Research On The Application Of Knowledge Tag In Network Learning

With the continuous development of Internet technology,online learning plays an increasingly important role in education.Especially during the epidemic period in the spring of 2020,under the requirement of "no suspension of classes",online learning played an incomparable advantage,solved the problem that students could not go to school during the epidemic period,and demonstrated the great significance of online learning in modern life,work and study.The auxiliary function in the network education platform can bring a better sense of use to learners and improve the efficiency and quality of learning.Therefore,the research and application of auxiliary functions have certain practical significance.This thesis studies and implements the auxiliary function of knowledge tag in teaching video,which can help learners understand the video better and also help to realize fast and accurate video retrieval.Combined with semantic information,this thesis proposes an improvement to the LDA theme model,which makes the theme division more accurate and the theme words have higher correlation.Afterwards,S-LDA keyword extraction algorithm is proposed based on the improved LDA theme model.Finally,research and construct a network education platform.The knowledge tag module of teaching video is designed and implemented on the platform based on S-LDA keyword extraction algorithm,and the knowledge tags are assigned to the video.The main work of this paper is as follows.1.Aiming at the deficiency of LDA subject model,a new model is generated by improving LDA subject model with semantic information.Build a word vector model of Chinese Wikipedia corpus by using Word2vec.In the process of LDA model training iteration,the word distribution calculation process in LDA model is improved by using the word vector model to calculate the similarity between the word vectors,and the prior distribution of theme-word,namely Dirichlet distribution,updates its superparameter dynamically.2.Propose a keyword extraction algorithm based on LDA theme model,and propose S-LDA keyword extraction algorithm based on improved LDA theme model.Based on the LDA model of the given text set,the target text file is predicted,and the theme distribution of the target text file and the theme-word distribution of the model are obtained.For the word distribution of each topic in the target text,calculate the similarity between it and the auxiliary vector with uniform word distribution and filter out the topic with high similarity,so as to get the filtered topic distribution of the target text.A certain number of words are selected from each topic to form the keyword candidate word set according to the proportion of the topic in the document topic.Set an auxiliary vector evenly distributed on all topics,and filter out the words with high similarity with the auxiliary vector in the keyword candidate words set.Finally,words with noun or verb in part of speech,appearing in the target text and ranking in the top s are selected from the candidate word set as the keywords of the target text.3.Research and construct a network education platform,then design and implement the knowledge tag module of teaching video on the platform.First,extract the speech from the learning video.Audio files are converted into text files using voice transcription technology.Then by using S-LDA keyword extraction algorithm to extract the keywords of the text file,the keyword database is established to give knowledge tags to the video.Finally,test and analyze the knowledge tag module deployed to the platform.