The Design And Implementation Of Web Services Tag Mining Module

Nowadays, more and more Web services are published on the Internet and provided to users. With so many services, mining tags is an effective way to solve the problem of service features reflect, and it can help discover and classify services. But traditional tag mining methods use small number of information source, and type of the resulting tags is single. Meanwhile, there are several problems in existing service search engine, such as not fully reflecting the features of service, the single search mode and the lack of interaction with users. Through expanding information source and mining tags from multiple perspectives, like emotion and function, this dissertation designs and implements the tag mining module of service search engine, and applies the mined tags to service search engine, which solves the not fully reflecting service features problems, etc.This dissertation solves the problem of insufficient information source by the way of a combination of multiple information sources. Existing information sources are mainly come with service-related information, such as WSDL (Web Services Description Language) documents and service descriptions. These information is fixed and lack of species. In this dissertation, the service-related web pages on the Internet will be used as an extended information source, and combined with the service-related information into tag mining information sources.This dissertation solves the problem of inadequate tags types by mining tags from multiple perspectives, such as emotional and functional perspectives. The tags generated by traditional mining methods are mainly functional tags and not rich in content. This dissertation mines tags from functional, emotional and the combination of these two perspectives by using feature extraction, sentiment analysis and LDA topic model technologies, then gets three types of tag sets.The dissertation also integrated the mined tags into service search engine to solve the problems of not fully reflecting the features of service, the single search mode and the lack of interaction with users. The problem of not fully reflecting service features is solved by three ways of displaying tags. Through tag searching and the combination of tag and keyword searching functions, the problem of simple searching way is solved. By providing the functions of adding and deleting tags and comments for users, this dissertation completes the interaction with users.Comprehensive testing and validation based on service search engine are carried out to verify the tag mining module. The result shows that the design and implementation of this paper have already satisfied the design goals.