Automating Key Phrase Extraction from Fault Logs to Support Post-Inspection Repair of Software Requirements

Abstract

This research paper aims at developing an automated approach to identify fault prone requirements in a software requirement specification (SRS) document to mitigate the fault propagation to later phases where the same faults are harder to find and fix. This research work proposes an automated approach (i.e., KESRI) for the identification of “problematic areas” (i.e., faulty requirements) from fault logs generated during inspections. Our automated approach uses machine learning-based key phrase extraction (KPE) algorithms (both supervised and unsupervised) that can extract key phrases from fault logs and map them to an SRS document (using semantic analysis) to locate faulty requirements. To validate our proposed approach, an inspection study conducted at North Dakota State University (NDSU) with 41 inspectors using an industrial-strength SRS document that resulted in fault logs. When compared against human experts, our approach achieved F-measure of up to 83% in extracting the relevant key phrases using supervised KPE algorithms. In conclusion, our automated KPE and mapping approach has the potential to reduce manual overhead and assist authors during the fault-fixation post-inspection.