Abstract: Geological literature includes information of genesis, process and prediction of ore deposits.It is important to extract and summarize the characteristics of ore deposits from structured and unstructured data.Therefore, this paper summarizes the feature model of the tungsten ore knowledge base, and constructs the tungsten mineral knowledge base based on the potential evaluation data and literature data.Based on python's string fuzzy matching algorithm, data classification and similarity calculation are realized by using tungsten ore knowledge base as data source.The results show that the algorithm can identify and distinguish different prediction types.