An efficient wikipedia-based approach for better understanding of natural language text related to user requirements

Requirements engineering is the initial step in the development of software. Eliciting and understanding requirements are crucial in that the quality of the resulting system is highly dependent on a clear understanding of the customers' needs. Customers frequently express requirements in natural language, and software engineers then transform these natural language requirements to a more formalized representation that is useful during the remaining steps of software development. This transformation process has the potential to introduce errors and misunderstandings in the requirements because of ambiguities and incompleteness found in the natural language requirements. A communication gap typically exists between customers and software developers who may not share the same technical expertise. Software developers typically have computer science backgrounds, while the software application areas and thus the expertise of the customers may be from different domains, including such domains as chemistry, aerospace, healthcare, and economics. This communication gap can lead to software failures that can have devastating consequences in many different domains, including the aerospace domain, in terms of time, money and mission-critical efforts. Numerous research efforts are found in the literature that describe how a large textual corpus such as Wikipedia can be useful for helping fill that communication gap between the customer and the developer when natural language requirements are used in the requirements engineering process [1][2][3]. Wikipedia is a large online information repository that has an interface that emulates a paper encyclopedia in its coverage of diverse topics. It provides a large number of real-world concepts organized in hierarchical semantic structures. The quantity of the information and the ability to search the information in Wikipedia make it a valuable resource for research related to requirements and natural language processing [4][5][6]; however, the quantity of information makes searching it time consuming. In this paper, we describe research to determine the most efficient approach to retrieve data from Wikipedia for use in improving the requirements engineering process. We refer to this step of retrieving the information of interest as knowledge acquisition. We present our analysis and compare time efficiency of the use of regular expressions and string search algorithms for decreasing the knowledge acquisition bottleneck caused by the high computational time needed to retrieve these data from Wikipedia.