Influential Global and Local Contexts Guided Trace Representation for Fault Localization

• DOI: 10.1145/3576043
• Publication Date: 2022-12-15T14:51:53Z
• Authors: Zhuo Zhang; Yan Lei; Ting Su; Meng Yan; Xiaoguang Mao; Yue Yu
• Keywords: TRACE (psycholinguistics); Test suite; Representation

Trace data is critical for fault localization (FL) to analyze suspicious statements potentially responsible a failure. However, existing trace representation meets its bottleneck mainly in two aspects: (1) the information of statement restricted local context (i.e., test case) without consideration global all cases suite); (2) it just uses ‘occurrence’ strong FL semantics. Thus, we propose UNITE : an infl U ential co N text-Gu I ded T race r E presentation, representing from both and contexts with influential semantics FL. embodies implements key ideas: leverages widely used weighting capability retrieval reflect how important (a word) case document) suite collection), where collection) represent respectively; further elaborates (weak semantics) ‘influence’ (strong by combing program dependencies. The large-scale experiments on 12 techniques 20 programs show that significantly improves effectiveness.