Research And Analysis Of Bilingual Reading Behavior Based On Artificial Intelligence

In the context of the big data era,there have also been many problems with the development of online reading,such as the low completion rate of book reading and the inability of teachers to timely identify students’problems and provide effective guidance.In order to solve the above problems,this article mainly analyzes the bilingual reading behavior data of college students from two aspects:constructing a knowledge graph and analyzing emotions,and proposes improvement suggestions.The main research content is as follows:This article will use the database server and log server of the teaching platform as the main data sources,construct a behavior analysis index,determine the learning population,and construct a multi-dimensional and multi-level online reading behavior data model.A bottom-up construction method was adopted to construct a knowledge graph of bilingual reading behavior among college students based on a graph database.Dictionary based sentiment analysis,as the most fundamental method of sentiment analysis,usually does not have a very suitable dictionary for specific fields to use.This article is based on the How Net and NTUSD sentiment dictionaries,and adds auxiliary sentiment dictionaries to form a comprehensive sentiment dictionary,which includes negative words,degree adverb dictionaries,and turning point dictionaries.And use the book review data as the corpus for expanding the dictionary,and use the SO PMI algorithm to construct an emotion dictionary suitable for book reviews.Use a universal dictionary,a comprehensive dictionary,and the sentiment dictionary constructed in this article to classify emotions in the same corpus,and compare the classification results.The experimental results show that the accuracy of the dictionary constructed in this article is 83.44%,which is better than 72.90%of general dictionaries and 79.62%of comprehensive dictionaries.Prove that the domain emotion dictionary constructed in this article has more effective emotion classification and recognition.In Chinese texts,the context is complex and there is often a phenomenon of polysemy,which leads to certain errors in sentiment analysis methods based on sentiment dictionaries.In order to improve the performance of sentiment classification,this article proposes a method that combines the optimized and expanded sentiment dictionary with machine learning to perform sentiment analysis on book reviews.The sentiment dictionary is used to calculate the sentiment values of book reviews.Based on the sentiment values,a definite set with obvious emotions and an uncertain set with fuzzy emotions are obtained.The definite set is used as machine learning training corpus,and the classifier is trained using two different algorithms,Naive Bayes and SVM,The uncertain set is used as the corpus to be classified,and the classification results of the determined set are based on the dictionary classification results.The classification results of the uncertain set are corrected by combining two methods.The experimental results show that the accuracy of combining sentiment dictionaries with Naive Bayes is 85.32%,which is better than 83.56%using Naive Bayes alone.The accuracy of combining sentiment dictionary and SVM is 89.43%,which is better than 85.12%using SVM alone.Both are superior to 83.44%of the emotional emotions constructed solely using this article.In terms of accuracy,SVM is 4.11%higher than Naive Bayes in the fusion algorithm.Finally,based on the knowledge graph of reading behavior and the emotional analysis method of book reviews,the following four suggestions for improving college students’ online reading performance are summarized:(1)improving students’ self-monitoring ability and cultivating good reading habits;(2)Teachers enrich online activities and enhance platform application capabilities;(3)Increase extracurricular reading time and introduce a reading grading system in schools;(4)The platform integrates learning resources and improves teaching support services.