Statistical Approach to Increase Source Code Completion Accuracyстатья
Информация о цитировании статьи получена из
Scopus
Статья опубликована в журнале из списка Web of Science и/или Scopus
Дата последнего поиска статьи во внешних источниках: 2 января 2019 г.
Аннотация:Code completion is an essential feature in every IDE’s toolbox, boosting a developer’s productivity and significantly reducing time spent on code exploration. In this paper, we introduce the extension of a typical code completion system. At each point, we construct a list of all possible functions, which are then sorted according to our probabilistic model. We draw our inspiration from natural language processing (NLP). As the foundation, we select the N-gram model, which works on top of abstract syntax tree (AST) nodes. Since our approach is not bound to any other analyses, our model is language-agnostic, and thus, can be applied to any programming language. Experiments on several well-known open source projects show that the described method is sound. It has an execution time comparable to naïve approaches and achieves much more accurate results.