An Augmented Vector Space Information Retrieval for Recovering Requirements Traceability

This paper discusses augmented information retrieval models based on the vector space model by employing additional retrieval strategies for improving automatic recovering of requirements trace ability. We propose two approaches for augmenting the vector space model. The first approach employs document identifiers of a term that indicate where the term has been found. The other approach deals with a set of closely related terms called "family terms" in a query. We assume that requirements and design models are written according to SysML specification with some degrees of consistent terminology. The contribution of the paper is three-fold: (1) to propose enhanced information retrieval models, (2) to establish processes to generate documents for information retrieval, (3) to evaluate the information retrieval models using SysML diagrams in industrial settings.