Study On The Learner Sentiment Mining

Sentiment is an important part of the quality of students,It has a greatly effect onthe learning outcomes of learners. For example: learners who have positivesentiments (happy, pleasant, satisfied, adore etc.), study enthusiasm is to bereassignmented so as to improve the study effect; conversely, who have negativesentiments (irritability, hate, anxiety), study enthusiasm is to be suppressed,and thenthe study effect declines. So if the qualitative and quantitative analysis unified andmining the sentiments of learners, it will effetely improves the sentimentalcommunicate with learners, adjust the sentimental state of the learners, so as tostimulate the learners’ intrinsic motivation, improve the study effect.The learner sentiment mining as education science, software technology,psychology, and other areas of the multi-disciplinary cross, is increasinglydevelopment as one of the most important in the reseach field of instructionaltechnology. In this paper, the data of learner sentiment were collected and analyzedto look into the research trends in this field, and the prospect of the learner sentimentmining was previewed from the theory and practice exploration.Classic sentiment mining on learners research, with interviews and othermethods, the partial qualitative research, this paper sticking to the currentinformationalizing educational background, Based on information theory, control theoryand system theory(integrated Information Interaction System), based on the data from blogs,qualitative and quantitative combined to explore sentiment mining on learners. That is,based on the concept of information interaction system architecture to explorelearners emotional vocabulary, based on the lexical structure, will classify the textdata from the learners’ blogs, this paper will judge the learner’s affective categories bycomputing quantization. Sentiment analysis by using ROST software and artificialhigh frequency characteristic word, which results in0as the critical point. more than0as the positive sentiment, less than0as the negative sentiment. This paper found alarge amount of text information is best to use ROST software, on the other hand, byusing the artificial high frequency characteristic word analysis.