An Approach to Improve the Language Quality of Requirements

Abstract

In this article, requirement authoring methods are investigated together with the way they impact the tasks carried out by technical writers. A method is proposed, based on natural language processing technology, to improve requirement production and writing. This method is based on the notion of correction memory and correction patterns derived from the observation of the corrections made by technical writers. This work remains preliminary and exploratory. We address in this article the case of fuzzy expressions and of complex sentences, which are major errors found in technical documentation.