Concept localization using n-gram Information Retrieval model and Control Flow Graph

Developing software involves many phases such as designing, coding and testing. Once the software is released, a separate team is responsible for maintaining the software. Nowadays many researchers and users work on Open Source Software to enhance its functionalities and to mould it according to their needs. Most of the time, a user or developer wants to locate a specific feature in software for the purpose of enhancement or removing a fault, which is known as concept localization. Automatic concept localization gives relevant files to the users as per the requirement. We have implemented n-gram, an Information Retrieval model to retrieve the names of the relevant files from the source code and incorporated Control Flow Graph (CFG) which helped us to determine the files encapsulating the functionality, in the correct order. We conducted tests on numerous grounds such as different threshold values (0.4, 0.6 and 0.8), N value (2 and 3) and varying query length. On examination, we obtained recall of 74% and precision of 65% on threshold value of 0.6 using trigram (i.e. n=3). Control Flow Graph significantly contributed in improving the ranking of relevant files.