CREeLS: Crowdsourcing based requirements elicitation for eLearning Systems

Abstract

Crowdsourcing is the process of having a task performed by the crowd. Because of the Web evolution, recently crowdsourcing is being used in the field of Requirements Engineering to help in simplifying its activities. Among the information systems that were highly affected by the Web evolution are the eLearning Systems (eLS). eLS has special characteristics, such as the large number and diversity of users who could be geographically dispersed. To the best of our knowledge, there is little evidence that a crowdsourcing based requirements elicitation approach especially tailored for eLS that addresses their special characteristics exists. In this paper we attempt to fill in this gap. We present Crowdsourcing based Requirements Elicitation for eLS (CREeLS), which is made up of a framework of the necessary elements of crowdsourcing suggesting specific tools for each element, and a phased approach to implement the framework. We evaluated our approach through analyzing real-life users' reviews and extracted keywords that represent users' requirements by using topic modeling techniques. The reached results were then evaluated by manual text reviewing and the extracted features were found to be coherent. CREeLS has 0.66 precision and 0.79 recall. Hence we contend that CREeLS can help requirements engineers of eLS to analyze users' opinions and identify the most common users' requirements for better software evolution.