Opinion Mining Applied In The Evaluation Of Teachers’ Educational Technology Ability Training In Primary And Secondary Schools

Opinion mining based on data mining and text mining technologies, is the current hot topic of natural language processing,, which has a certain ability to understand and analysis text, and it can help people quickly get the overall emotional tendency from a lot of comment information. This article comments apply to mining education technology training school teachers evaluation, designed a training comment for opinion mining platform. Evaluation object extraction and emotional orientation analysis is the basis and key opinion mining. This paper reviews the use of training as a research corpus, focusing on these two core tasks in-depth study, are summarized as follows:(1) A pattern matching method to be extracted based on the evaluation. First, the establishment of the corpus, then the seeds by constructing rules, extraction effective evaluation sentence. Re-use POS combinations and speech from the correlation algorithm to extract the evaluation.(2) In the evaluation of the extracted object associated aspects associated with the object to evaluate the proposed object-centric model based on the evaluation of syntactic tree identification method. Proposed a capable of recognizing objects associated with the evaluation of the method, the syntax tree model will be evaluated as a center, in order to achieve the object associated with the evaluation of the extraction. First, the Stanford Parser parsing tool to generate syntax tree constructed syntax tree pattern library, and the syntax tree nodes generalization in combination to improve the evaluation of the word dictionary and emotional match rate, and then use the similarity algorithm for pattern matching.(3) the emotional propensity analysis, build a suitable network comment emotions dictionary sentiment analysis, and polarity analysis method using the dictionary. At the same time, taking into account the evaluation of the word is a combination of words after the generalization process, and the increasing scale, existing emotional dictionary has been unable to fully covered, therefore, we propose to split the identification and semantic similarity algorithm into the emotional dictionary matching method.(4) Based on the above research, design and implementation of an opinion mining platform, and build experimental environment for the design of the platform for testing. Experimental results show that the proposed mining method comments can effectively reduce false positives affective tendency to improve the accuracy and recall rate.