This book constitutes the refereed proceedings of the 13th CCF Conference on BigData 2025, which took place in Tianjin, China, during September 1214, 2025.

The 25 full papers presented in this volume were carefully reviewed and selected from 196 submissions. The topics of accepted papers include theories and methods of data science, algorithms and applications of big data.

Inhalt

.- SlideDP: A Time-Aware and Private Sketch for Cardinality Estimation over Sliding Windows.
.- PreSL: Document Level Pre-training of Scientific Literature via Self-supervised Learning.
.- EACon: Finding Key Information for Claim Verification.
.- An AI-Powered System for Predicting Energy Consumption in Ore Comminution Operations.
.- PMTS: Parameter Masking meets Temporal Smoothness in Backdoor Defense for Federated Learning.
.- AFLA-V2VDA: A Virtual-real Domain Adaptation Method for V2V Collaborative Object Detection.
.- DLSFNet: A Modular Architecture for Long-Term Time Series Forecasting via Structural Decoupling and Multi-Scale Feature Fusion.
.- A Gated Attention Transformer with Dynamic Spatio-Temporal Decoupling for Traffic Flow Prediction.
.- HMSC-FuseMaskNet: A Hybrid Multi-Source Neural Architecture for Context-Aware Data Security Classification.
.- Dynamic Pattern-Aware Privacy Protection for Data Streams via Local Differential Privacy.
.- Heterogeneous Graph Neural Network Based on Self-Knowledge Distillation and Transformer.
.- Causality-Informed Machine Learning for ICU Length of Stay Prediction: A Healthcare Case Study.
.- LTE-DETR: Long-Tail Scenes Object Detection via Enhanced Feature Representation.
.- Dual-branch Transformer with Temporal-Feature Fusion for Multivariate Time Series Imputation.
.- Heterogeneous Graph Neural Network Based on Self-Knowledge Distillation and Transformer.
.- PCNNSR: Facial Image Super-Resolution via Pulse-Coupled Neural Network Attention Mechanism.
.- Retrieval-Augmented Generation for News Recommendation.
.- Semi-Supervised Classification Based on Abductive Learning.
.- Shortest Path Tree Maximize Node Influence Propagation in Complex Networks.
.- BSGCL: Boosting Scene Graph Completeness and Localization under Weak Supervision.
.- Multi-Modal LLM based Image Captioning in ICT: Bridging the Gap Between General and Industry Domain.
.- PNRM: A Pre-training News Recommendation Model with Contrastive Sampling.
.- M2STFormer: A Mixture-of-Experts Multi-Level Spatio-Temporal Transformer for Traffic Flow Prediction.
.- QAugment-Summary: Multi-Stage Summarization and Question Decomposition for Open-Domain QA.
.- M3Bench: A Benchmark for Multi-Model Databases on Analytical Query and In-Database Machine Learning Workloads.