The two-volume set CCIS 2910 + 2911 constitutes the refereed proceedings of the 20th CCF Conference on Computer Supported Cooperative Work and Social Computing, ChineseCSCW 2025, held in Guangzhou, China, in November 2025.

The 50 full papers and 13 short papers presented in the proceedings were carefully reviewed and selected from 196 submissions.These papers were organized in topical sections as follows:

Part I: CSCW and Social Computing and Human-Machine-Things Fusion and Human-AI Collaborative Computing.

Part II: Crowd Cooperative Computing and Evolutionary Computation and Collaborative Data, Software and Services.

Inhalt

.- Topic A: CSCW and Social Computing
.- Feature-aware Variational Graph Autoencoder for Link Prediction.
.- Exploring the Impact of AI Technology on Doctor-Patient Relationship Based on Topic Modeling and Sentiment Analysis.
.- HyperCLKT: Enhancing Knowledge Tracing via Hypergraph-Contrastive Learning with Self-Attention.
.- Multi-layer Heterophilic Graph Representation Learning with Relation Consistency Preservation.
.- A Multi-Dimensional Analysis of Academic Social Networks: A Case Study on SCHOLAT.
.- Graph Attention Based on Kolmogorov Arnold Network for Social Recommendation.
.- A Generalized Debiasing Framework to Remove Mixing Bias in Recommender Systems.
.- Straggler-Aware Dynamic Selection for Federated Learning.
.- Membership Uncertainty-Aware Overlapping Community Detection with Adaptive Cluster Number Estimation.
.- A Study on Users' Privacy Risks and Disclosure Behaviors in Generative Artificial Intelligence Platforms.
.- A Large Language Model for Traditional Chinese Medicine Prescription Recommendation Based on Multi-Stage Learning.
.- Fusing Hierarchical Attention and Hawkes Processes for Embedding Dynamic Heterogeneous Graphs.
.- Detecting Collusion in Play-to-Earn Games: A Case Study of Axie Infinity.
.- A Graph Neural Network Based on Feature Network for Identifying Influential Nodes.
.- Enhanced BERTopic Topic Modeling with LLM-Based Text Preprocessing.
.- Contrastive Enhanced Multi-View Session Recommendation Model With Collaborative User Interest and Intention.
.- FCA-based Maximal Clique Summary Detection in Social Networks.
.- Sampling-Driven Multi-State Epidemic Model.
.- "AgentSEIV: Rumor Diffusion Simulation and Dynamic Intervention in Social Networks".
.- Chinese Relation Extraction based on Data Augmentation and Dual-Channel Attention.
.- Research on Retrieval-Augmented Generation Technology Utilizing Chain-of-Thought and Hybrid Search.
.- A Cost-Effective Training Method for Large Models in Pedestrian Search Task.
.- Evidence-Chain-Based Analysis for the Factual Authenticity of Social Media Report.
.- Topic B: Human-Machine-Things Fusion and Human-AI Collaborative Computing
.- Spectral Clustering-driven SimpleMKKM Algorithm via Multi-level Aggregation.
.- Interval Type-2 Symmetrical Quintuple Implicational Algorithm for the Multi-rule Case.
.- Research on the LLMs Trust Mechanism and Users' Active Use IntentionThe Role of Usage Experience and Model Usage Richness.
.- TriAD: Tri-Level Augmentation Distillation for Uncertainty-Aware Consistency in Semi-Supervised Semantic Segmentation.
.- PokéElder: Building a MLLM-Based Interactive Tool to Guide Older Adults in Learning APP Usage.
.- ScholarRAG: Path-Aware Knowledge Graph-Augmented Retrieval for Scholarly QA.
.- Defense against Prompt Injection Attacks via Intention Emphasis.
.- Deep Learning for Ship Re-identification: A Survey and Outlook.
.- MTS-STNet: A Spatio-Temporal Fusion Model for Traffic Flow Forecasting Considering Multivariate Relationships.
.- TLLight: A Two-layer Reinforcement Learning Approach based Intelligent Traffic Signal Control System for Emergency Vehicle Priority.
.- A KG-LLM Synergistic Framework for Endless Learning in Chinese Classical Studies.
.- Enhanced Professional Semantic Similarity Measurement: Leveraging LLMs and Human Experience.