Large Scale Evaluation of Natural Language Processing Based Test-to-Code Traceability Approaches

Traceability information can be crucial for software maintenance, testing, automatic program repair, and various other engineering tasks. Customarily, a vast amount of test code is created systems to maintain improve quality. Today's may contain tens thousands tests. Finding the parts tested by each case usually difficult time-consuming task without help authors tests or at least clear naming conventions. Recent test-to-code traceability research has employed approaches but textual methods as standalone techniques were investigated only marginally. The convention approach well-regarded method among developers. Besides their often voluntary use, however, one its main weaknesses that it identify one-to-one links. With use more versatile text-based methods, candidates could ranked similarity, thus producing number possible connections. Textual also have disadvantages, even machine learning provide semantically connected links from text itself, these refined with incorporation structural information. In this paper, we investigate applicability three both link recovery technique regarding combination possibilities paper presents an extensive evaluation using several source representations meta-parameter settings on eight real, medium-sized combined size over 1.25 million lines code. Our results suggest suitable settings, used purposes, where conventions not followed.