Advanced Search
    Peng Zhenlian, Wang Jian, He Keqing, Tang Mingdong. A Requirements Elicitation Approach Based on Feature Model and Collaborative Filtering[J]. Journal of Computer Research and Development, 2016, 53(9): 2055-2066. DOI: 10.7544/issn1000-1239.2016.20150426
    Citation: Peng Zhenlian, Wang Jian, He Keqing, Tang Mingdong. A Requirements Elicitation Approach Based on Feature Model and Collaborative Filtering[J]. Journal of Computer Research and Development, 2016, 53(9): 2055-2066. DOI: 10.7544/issn1000-1239.2016.20150426

    A Requirements Elicitation Approach Based on Feature Model and Collaborative Filtering

    • With the rapid development of Internet and Web service related technologies,developing software system on Internet is increasingly popular. Software development is a multi-knowledge-intensive process in which requirements elicitation plays a key role. Software systems deployed on Internet need to meet the needs of various kinds of customers and users who are geographically distributed,which increases the difficulties of software requirements elicitation. Meanwhile,more and more software systems that share similar functions are deployed on Internet,which provides a new way to elicit software requirements. Toward this end,recommender systems have been leveraged in the requirements elicitation to recommend missing features for software products by comparing the requirements descriptions of existing similar software systems. In order to improve the prediction accuracy of the recommended features of the software system,a requirements elicitation approach based on feature model and KNN (K-nearest neighbors) collaborative filtering recommendation system is proposed in this paper. An algorithm named FM_KNN is presented by utilizing constraint relations between features in KNN collaborative filtering recommendation system. Experiments conducted on a real data set and a simulated data set, by comparing the proposed FM_KNN with two existing methods, i.e., KNN and an approach that combines association rule mining with KNN, show that the proposed approach can achieve higher precision.
    • loading

    Catalog

      Turn off MathJax
      Article Contents

      /

      DownLoad:  Full-Size Img  PowerPoint
      Return
      Return