Focusing synonymy on source code repositories

Del Carpio P.M. · RISTI - Revista Iberica de Sistemas e Tecnologias de Informacao (2020)

Resumen

We expect that a huge corpus of code could be rich in patterns, and the corpus of software has statistical properties, which are similar to the corpus of natural language. In spite of code and text are similar, written code is a new problem domain for the techniques of natural language processing. Taking in account that synonymy is the base for building a WordNet, this work addresses the synonymy in source code repositories, presenting an approach and techniques based on naming patterns and term frequency for that. The proposal has been evaluated in popular repositories of Apache, Eclipse and Red Hat. The results show that the proposal does not require an intensive processing.

Instituciones: Universidad La Salle