Code search query expansion method based on large language model

Abstract

The invention discloses a code search query expansion method based on a large language model, and belongs to the related technical field of code search query expansion, the method comprises the following steps: obtaining an original query input by a user; inputting the original query into a query expansion model subjected to corpus fine tuning in the code search field, and performing semantic expansion on keywords in the original query through the query expansion model to generate an expanded query; and inputting the extended query into a neural network code search model, and retrieving the matched code snippets from the code library by calculating the similarity between the query and the code snippets in the vector space. The method has positive influence on query expansion in a code search task, and effectively enhances the query expansion, so that the code search accuracy is improved.

Images