| With the Internet stepping into the age of WEB 2.0,people are not only consumers of Internet information,but also the producers.Nowadays,it's getting much easier to publish our opinions of products,services and real-time events to the Internet.Under this trend,the mining of online reviews has become a hotspot in recent research.The research of this thesis is focused on tag extraction of online reviews.First,a method of extracting comment object and comment evaluation based on grammar dependency relationship analysis and graph model is proposed.Based on the grammar dependency relationship analysis,the graph is constructed with words as nodes,the direct and the first order indirect dependency as the weight of the edges.Extract the tuple of comment object and comment evaluation,with which generate the review tag with the classified comment object and joint sentiment.Then regarding tags as text sequence,an improved local attention mechanism applied in the sequence to sequence learning neural network model is proposed,which constructs a review summarization model.By taking advantage of the feature of review summarization samples,that the words in the abstracts appear more at the start of the source reviews,the local attention mechanism is improved which has more attention weight on the start of the source sentence.Finally,a system of rural tourism reviews mining is designed in this thesis using the above tag extraction method.The system consists of data acquisition module,data preprocessing module,tag extraction module,sentiment analysis module,local corpus module,storage module and interface gateway module.As rural tourism is developing rapidly in recent years,applying the tag extraction method to the mining of rural tourism online reviews and showing in the style of tags provides a more convenient and quick way for rural tourism attendants,monitors and tourists to get the desired information,which will promote the development of the rural tourism industry. |
PDF Full Text Request