Research On Application Of MOOCs Reviews Based On A CRF Sentiment Classification Model

After the year of MOOCs,MOOCs has been more and more popular.MOOCs turn the high quality lesson from college to peers which is suitable for online study and put it in Internet which everyone from the world can easily get access to.Different from the way which traditional college teaches the students,MOOCs provide a new way which has great influence on education idea,education system and the way teaching.The feature of MOOCs such as open,massive and online give learners a good way to study,it also brings challenge that how to handle the large scale of reviews from the learners in MOOCs forums to prove teaching.However,traditional methods which use the sentiment lexicon which is used to statics the quantities of the positive world and the negative world to classify the sentiment polarity of the reviews,through the analysis,the sentiment polarity in document level and sentence level can be obtained.However,the emotion in one sentence may be different,so it is difficult to classify the sentiment of reviews accurately.And for the sentiment lexicons,it is difficult to cover all the sentiment lexicons in the text,especially the online reviews,online reviews which is informal,the data of which is sparse,and have many unregistered words in it,the new use of the old words also make the sentiment analysis difficult.In this paper we designed a sentiment classification model based on conditional random field and use it to analysis the reviews from MOOCs forum.We collect reviews from MOOCs forum,and divides the sentence of the reviews into basic emotion units.The sentence sentiment classification based on the basic emotion unit is more accurate and more reasonable than the traditional classification based on the document or sentence level.Then we use Stanford Parser to get the grammatical information of the clause,synthesize the NTUSD emotional dictionary,and extract the multidimensional features of the reviews.We use the method of machine learning to construct a semi-supervised sentiment classification model.At the same time,through the syntactic information of the comment text and the result of emotion classification,a Chinese emotional dictionary based on the text from domain field is established.And verified the effectiveness of this method in MOOCs reviews through experiments.