Crowdsourcing service requirement oriented requirement pattern elicitation method

摘要

"Pattern" can always help machine to recognize the new encounters, so does "requirement pattern." Requirement pattern is one of the essences for the cognitive service to understand customer's intention. Since crowdsourcing service platform holds abundant user demands in the form of text information, the method proposed in this paper aims at eliciting valuable patterns from this "treasure." This method is based on a knowledge graph, which is constructed with the refined concepts of those text information of several different domains. Due to the irregularity and difference of user demand expressions, this paper will firstly explain the knowledge extraction method for heterogeneous text and the knowledge fusion-based knowledge graph construction method. Afterward, we will introduce the requirement pattern elicitation method based on this knowledge graph. The pattern could either be a frequent demand sequence or a domain-oriented rule or link. Finally, this paper will demonstrate a case study to show how those patterns can help to understand customers' intention effectively and accurately.