Research And Implementation Of APP Software User Intention Mining Method Based On User Reviews

At present,the variety of App software is becoming more and more diverse,which affects people’s life from all aspects.With the popularity of App software applications,the number of user’s comments for App software has increased dramatically.Mining valuable user’s intention for App software based on user’s comments can help developers to maintain and improve App software pertinently.Under the environment of online network,the update of user’s comments is fast.The amount of user’s comments is large.A large amount of useless data will affect the mining of valuable user’s intention information.It is not useful to mine user’s comments reflecting actual usage of software through sentiment analysis technology.If supervised learning is adopted,data resources manually annotated need to be constantly supplemented or even reconstructed.In addition,if only the comment content is adopted to mine user’s comment,the importance of sentence structure is ignored.Therefore,in this dissertation,user’s intention mining of App software based on user comments is studied.The comment content and sentence structure of user’s comments are analyzed comprehensively.User’s comments reflecting user’s intention are mined automatically through semi-supervised learning.The main work of this dissertation is as follows:(1)User’s intention of App software is defined as two types:feedback for current usage of App software and expectations for future improvement of App software.(2)A method of mining user’s comments reflecting feedback for current usage is proposed.The abstract representation of feedback for current usage is defined,including feature words and comment seeds of feedback for current usage.Combined with the definition of abstract representation,the semi-supervised learning method is adopted to mine the user’s comments that reflect feedback for current usage of App software.(3)A method of mining user’s comments reflecting expectations for future improvement is proposed.The extraction rules of evaluation object and evaluation opinion are proposed.The definition of comment seed and the matching algorithm between user’s comment and comment seed are improved based on the extraction rules.For the user comments that fail to match comment seeds,the type of expectations for future improvement reflected in user’s comment is identified(4)A method of mining user’s comments reflecting user’s intention for App software is proposed.Abstract representation of user’s intention of App software is defined.A semi-supervised learning method is used to dynamically expand the comment seed library and feature words table.Convergence conditions is defined for cyclic mining.User’s comments reflecting user’s intention of App software are mined until cyclic mining converges(5)According to the proposed mining method of mining user’s comments reflecting user’s intention for App software,a prototype is developed.The effectiveness of the method in this dissertation is proved through experiments.