Research On Text Orientation Based On Template Matching

Text sentiment analysis technology uses computer to automatically analyze the emotionalinformation,such as the opinions and attitudes expressed in the text with emotional and subjective. Textsentiment analysis has been widely used in spam filtering, product evaluation, public opinion analysis andother fields.In Chinese, some of the words in different contexts have different means. However, the traditionalsentiment analysis have ignored the influence of context on the sentiment orientation. In order to solve theproblems, this paper uses the template matching method to judge the sentiment orientation of the words inthe context, and then analyzes the emotional orientation of the text based on the words sentimentorientation. So that we can determine the word sentiment orientation in the specific context, and improvethe accuracy of the text sentiment orientation analysis. The main work of this paper is as follows.(1)Word similarity computation. The remaining word similarity questions based on HowNet has beenanalyzed. Improved methods of word similarity computation, making the word similarity computationmore accurate and rational. The word similarity computation lay a foundation for the word sentimentorientation.(2)The construction of sentiment vocabulary ontology. According to HowNet and semantic dictionarybuild sentiment vocabulary ontology. Using the sentiment vocabulary ontology filter out corpus ofsentimental orientation, which provides high-quality corpus for the text orientation analysis experiment.(3)Text sentiment orientation analysis. First, in view of some words with different orientation indifferent contexts, we divide the sentiment words into two categories: single words and ambiguous word.The orientation of the single word is judged by the close degree between single word and standard word.As for ambiguous word, first judge the orientation in the context based on template matching method andthen calculate tendency strength. After that, the orientation of text is analyzed according to the orientationof the words and the impact of adverbs and rhetorical questions. Finally, experimental results show that theaccuracy of web review orientation analysis has been improved.(4)A application of text orientation. The text sentiment orientation analysis based on template matching is applied to the analysis of writing sentiment. According to analysis of the orientation to thestudents’ composition teachers can have timely understanding of the students’ sentiment orientation, whichhas a function of guidance for teachers to understand the mental health status of students.Theimplementation of this function is a new attempt of text orientation analysis applications, and alsodemonstrated the feasibility of template matching method applied to the orientation analysis of the writingsentiment.