Terms Extractions: An Approach for Requirements Reuse

Abstract

This paper presents a solution to a requirements reuse problem that utilises natural language processing and information retrieval technique. We proposed a semi-Automated approach to extract the software features from online software review to assist the process to reuse natural language requirements. We have conducted an experiment to compare the manual feature extraction versus the semi-Automated feature extraction. We used compilations of software review from the Internet as a source of this extraction process. The extracted software features are compared against the features obtained manually by human and the evaluation results obtained in terms of time, precision, recall, and F-Measure indicate a promising result. © 2015 IEEE.

Data-Driven Requirements Elicitation: A Systematic Literature Review

“…Only three studies compared the performance of the proposed artifact with the results of manual analysis without the aid of automation [57,62,78]. For example, Bakar et al [62] compared the software features that were extracted using their proposed semi-automated method with those that were obtained manually.…”

“…com [60]. Expert reviews were used in the 8% (n = 3), of which two were from multiple platforms [61,62], and one was from a single platform [63]…”

“…• A semi-supervised learning technique was used to classify app reviews into either functional or non-functional requirements [43]. Rule-based classification • Rule-based classification has been applied to identify and classify relevant requirements information, to identify software features [40,41,46,61,62], and to elicit requirements [94]. • Rule-based classification has been performed: using collocation finding algorithms [40,46]; grammar-based and delimiter-based strategies [76]; phrase search [86]; conducting language pattern matching which results are postprocessed with Part-of-Speech tagging [57]; n-gram analysis [102].…”

“…• Rule-based classification has been performed: using collocation finding algorithms [40,46]; grammar-based and delimiter-based strategies [76]; phrase search [86]; conducting language pattern matching which results are postprocessed with Part-of-Speech tagging [57]; n-gram analysis [102]. It has also been performed based on language patterns [37,61], or linguistic tags [62], propagation rules based on syntactic relations [41], or the SAFE patterns which consist of Part-of-Speech patterns and sentence patterns [94].…”

“…• Identification of candidate features has been done, using rule-based classification [40,46,61,62,100], topic modeling [49,52], traditional clustering (i.e., K-means [75] and the BIRCH algorithm [90]), a conceptual model [72,73], and rule-based clustering [74]. • In the several studies, classification was first performed, using supervised machine learning [47,48,58,60,68] or ruled-based classification [41,46] before identifying candidate software features using topic modeling [46,47,68], traditional clustering [41,58], word clouds [48], or conjoint analysis [60].…”