Abstract: Based on big data method, the text of natural language user requirement is analyzed. TF-IDF method is used to calculate the frequency of feature words appearing in the requirement text. After normalization, the weight of feature words in the text is obtained. Chi-square statistics is used to measure the correlation between feature words and clusters. The feature set is clustered by k-means algorithm and bisecting k-means algorithm, and the user requirement text feature tree is generated, which is mapped to the corresponding requirement function in the requirement function decomposition tree. The semantics and structure of requirement feature are clearer. The ability of automatic requirement specification is enhanced. The efficiency of requirement analysis is improved, and the requirement specification is more scientific, reasonable and accurate.