arXiv subjects

Leihan Zhang

Publications and source records attributed to Leihan Zhang.

5 recordsLinked to original sources

arXiv2026

RiskNet: A large-scale dataset of AI risk incidents from news with alignment and multi-dimensional annotations

As artificial intelligence (AI) systems are increasingly deployed across socially consequential domains, reports of AI-related harms and failures have grown in frequency and diversity. Although existing governance frameworks articulate high-level principles for responsible AI, large-scale empirical resources for tracking and analyzing real-world AI risk incidents remain limited. Existing incident collections are often manually curated, relatively small in scale, and insufficient for continuous, data-driven monitoring and downstream computational analysis. To address this need, we present RiskNet, a large-scale dataset of AI risk incidents constructed from large-scale multilingual news sources. RiskNet applies a structured pipeline for AI risk news identification, event-level report screening, incident alignment, and multi-dimensional incident classification. The resulting resource organizes dispersed news reports into incident-centered records and provides benchmark datasets for event classification, incident alignment, and incident-level risk labeling. In its current release, RiskNet covers hundreds of millions of source records and yields a large-scale collection of AI risk-related reports, including aligned incident clusters and annotated benchmark subsets. The dataset is also accessible through an online platform for browsing and exploration. We describe the data sources, processing workflow, taxonomy design, and technical validation of the resource. RiskNet is intended to support downstream research on AI safety, governance, risk analysis, and benchmarking, as well as longitudinal and cross-source analyses of AI-related harms. By providing a structured and reusable empirical resource, RiskNet helps bridge the gap between high-level governance principles and the documented realities of AI risk incidents.

cs.LG↗

arXiv2026

Legal-DC: Benchmarking Retrieval-Augmented Generation for Legal Documents

Retrieval-Augmented Generation (RAG) has emerged as a promising technology for legal document consultation, yet its application in Chinese legal scenarios faces two key limitations: existing benchmarks lack specialized support for joint retriever-generator evaluation, and mainstream RAG systems often fail to accommodate the structured nature of legal provisions. To address these gaps, this study advances two core contributions: First, we constructed the Legal-DC benchmark dataset, comprising 480 legal documents (covering areas such as market regulation and contract management) and 2,475 refined question-answer pairs, each annotated with clause-level references, filling the gap for specialized evaluation resources in Chinese legal RAG. Second, we propose the LegRAG framework, which integrates legal adaptive indexing (clause-boundary segmentation) with a dual-path self-reflection mechanism to ensure clause integrity while enhancing answer accuracy. Third, we introduce automated evaluation methods for large language models to meet the high-reliability demands of legal retrieval scenarios. LegRAG outperforms existing state-of-the-art methods by 1.3% to 5.6% across key evaluation metrics. This research provides a specialized benchmark, practical framework, and empirical insights to advance the development of Chinese legal RAG systems. Our code and data are available at https://github.com/legal-dc/Legal-DC.

cs.CL↗

arXiv2026

KID: Knowledge-Injected Dual-Head Learning for Knowledge-Grounded Harmful Meme Detection

Internet memes have become pervasive carriers of digital culture on social platforms. However, their heavy reliance on metaphors and sociocultural context also makes them subtle vehicles for harmful content, posing significant challenges for automated content moderation. Existing approaches primarily focus on intra-modal and inter-modal signal analysis, while the understanding of implicit toxicity often depends on background knowledge that is not explicitly present in the meme itself. To address this challenge, we propose KID, a Knowledge-Injected Dual-Head Learning framework for knowledge-grounded harmful meme detection. KID adopts a label-constrained distillation paradigm to decompose complex meme understanding into structured reasoning chains that explicitly link visual evidence, background knowledge, and classification labels. These chains guide the learning process by grounding external knowledge in meme-specific contexts. In addition, KID employs a dual-head architecture that jointly optimizes semantic generation and classification objectives, enabling aligned linguistic reasoning while maintaining stable decision boundaries. Extensive experiments on five multilingual datasets spanning English, Chinese, and low-resource Bengali demonstrate that KID achieves SOTA performance on both binary and multi-label harmful meme detection tasks, improving over previous best methods by 2.1%--19.7% across primary evaluation metrics. Ablation studies further confirm the effectiveness of knowledge injection and dual-head joint learning, highlighting their complementary contributions to robust and generalizable meme understanding. The code and data are available at https://github.com/PotatoDog1669/KID.

cs.CL↗

arXiv2016

Sleeping Beauties in Meme Diffusion

A sleeping beauty in diffusion indicates that the information, can be ideas or innovations, will experience a hibernation before a sudden spike of popularity and it is widely found in citation history of scientific publications. However, in this study, we demonstrate that the sleeping beauty is an interesting and unexceptional phenomenon in information diffusion and even more inspiring, there exist two consecutive sleeping beauties in the entire lifetime of propagation, suggesting that the information, including scientific topics, search queries or Wikipedia entries, which we call memes, will go unnoticed for a period and suddenly attracts some attention, and then it falls asleep again and later wakes up with another unexpected popularity peak. Further explorations on this phenomenon show that intervals between two wake ups follow an exponential distribution and the second wake up generally reaches its peak at a higher velocity. In addition, higher volume of the first wake up will lead to even much higher popularity of the second wake up with great odds. Taking these findings into consideration, an upgraded Bass model is presented to well describe the diffusion dynamics of memes on different media. Our results can help understand the common mechanism behind propagation of different memes and are instructive to locate the tipping point in marketing or find innovative publications in science.

cs.SI↗

arXiv2014

Who creates trends in online social media: The crowd or opinion leaders?

Trends in online social media always reflect the collective attention of a vast number of individuals across the network. For example, Internet slang words can be ubiquitous because of social memes and online contagions in an extremely short period. From Weibo, a Twitter-like service in China, we find that the adoption of popular Internet slang words experiences two peaks in its temporal evolution, in which the former is relatively much lower than the latter. This interesting phenomenon in fact provides a decent window to disclose essential factors that drive the massive diffusion underlying trends in online social media. Specifically, the in-depth comparison between diffusions represented by different peaks suggests that more attention from the crowd at early stage of the propagation produces large-scale coverage, while the dominant participation of opinion leaders at the early stage just leads to popularity of small scope. Our results quantificationally challenge the conventional hypothesis of influentials. And the implications of these novel findings for marketing practice and influence maximization in social networks are also discussed.

cs.SI↗