Calculation Method On Text Orientation In Digital Campus System

With the rapid development of information technology and the Internet, more and more netizens prefer to express their views, opinions and comments on the Internet. Massive information resources distributed on the Internet have become the main source of information in people’s daily life. This thesis collects data information based on teaching evaluation system of digital campus. By using some natural language processing such as segmentation, part of speech tagging, keyword extraction, combined with the teaching evaluation of short text messages and the existing emotional dictionary. To expand the existing emotional dictionary by part of speech tagging method, building the vector space model and adjusting weights of sentimental influence factors dynamically. Analyzing the emotion of teaching evaluation information and calculating the score, finally get the report of corresponding emotional analysis. At the same time, there are important research significance and application value for the public opinion monitoring and evaluation etc. The main research works are as follows:(1) For the problems of applicability and coverage on the sentiment dictionary, it studies and puts forward the construction method of the sentiment dictionary based on semantic similarity. Collecting corresponding data information by logining the system of digital campus; constructing a domain emotional dictionary by combining with semantic similarity algorithm on the "HowNet"; Analysing the speech and the emotional strength of the constructed emotional dictionary. Experiments show that the research based on the construction method of the sentiment dictionary based on semantic similarity is feasible and effective.(2) For the problems of rate and accuracy on the Chinese participle, it studies and puts forward the method of Chinese participle based on the double-array Trie segmentation. Combined with some technology of natural language processing such as speech tagging, word sense disambiguation and unknown words recognition, marking the parts of speech based on the existing analysis system1CTCLAS which is developed by The Chinese Academy of Sciences and the open source Java segmentation tools Ansj and getting the final result of segmentation. We verify it from the segmentation speed rate and the space occupied by experiment. Experiments show that the research based on the double-array Trie segmentation is feasible and effective.(3) For the problems of numerical quantify on the emotional polarity, it studies and puts forward the method of the emotional polarity calculation based on keyword statistics. Analyzing the factors which influence emotional polarity, and calculating the factors’weight and building the model, and adjust the calculation result through the static and dynamic weight of the experiment, get the best value of the influence factor. Experiments show that the research based on the keyword statistics is feasible and effective.(4) We design and implement the sentiment analysis system based on digital campus based on Chinese segmentation and emotional polarity calculation. In order to make the result is more accurate, in this paper, it designs and implements two important modules such as Chinese segmentation and emotional polarity calculation. The effect of the running shows that the sentiment analysis system based on digital campus is running well.