API recommendation method based on word embedding technology

Abstract

The invention provides an API (Application Program Interface) recommendation method based on a word embedding technology, which recommends an appropriate API to developers, belongs to the technical field of code recommendation, and comprises the following steps: step 1, collecting a Java help document and a question and answer data set of a Java language on Stack Overflow; step 2, word embedding modeling is carried out through the collected data in the Stack Overflow; 3, collecting questions input by a developer through a UI interface, calculating the similarity with the target question set byusing a word model and a TF-IDF, and filtering a candidate question list by using an LSA model; and step 4, extracting subject terms through LDA, and finally obtaining API recommendation informationfrom the Java help document, so that a developer can recommend relatively accurate API information to the developer when using the method to perform API query.

Description

technical field [0001] The invention belongs to the technical field of code recommendation, in particular to an API recommendation method based on word embedding technology. Background technique [0002] With the continuous development of technology, people have increasingly rich functional requirements for software, and the structure of software presents the characteristics of complexity and large-scale. In the process of software development, program developers may face many difficult situations in software writing, such as forgetting the function of API (Application Programming Interface, application programming interface), not familiar with the return value and parameter list of an API, or wanting to know Specific implementation methods for some uncommon functions. Therefore, it is particularly important for developers to query APIs that meet development needs during the development process. [0003] When a suitable API cannot be found in the actual development process...

Images