Crowdsource requirements engineering: using online reviews as input to software features clustering

As to date, various software being produced to help in our daily routines. At times, there are complaints on errors or faults lodged by users over the internet. This information can be valuable for software development teams to enhance the software functionalities in the next releases. Not only that, these comments contain important software features that can be extracted and reuse for future development of similar software systems. Reviews provided by various user from unknown background is an example of open call involvement in crowdsource software engineering. In this paper, sample software reviews available in the internet were collected. In the experiment conducted, twenty-five groups of random software reviews within the domain of children online learning software were selected as input to crowdsource requirements engineering. T h e extracted reviews were then clustered into related groups by using K-Means algorithm. The clustering results achieved by K-Means were evaluated in terms of cluster compactness and cohesion. A statistically significant result with time efficiency obtained and reported at the end of this paper. Based on this information, this paper provides some recommendations on how user reviews can be used as input to the crowdsource requirements engineering either for improving existing software or for production of a new similar systems