[Purpose/Significance] By jointly mining multiple diaries related to National South-west Associated University (NSAU), a social network graph of NSAU that integrates information from multiple sources is constructed. The aim is to discover more potential social relationships through joint mining of multiple diaries, and break through the limitations of single diary social network mining. [Method/Process] Using multiple diaries related to NSAU from 1938 to 1941 as corpus, Python program is used to count co-occurrence relationships of characters, and Gephi is used to construct multi-diary social network graph. Through social network analysis methods, the network topology features, character centrality features and character group features based on modularity and K-core are analyzed and discussed. [Result/Conclusion] Compared with independent diary mining, multi-diary social network joint mining showed more obvious network structure features, more decentralized and rich social relationship information, which can reveal more hidden social relationships, and has good application value in the field of digital humanities.