Research And Implementation Of Learning Resource Personalized Retrieval Strategy Based On The Relationship Among Knowledge Points

This thesis mainly studied the problems of knowledge incoherence of learning resources,unsystematic arrangement of learning contents and search topic drift for learning resources.This is caused by lacking of structure relevance of knowledge and structures between the knowledge points and unclear searching intentions.By searching and electing knowledge through combined knowledge-based and corpus-based methods,identifying the searching intention of a user and analyzing the learning style and condition of the user to achieve the knowledge of the search results systematic,relevance,the accuracy of user intent recognition and personalized search results.First of all,we studied the extraction of knowledge points related to query content and the expansion of related knowledge points.Through the combination of corpus and knowledge base,the matching degree between query keywords and knowledge points of the knowledge base is achieved.Finally,explicit knowledge points of query content and extension of tacit knowledge points are realized.Explicit knowledge points can be used to determine the purpose of the user's learning.The tacit knowledge points can be used to predict the user's learning path and realize the knowledge system and relevance of user learning.Next,the thesis studied the identification of users'retrieving intentions.For solving the trouble of excursion in query topic often caused by expansion of knowledge points,this thesis puts forward a method of identifying the real intention of users.Through filtering all associated knowledge points so as to keep the queried knowledge points fitting the needs of users.And the thesis also proposes forgetting function based on temporal distance for updating weights of knowledge points within the topics of the users,in other words,ensuring the timeliness of retrieving results.In this way,the user's attentions to knowledge that appear repeatedly and that appeared in other times are carefully measured,thus improve the reliability of the model and make the retrieved results more precise.Finally,the thesis studied the user-individualized resources presentation.This thesis proposes an optimizing method of retrieving based on learning style analyses of users.User-individualized presentation of learning resources can be achieved through a three-dimension analysis of users' learning style including information input,content understanding and cognitive level,perfecting the retrieved results.