The crowdsourcing data essentially refers to a large amount, complex and potential spam in the Internet. The current central issue of spam research is how to use effective methods to mine and use the data of interest. This article introduced the latest knowledge graph technology in the computer field to extracts and stores knowledge of all crowdsourcing data. The development of knowledge graph mainly aims to solve two core problems. One is modeling knowledge and the other is how to store the model. This paper discussed the latter point. First , we studied the data storage model of all source data knowledge and analyzed the methods for the construction of storage models in artificial intelligence by using crowdsourcing and attribute knowledge. Then we focuses on a plurality of methods for knowledge storage of crowdsourcing data. There are currently more popular methods such as storing triples, storing horizontally, attributing store, dividing vertically, multiple query and mixing management. We analyzed the advantages and disadvantages of these methods and their performance in specific application scenarios. Finally , We took the current stable Neo4j knowledge database as an example to stores and visualizes the mined crowdsourcing data. The results showed that the current graph database can store and the operate the performance reliably.