For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
Spectrum-Based Fault Localization Using Fault Triggering Model to Refine Fault Ranking List
Yong WANG Zhiqiu HUANG Rongcun WANG Qiao YU
IEICE TRANSACTIONS on Information and Systems
Publication Date: 2018/10/01
Online ISSN: 1745-1361
Type of Manuscript: PAPER
Category: Software Engineering
fault localization, software debugging, testing,
Full Text: PDF(824.9KB)
>>Buy this Article
Spectrum-based fault localization (SFL) is a lightweight approach, which aims at helping debuggers to identity root causes of failures by measuring suspiciousness for each program component being a fault, and generate a hypothetical fault ranking list. Although SFL techniques have been shown to be effective, the fault component in a buggy program cannot always be ranked at the top due to its complex fault triggering models. However, it is extremely difficult to model the complex triggering models for all buggy programs. To solve this issue, we propose two simple fault triggering models (RIPRα and RIPRβ), and a refinement technique to improve fault absolute ranking based on the two fault triggering models, through ruling out some higher ranked components according to its fault triggering model. Intuitively, our approach is effective if a fault component was ranked within top k in the two fault ranking lists outputted by the two fault localization strategies. Experimental results show that our approach can significantly improve the fault absolute ranking in the three cases.