Research On Service Discovery And Recommendation Methods For Agile Mashup Development

Mashup application development is a new type of development technology that creates services that are more user-friendly,usable,and more versatile by mining and integrating external data sources(usually a single-featured Web service).With the maturity of Mashup application development technology,the low cost and rapidity of Mashup application development has been recognized by users(ie,mashup application developers),which has led to an increasing number of users of Mashup application development technology.However,when Mashup application developers face the explosive growth of services and data on the Internet,they often fall into the trap of service information overload.Therefore,there are two major research issues for Mashup application development:(1)How to efficiently and quickly discover available services in Mashup application development to meet the changing needs of users,so that existing services can be fully utilized.(2)How to recommend high-quality Web services based on user interests,thereby promoting the ecological balance of Web services.In order to cope with the above problems,A method of service discovery and service recommendation for agile mashup application development is proposed in this paper,and specifically carried out the following work:(1)A Service Discovery approach for Agile Mashup Development(SDAMD)is proposed,which breaks the limitation of single documentation.Firstly,the user story in agile development is search-driven.The service requirement forms the original search text under the bearing of the user story.Then,natural language processing technology is adopted to extract the three elements of agile demand from the original search text and the three attributes of agile service from the service description style.Then,a new similarity calculation method is proposed,which takes into account the part of speech,semantic weight and other information.Finally,by calculating the similarity between the three elements of agile requirements and the three attributes of agile services,the suitable service is found.(2)A service recommendation method for agile Mashup application development is proposed,which is composed of two parts: the construction of multidimensional information matrix(MIM)and predictive scoring by factor machine model(FM).First,the relationship between the Mashup application,the service and its links is represented by MIM.During the MIM construction process,we propose a similarity calculation method for the Mashup application,which combines the Mashup description text with the Mashup call service information for the similarity calculation of the Mashup.FM was then used to predict potential association scores between Mashup applications and services.Finally,the top-k related services are recommended for the target Mashup application development based on the predicted score.