Expert Example Acquisition Based On Stack Overflow Ontology Modeling And Its Application In API Mining

Application programming interface(API)is a set of accessible interfaces provided by the software library.The software library provides services through the API.Developers usually use the relevant functions provided by the software library by calling the API to carry out software development.Software developers use APIs to achieve software reuse,which also greatly improves the efficiency of their software development.Now,the complexity and scale of software are increasing,various large and complex software is emerging in an endless stream,and software reuse technology is also developing.Therefore,the types and number of APIs that developers need to use in the software development process are increasing.The use methods of APIs held by developers themselves are very limited.For some APIs that are not familiar with how to use,they need to consult relevant API reference documents.However,some APIs lack relevant reference documents or have reference documents.However,most of the reference documents are mainly text explanations and lack corresponding code examples.Therefore,developers are hindered in learning to use APIs.To help developers learn to use the API faster,reduce the time spent by developers in learning to use the API,and improve the development efficiency of developers,researchers began to mine the API call patterns from a large number of open source codes according to certain rules by analyzing open source code examples.These mined API call patterns can solve the problem of insufficient API reference document code examples to a certain extent,that is,help developers quickly and effectively query,understand,and use the API.Our research team also conducted relevant research on the mining methods of API call patterns before this and proposed an example-based API call pattern mining method,BE-Miner.This method extracts the API call sequences from a large number of code examples in the Stack Overflow Q&A website.Through these extracted API call sequences,we can mine the API call patterns with fewer occurrences and improve the recall rate of the mining results.However,in terms of code examples,this method does not take into account the quality of code examples obtained from the Stack Overflow Q&A website.Some code examples with poor quality do not help the mining of API call patterns,and may even adversely affect the mining effect.To solve the problem of selecting code samples for the example-based API call pattern mining method,this paper proposes an expert example acquisition method based on Stack Overflow ontology modeling,builds an ontology model from the Stack Overflow Q&A website,applies the ontology model to the Jena reasoning machine,customizes the Jena reasoning rules,and collects and obtains some metric reasoning rules from some relevant research documents of Stack Overflow,The expert example reasoning rules are obtained by combining the rules,and expert users and expert examples are deduced.Finally,this paper uses the obtained expert examples as training sets in the BE-Miner method.The experimental results show that the high-quality code examples written by expert users,namely expert examples,can effectively improve the mining effect of the BE-Miner method.