small degree nodes have the most significant impact on the prediction accuracy. The intralayer links connected with large degree nodes may have side effects on the interlayer link prediction.

physics.soc-ph↗

arXiv2022

ClueGraphSum: Let Key Clues Guide the Cross-Lingual Abstractive Summarization

Cross-Lingual Summarization (CLS) is the task to generate a summary in one language for an article in a different language. Previous studies on CLS mainly take pipeline methods or train the end-to-end model using the translated parallel data. However, the quality of generated cross-lingual summaries needs more further efforts to improve, and the model performance has never been evaluated on the hand-written CLS dataset. Therefore, we first propose a clue-guided cross-lingual abstractive summarization method to improve the quality of cross-lingual summaries, and then construct a novel hand-written CLS dataset for evaluation. Specifically, we extract keywords, named entities, etc. of the input article as key clues for summarization and then design a clue-guided algorithm to transform an article into a graph with less noisy sentences. One Graph encoder is built to learn sentence semantics and article structures and one Clue encoder is built to encode and translate key clues, ensuring the information of important parts are reserved in the generated summary. These two encoders are connected by one decoder to directly learn cross-lingual semantics. Experimental results show that our method has stronger robustness for longer inputs and substantially improves the performance over the strong baseline, achieving an improvement of 8.55 ROUGE-1 (English-to-Chinese summarization) and 2.13 MoverScore (Chinese-to-English summarization) scores over the existing SOTA.

cs.CL↗

arXiv2021

Interlayer Link Prediction in Multiplex Social Networks Based on Multiple Types of Consistency between Embedding Vectors

Online users are typically active on multiple social media networks (SMNs), which constitute a multiplex social network. It is becoming increasingly challenging to determine whether given accounts on different SMNs belong to the same user; this can be expressed as an interlayer link prediction problem in a multiplex network. To address the challenge of predicting interlayer links , feature or structure information is leveraged. Existing methods that use network embedding techniques to address this problem focus on learning a mapping function to unify all nodes into a common latent representation space for prediction; positional relationships between unmatched nodes and their common matched neighbors (CMNs) are not utilized. Furthermore, the layers are often modeled as unweighted graphs, ignoring the strengths of the relationships between nodes. To address these limitations, we propose a framework based on multiple types of consistency between embedding vectors (MulCEV). In MulCEV, the traditional embedding-based method is applied to obtain the degree of consistency between the vectors representing the unmatched nodes, and a proposed distance consistency index based on the positions of nodes in each latent space provides additional clues for prediction. By associating these two types of consistency, the effective information in the latent spaces is fully utilized. Additionally, MulCEV models the layers as weighted graphs to obtain representation. In this way, the higher the strength of the relationship between nodes, the more similar their embedding vectors in the latent representation space will be. The results of our experiments on several real-world datasets demonstrate that the proposed MulCEV framework markedly outperforms current embedding-based methods, especially when the number of training iterations is small.

physics.soc-ph↗

arXiv2020

Interlayer link prediction in multiplex social networks: an iterative degree penalty algorithm

Online social network (OSN) applications provide different experiences; for example, posting a short text on Twitter and sharing photographs on Instagram. Multiple OSNs constitute a multiplex network. For privacy protection and usage purposes, accounts belonging to the same user in different OSNs may have different usernames, photographs, and introductions. Interlayer link prediction in multiplex network aims at identifying whether the accounts in different OSNs belong to the same person, which can aid in tasks including cybercriminal behavior modeling and customer interest analysis. Many real-world OSNs exhibit a scale-free degree distribution; thus, neighbors with different degrees may exert different influences on the node matching degrees across different OSNs. We developed an iterative degree penalty (IDP) algorithm for interlayer link prediction in the multiplex network. First, we proposed a degree penalty principle that assigns a greater weight to a common matched neighbor with fewer connections. Second, we applied node adjacency matrix multiplication for efficiently obtaining the matching degree of all unmatched node pairs. Thereafter, we used the approved maximum value method to obtain the interlayer link prediction results from the matching degree matrix. Finally, the prediction results were inserted into the priori interlayer node pair set and the above processes were performed iteratively until all unmatched nodes in one layer were matched or all matching degrees of the unmatched node pairs were equal to 0. Experiments demonstrated that our advanced IDP algorithm significantly outperforms current network structure-based methods when the multiplex network average degree and node overlapping rate are low.

physics.soc-ph↗