Research On Resource Portrait And Recommendation Based On Online Learning Platform

With the continuous development of hybrid learning,it provides an online and offline learning mode for higher education.Online learning platform is increasing,and digital resources in the platform are constantly being learned,accumulating a large amount of resource centered behavioral data.The application and construction of digital resources has become the focus of relevant researchers for a while.At present,the domestic research on teaching resources focuses on how to optimize the construction of digital resources,the construction of teaching resources platform and other topics,and has achieved some results.For the application analysis of different types of digital resources and the construction of resource portraits,it is still a little rare.In view of the above problems,the paper collects the behavior logs of learners,excavates the resource attribute characteristics behind the data,constructs resource portraits and learner portraits,helps resource designers improve the quality of resources,provides personalized resource recommendations for learners,and improves the resource utilization rate of online learning platform.The research content of this paper includes the following aspects:1.On the basis of summarizing the dimensions and models of resource analysis at home and abroad,this paper constructs an online learning platform resource analysis model from three aspects:basic attribute,behavior attribute and result attribute,and explains the calculation of some indicators.2.After expert evaluation,1400 resources,including three class a curriculum resources,three class B curriculum resources and two class C curriculum resources,are selected as research objects.After data collection,cleaning and preprocessing,analysis and data visualization are conducted based on the resource analysis model of online learning platform.Through data analysis,we found that:(1)Class A curriculum resources and class B curriculum resources are in a healthy state,and class C curriculum resources are in a healthy state.(2)Text resources are more popular than video resources,but there is a waste of resources.(3)The learning time of A/B/C course resources on the web is much longer than that on the mobile.The average learning time of class a course resources is much longer than that of class C course resources.Most learners use the web for online learning.(4)When exploring the characteristics of text resources,the resource heat of procedural knowledge in eight courses of A/B/C curriculum resources is greater than that of descriptive knowledge.In addition to course C2,the learning standard of descriptive knowledge is greater than that of procedural knowledge,and the procedural knowledge in the other seven courses is greater than that of descriptive knowledge.(5)When exploring the characteristics of video resources,the average value of resource heat of resources over 10 minutes is the highest,the average value of resource heat of resources over 5-10 minutes is the second,and the average value of resource heat of resources between 0-5 minutes is the lowest.However,although the popularity of 0-5min video resources is low,the learning standard of curriculum resources is higher than the average;the resource popularity and learning standard of teachers and captions are higher than the other three categories;the resource popularity and learning standard of procedural knowledge are higher than those of descriptive knowledge.3.Resource portraits are explored from the construction process of resource portraits,the design of resource portraits and the drawing of resource portraits.Among them,the drawing of resource portraits includes the drawing of individual resource portraits and the drawing of group resource portraits.4.From the technical modeling,data model design,label system design,portrait drawing to explore the learning portrait.5.The Resource Recommendation Model Based on learners is constructed.PMF+and BiasSVD+models are used for recommendation,and RMSE and Mae are used for algorithm model evaluation.Among them,RMSE of PMF+is 1.5615,and RMSE of BiasSVD+is 1.4123.