A Recommendation System for Functional Features to aid Requirements Reuse

Abstract

Software product lines (SPL) engineering is an efficient means to enhance software quality, support requirement reuse and develop variants of products. Functional and nonfunctional features can be extracted from SRS docs of ancestry built artifacts to aid RR. In this paper we offer a recommendation web tool (prototype) to extract functional features and calculating reusability for amount of data available in the form of SRS of already developed systems. In initial-level, SRS docs are feed into system. System accesses natural language requirements automatically from SRS. Terms extraction is performed which depends on keyword occurrences from several combinations of nouns, verbs, and/or adjectives. Phrases that reflect functional features reside on SRS docs were extracted by using information retrieval (IR). FRs are then stored in knowledgebase automatically. In Secondary-level, requirement analyst inputs summary of prospective system and selects the operation to perform i.e. simple and advance search. System applies POS-tagger technique on software summary for tokenization to search functional features. These tokens are then passed to inference engine to match between knowledgebase to identify which features could be recommended to analyst to aid RR. Matched features with queried features are prioritized using collaborative filtering to assist requirement analyst in making right decision in different software engineering tasks, starting from forming the teams and specifying the requirements to subsequent projects.