近年来，社会网络快速发展，产生海量且杂乱的数据。如何建模社会网络并推理社会知识是一个关键问题。现有研究仅关注网络演化过程的一个静态截面，忽略了各要素和关系在时间维度上的关联，导致网络演化模型缺失、知识推理精度不足。针对该问题，本项目拟从知识演化角度出发，围绕“机理发现-模型构建-知识推理”主线，研究异质社会网络演化建模和知识推理问题：1）机理发现：基于真实异质社会网络数据，观察分析网络演化规律，对网络演化机理进行发现总结，为后续建模及推理提供原理性支撑；2）模型构建：基于异质社会网络演化机理，采用随机过程与概率论等工具，对网络演化过程进行数学建模；3）知识推理：从知识演化角度出发，采用图神经网络等深度学习工具，挖掘已有知识演化模式，推理隐藏社会知识。上述三个方面层层递进，对异质社会网络中的演化机制和知识推理形成有益探索，研究成果在舆情监控、突发事件分析等国家重大需求中具有重要应用价值。

In recent years, social networks develop rapidly and generate the chaotic and massive data. Evolution modeling and knowledge reasoning in heterogeneous social networks has become an important problem. Existing studies focused on a snapshot of network evolution, but neglected the relevance of social entities and relationships in time dimension, which led to lack of evolution modeling and inaccuracy of knowledge reasoning. To solve this problem, this project aims to study evolution modeling and knowledge reasoning in heterogeneous social networks from the knowledge evolution perspective, following three parts: mechanism uncovering, network modeling and knowledge reasoning. 1) mechanism discovering: offer theoretical support to evolution modeling and knowledge reasoning by observing and summarizing the pattern of network evolution based on real data; 2) network modeling: based on the mechanism, mathematically model the network evolution using probability theory, stochastic process， etc.; 3) knowledge reasoning: using deep learning methods like graph neural networks, analyze the evolution mode of existing knowledge and discover the hidden knowledge from the perspective of knowledge evolution. The above three parts provide benefits to the study of evolution modeling and knowledge reasoning in heterogeneous social networks, which contributes to national development like public opinion monitoring and emergency analysis.