Service Discovery Based On Agile Software Requirement

In the case of expanding service scale,increasing service types,accelerating user demand changes,and increasing business cross-cutting,how to better understand customer requirements and respond quickly to user needs,and more accurately perform service discovery or service recommendation.The traditional method of service discovery based on keyword matching does not understand user requirements well,and in the case of big data,full matching is already overwhelmed.There are many research found in domestic or foreign countries,but there are several shortcomings in the existing methods:1)there are certain requirements for the type of document;2)the semantics of service requirements are rarely considered;3)few users understand requirements from the perspective of team development.In order to come over the above deficiencies,this paper work on how to improve the effect of service discovery and make up for the shortcomings of traditional service discovery.As we all know,the service discovery method include two phases:service clustering + matching services.Therefore,the main contents are as follows:(1)Service clustering(Phase 1):This paper proposes Service Clustering based on Semantics of Requirements(SCSR).Clustering can narrow the space of service query and improve the efficiency of service discovery.The service information set is extracted by dependency syntax analysis,and the similar services are clustered into clusters by the clustering algorithm.(2)Services Matching(Phase 2):Introducing a Service Discovery Approach for Agile Development(SDAD),taking the clustering result of the first stage as input and introducing agile in agile development.The software requirements define a three-state structure corresponding to the user story,and the service description documents are converted into a triple form.The similarity is calculated and sorted to obtain the recommended result.This paper uses the real data of the API service on ProgrammableWeb to verify the effectiveness of the SCSR method and the S DAD method.