Feature and variability extraction from natural language software requirements specifications

Abstract

Extracting feature and variability from requirement specifications is an indispensable activity to support systematic integration related single software systems into Software Product Line (SPL). Performing variability extraction is time-consuming and inefficient, since massive textual requirements need to be analyzed and classified. Despite the improvement of automatically features and relationships extraction techniques, existing approaches are not able to provide high accuracy and applicability in real-world scenarios. The aim of my doctoral research is to develop an automated technique for extracting features and variability which provides reliable solutions to simplify the work of domain analysis. I carefully analyzed the state of the art and identified main limitations so far: accuracy and automation. Based on these insights, I am developing a methodology to address this challenges by making use of advanced Natural Language Processing (NLP) and machine learning techniques. In addition, I plan to design reasonable case study to evaluate the proposed approaches and empirical study to investigate usability in practice.