Nocuous Ambiguities in Requirements Specifications

Chantree, Francis; Nuseibeh, Bashar; De Roeck, Anne and Willis, Alistair (2005). Nocuous Ambiguities in Requirements Specifications. Technical Report 2005/03; Department of Computing, The Open University.

DOI: https://doi.org/10.21954/ou.ro.00016035

Abstract

In this paper we present a novel approach that automatically alerts authors of requirements specifications to the presence of potentially dangerous ambiguities in their text. We first establish the notion of "nocuous" ambiguities, i.e. those that are likely to lead to misunderstandings. We focus on coordination ambiguity, which occurs when words such as "and" and "or" are used. Our starting point is a dataset of ambiguous phrases from a corpus of requirements specifications, and a collection of associated human judgements about their interpretation. We then use machine learning techniques combined with syntactic, semantic and word distribution heuristics to eliminate instances of text which people interpret easily. We report on a series of experiments and evaluate the performance of our approach against the collection of human judgements. Our machine learning algorithm has an accuracy of 75% compared to a 59.6% baseline.