Premium
Feature location enhancement based on source code augmentation with synonyms of terms
Author(s) -
Saifan Ahmad A.,
Obeidat Lana
Publication year - 2021
Publication title -
software: practice and experience
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.437
H-Index - 70
eISSN - 1097-024X
pISSN - 0038-0644
DOI - 10.1002/spe.2900
Subject(s) - source code , computer science , feature (linguistics) , code (set theory) , information retrieval , search engine indexing , precision and recall , process (computing) , data mining , kpi driven code analysis , wordnet , natural language processing , artificial intelligence , software , programming language , static program analysis , software development , set (abstract data type) , philosophy , linguistics
Summary In software maintenance the developers may add new feature to program, improve existing function, and remove bugs. In this case the developer should identify the location in the source code that corresponds to a specific functionality; this is known as feature location. This presents a new approach for enhancing the process of feature location using Information Retrieval (IR) and Natural Language Processing. The approach presented augments the source code with additional semantic information that was extracted and derived from the synonyms of source code terms. This approach works in a pipeline structure, starting by augmenting the source code corpus with synonyms of the original terms and ending by inferring the source code with a particular user query. More specifically, the WordNet platform is used for extracting the synonyms of terms. Moreover, the approach uses an advanced IR technique, namely the Latent Semantic Indexing, for searching and inferring the source code. The used approach was tested and evaluated on two open source systems, namely the Qt and Hippodraw. Four experiments were conducted on each system using 21 features and the results showed that enriching the source code with synonyms of terms clearly and significantly improved the process of feature location efficiently. The experimental results showed that the approach presented obtained higher levels of Recall and Precision. For that reason, it has been shown to improve the state‐of‐the‐art techniques for feature location process.