Abstract-Requirements are commonly vague and ambiguous. In this paper, we describe an automated Inconsistency Checker called Marama AI for checking for high-level inconsistency between textual requirements, abstract interactions and Essential Use Cases. We use concepts of phrase extraction and essential interaction patterns to carry out these checks. We provide further support for checking of requirements quality attributes such as completeness and correctness using visual differencing.
Keywords-higher level inconsistency, essential interaction pattern, visual differencing
I.INTRODUCTION It is acknowledged by many that natural language requirements are very often error prone, imprecise and ambiguous [1], [2]. In order to deal with these issues much research has been aimed at checking of requirements consistency, completeness and correctness either by using heuristic algorithms and formal models [3], [4] or semiformal models [5], [6]. In our previous work [7,8], we have introduced the MaramaAI tool for capturing requirements from natural language descriptions and helping to manage requirements inconsistency. The tool captures the essential requirements in the form of abstract interactions from natural language requirements and then transforms it to a semi-formal representation called Essential Use Cases. Besides capturing requirements, it also triggers inconsistency warnings if any inconsistency occurs between the textual requirement, abstract interaction and Essential Use Cases. However, triggering simple inconsistency warnings between these requirement elements is not adequate to make sure the requirements are completely consistent.