Extraction And Analysis Of Teaching Content In Group Chat

Internet group tools such as We Chat and QQ are usually used to assist teaching.New content and new knowledge can be provided in the process of using these tools,and the situation of "teaching" and "learning" can be reflected to a large extent.However,the contents are rarely analyzed and organized.Combined with the teaching process,the analysis,mining and sorting of these contents is conducive to improving the quality of teaching and providing new resources.The group chat generated in the teaching process is selected as the research object in this paper.At first,the group chat dataset of the automatic control principle course is constructed.Then the relevant textbook is introduced for keyword extraction,and the word vector model of the textbook and the group chat is trained.The cosine similarity between them is calculated,and then data-cleaning is carried out to remove irrelevant information.And Text CNN is selected as the benchmark model for question extraction.To solve the problem of the uneven distribution of categories,combining with the text’s word characteristics,three optimization strategies are proposed: attention mechanism,balanced cross-entropy loss function and Borderline1-SMOTE algorithm.Next,the dataset is pre-segmented,then dialogue fragments and topic words are marked.After that,the single-pass algorithm is selected as the benchmark dialogue segmentation model.TFIDF training word vector and time features are used as the optimization strategy in a targeted manner.Finally,the Text Rank algorithm is selected for topic word extraction based on the features of the data set.From the experiment results,the test set accuracy of the optimized question extraction model is 92.03% and the F1-score is 0.7150.Compared with the benchmark model,the results are increased by 1.28% and 0.0847.The accuracy rate of the optimized single-pass clustering model is 66.9%,with the increase of 13.7%.According to the case and surrounding the dialogue fragments after segmenting,first the topic core question is extracted by the question extraction model.Then the keyword sentence technology is used for analyzing topic keywords and topic center sentences.Finally,a more systematic dialogue fragment system can be formed to provide help for teaching.Applying the extracted content to actual teaching,the teaching quality is improved.