Study On Mapping Between Heterogeneous Requirements From Multi-users

In the field of software engineering, the model for the commonalities and variabilities among a certain domain has become a hot topic. In the software product line(SPL) and Software as a Service(SaaS), we would like to provide more customized and distinct service to different customers. Since customers would describe their requirement for the software with different kinds of natural language, it is difficult for us to take full use of the automatic methods to implement the process of modelling. To solve this problem, it is necessary to search for the automatic mapping between the different requirements, which can help analyze the requirement for a certain domain.In this search, we discuss semantic similarity between use cases for the task of mapping between them. We will introduce the work from two aspects. Firstly, we will take the database from STS-2012 to research on parameter tunning, feature selection in SVM model on the task of semantic textual similarity. Secondly, based on the e-commerce background, we build the concept ontology. According the diffirent semantic frame, we develop the set of rules to abstract the semantic elements in the sentence. We introduced a concept similarity algorithm on the ontology network, and implement the similarity algorithm between the sentences using the similarity of the two conresponding semantic elements.