| College students,as one of the main forces in a large network of people,are in the stage of long knowledge and most keen on new things.They like to express their opinions and dare to express their attitudes and positions publicly.Mining the concerns of college students is conducive to further understanding of what college students think and what they are looking forward to,early detection of problems with signs and tendentiousness,strengthening the scientific,targeted and initiative of university managers’ decision-making,so as to resolve conflicts,straighten out sentiments and solve problems as soon as possible.Therefore,it is imperative to establish a system of college students’ concerns recognition and sentiment analysis.Based on the analysis of college students’ concerns and their sentiment analysis at home and abroad,this paper constructs a classification model of college students’ concerns by crawling Baidu Post Bar data as an important source of college students’ concerns.At the same time,it uses neural network method to analyze sentiment tendency and uses rule-based method to analyze the reasons for the emergence of sentiments.This paper mainly carried out the following research work:(1)Recognition of college students’ concerns based on classification.First of all,a three-level classification framework of concerns is established.The first level is three main categories,the second level is three sub categories,a total of ten sub categories,and the third level is the sentimental classification results of each sub category.After that,the method of multi-label classification is used to classify the crawling data.Based on the classification framework,this paper proposes a focus classification model combining TF-IDF and Word2Vec.The model first extracts the category keywords after post classification by TF-IDF method,then calculates the similarity between the extracted category keywords and the keywords in the text to be classified,and then obtains the category of the post to be classified.(2)Sentiment analysis of concerns.First of all,the paper uses word embedding to construct word vector and get the semantic expression of the text by extracting the sentimental features between words.After that,the neural network model is trained to get the sentimental classification of college students’ concerns.Based on the time series,the evolution of sentiment tendency is analyzed.Then,we identify the reasons for the emergence of the concern sentiment.In this paper,by expanding the sentimental words list,improving the matching degree between the text and the sentimental dictionary,further improving the sentimental rules,and accurately identifying the reasons behind the sentiments of college students’ concerns.(3)Establish a visualization system.Based on the theoretical research,the paper comprehensively uses graph theory graphics,visualization technology,object-oriented programming ideas,and algorithm design and analysis theory.This paper design and implement a student’s concern and sentiment cause recognition system based on data mining.Modules such as analysis of college students’ concerns and sentimental tendencies are integrated and displayed to visualize the latest development trend of concerns and provide technical support for supervisors’ decision-making. |
