With the advent of the era of big data, people are increasingly demanding for queries of mass cultural and sports activities, which have been given high attention. However, due to the challenges of such activities and the outstanding performance of knowledge graphs in recent years, in this paper we solve the problem from the perspective of knowledge graph. We first present a method to construct a knowledge graph of mass cultural and sports activities based on the fusion of multisource data for user activity query, then present a hierarchical modeling method combining ontology and taxonomy, design a BERT-based pipeline model to extract activity entities, solve the extraction of unstructured text data, and use a rule-based method to extract the relationship between activities and labels. Finally, the knowledge graph is constructed using the real activity data of Beijing region, and the multi-condition query of the activity is realized. The experimental results show that this method can effectively provide personalized activity query for users with different needs, and solve the problem of wide activity information and low matching between activities and users. A detailed experimental analysis was carried out to construct a data set using real activity data in Beijing, and the validity of the experiment was evaluated.