The Emotional Tendency Analysis Tesearch Of Online Comments On Teaching Materials

With the rapid development of e-commerce,online bookstore has become an important platform for many businesses to sell books,owing to the reasonable price,easy to buy and other advantages of the online shopping,it has gradually become the preferred way to purchase books.After reading the books,more and more users are also keen to share their real views or experience about the books through the internet.A large number of book reviews appear in the e-commerce site,these reader comments contain the estimates of the books,through it potential consumers can reduce the risk of purchase and obtain a satisfactory shopping results,businesses and publishers can make reasonable and effective decisions based on them.It can be seen that the excavation of online book reviews is of great significance and practical value for consumers,merchants and publishers.Therefore,we use the fine-grained sentiment analysis technology to analyze the online reviews of the teaching material books,and obtain the emotional analysis results of the teaching material's feature level,to provide valuable reference information for consumers and businesses.First of all,we analyzed the research status of the emotional tendency analysis of online comments in both coarse and fine grain levels at home and abroad,then studied the related theory and technology of fine-grained emotion analysis,ascertained the steps of sentiment analysis and the key technologies in each steps.On this basis,through the web crawler software to collect the online comment information of the teaching materials,and then to do the data processing,such as redundant data cleaning,Pinyin English replacement and so on.formed the training and test corpus to analysis textbook reviews.Then use the Chinese word segmentation software and custom word segmentation dictionary to complete and optimize the word segmentation and POS tagging.Based on the annotation results,according to the rules that product features are usually nouns and noun phrases,summed up the word-formation rules of noun phrases,making use of this rules,extracted the candidate product features from the textbook reviews,filtered it by frequency filtering and manual screening,to complete the construction of feature thesaurus of teaching materials.Then,according to the domain characteristics of the textbook reviews,on the basis of the general emotional dictionary,using the training corpus to build domain affective lexicon,network emotional dictionary and polarity modified emotional dictionary,to form an sentiment lexicon in the field of textbook reviews.Finally,based on the construction of the polar dictionary and feature thesaurus,we found that the existing SBV algorithm is applied to the textbook comments still can not recognize some of the characteristics and views,throughanalyzing the problems,we puts forward improving ideas,finally designed an emotion analysis algorithm which is suitable for textbook reviews.Through the test corpus to do the experiments,the results show that this algorithm and dictionary resources compared to the general emotional dictionary and SBV algorithm,the evaluation index is obviously improved,which proves the validity of the resource and algorithm design.