Automatic Summarization Of Book Review Based On Multi Source

As an important form of current book evaluation,book reviews reflect readers' subjective opinions and value judgments on the quality of a book.Mining book reviews not only helps users to understand the content of books,but also serves as a feedback mechanism to help publishers improve book quality and optimize marketing strategies.Although the traditional expert reviews are authoritative,they are too few to cover all the books.There are a large number of online reviews generated by the Internet environment,but the quality is uneven,which to some extent improves the difficulty of obtaining useful information.How to get the information what users need from the massive book reviews is the core of the review mining.By presenting the results of comment mining in the form of summaries,users can get a glimpse of the key content of comments by simply reading the summaries,which greatly improves the efficiency of users' acquisition or understanding of information.Therefore,how to provide users with concise and accurate summaries of book reviews has important research significance.In addition,different book review platforms have great differences in reviews.For example,Amazon and other e-commerce platforms contain abundant information such as book price performance,packaging and so on,while Douban.com contains information related to book content quality.Only based on the comments of a single platform,users can't get a comprehensive understanding of books.Therefore,this paper uses multi-source data,with the help of natural language processing,automatic summarization and other technologies to dig indepth reviews,so as to build a book review summary.In the research of difference analysis on multi-source difference data,this paper takes online book review data of e-commerce platform,community reading platform and expert review platform as experimental data,and analyses the differences of data of different platforms from four aspects: rating,number of reviews,number of positive reviews and number of negative reviews.Then,this paper takes the library collection information provided by Online Computer Library Center(OCLC)as the validation data of book evaluation,and evaluates the effect of book evaluation based on multi-source data,so as to provide a basis for the design of book review summary model based on multi-source data.In the research of theoretical model construction of book review summary,in order to fully consider the difference of data between different platforms,this paper puts forward two basic principles of building review summary,and gives the structural definition of book review summary.On the basis of this review summary theory,this paper proposes a method of generating book review summary based on multi-source heterogeneous data,which combines aspect sentiment analysis technology and multi-document automatic summarization technology.In the module of book review summary generation,this paper divides it into two stages: automatic construction of book aspect summary and automatic construction of book content summary.In the research of automatic construction of book aspect summary,we use attribute sentiment analysis method to mine book review data in fine granularity,and extract book aspect information automatically from two dimensions of aspect attention and aspect satisfaction,so as to form book aspect summary.In the research of automatic construction of book content summary,this study uses content sentences in dataset.On the one hand,it uses multi-document automatic summarization technology to extract book content information automatically.On the other hand,it uses rule-based method to extract important opinion sentences from content sentences,and then integrates book content information and core opinion sentences to obtain book content summary.Finally,by integrating the basic information of books,the abstract of book aspects and the abstract of book contents,the abstract of book reviews based on multisource data is obtained.This study can not only provide a new perspective for the study of book review abstracts,but also provide a reference for users' book selection practice.This study can not only provide a new perspective for the study of book review abstracts,but also provide a reference for users' book selection practice.