Using Semantic Analysis and Graph Mining Approaches to Support Software Fault Fixation

Software requirement specification (SRS) documents are written in natural language (NL) and are prone to contain faults due to the inherently ambiguous nature of NL. Inspections are employed to find and fix these faults during the early phases of development, where these are the most cost-effective to fix. Inspections being too manual are very tedious and time consuming to perform. After fixing a fault, the SRS author has to manually re-inspect the document to make sure if there are other similar requirements that need a fix, and also if fixing a fault does not reintroduce another fault in the document (i.e., change impact analysis). The proposed approach in this paper employs NL processing, machine learning, semantic analysis, and graph mining approaches to generate a graph of inter-related requirements (IRR) based on semantic similarity score. The IRR graph is next mined using graph mining approaches to analyze the impact of a change. Our approach when applied using a real SRS generated IRR and yielded promising results. Graph mining approaches resulted in a G-mean of more than 90% to accurately identify the highly similar requirements to support the CIA.