Research And Design Of Personalized Learning Resource Recommendation System

With the rapid development of network technology,cyber source showing explosive growth,people can get rich learning resources from the Internet,but the Internet has become one of the main ways for people to obtain information.However,with the increase of network information,when the user is faced with a large number of network information,easy to drown in the knowledge of the "sea",causing the problem of "information overload",people usually need to spend a lot of time or energy from the massive network of information to find the information they need.People usually use the way of information retrieval is to find the information they need,but this way can't satisfy with the increasing personalized needs,information retrieval methods usually require people to describe the information,but in many cases,the user actually to own demand is not very clear or very difficult to use the simple keyword to describe,at the same time,the traditional information retrieval technology is based on the key search,therefore,it can't tap the potential interest points of users,users can only find information that they are interested in,so in order to meet the growing demand for personalized services,personalized recommendation technology will be generated.At present,the application of personalized recommendation technology in the field of education is still relatively small,the related theory and application research is also scarce,and published related papers are few.On the whole,the research on personalized recommendation in China is still in its infancy,compared with foreign countries,there is still a certain gap,for example,the method of personalized recommendation in China is too single,and has not implemented diversification recommendation.Single recommendation is often more monotonous and can't meet the diverse needs of users,at the same time,in the recommendation process,the lack of intelligent processing level,the degree of automation is relatively low,Therefore,through the use of hybrid recommendation technology to play its respective advantages,so as to make the diversity of the recommended results,this article will adopt a variety of recommendation techniques to achieve diversification recommendations.With the continuous construction of learning resources network,learning resources are also increasing and rich.Facing a large number of learning resources,it is easy to cause the problem of information overload.In this paper,a personalized resource recommendation system is designed.By analyzing the learner's learning behavior,the initiative to recommend learning resources to users,by using hybrid recommendation algorithm,combining the advantages of different recommendation technology,achieve the diversification personalized recommendation results,better meet the needs of users,solve the information overload problem to some extent,the research in this thesis mainly includes the following parts:(1)Through the existing literature review,this paper will introduce the relevant theories of personalized recommendation technology,and explain its research status and application fields,this paper compares the advantages and disadvantages of the most popular recommendation techniques,and selects the most suitable recommendation technology for the resource recommendation system designed in this paper.(2)Develop user's scoring criteria for learning resources,there are two main ways to obtain the user's scoring data,namely,implicit mode and display mode.The display mode is that users directly score the learning resources,and the implicit way is achieved by the system.After the user logs in,the system automatically tracks the user's learning behavior in the system,when users browse,collect and recommend learning resources,the system converts the operation of learning resources to corresponding scoring data according to the scoring criteria,thus forming a scoring matrix between users and resources..(3)Through the comparison of some common personalized recommendation technology,this paper chooses collaborative filtering recommendation algorithm and social label recommendation algorithm,Collaborative filtering recommendation technology is widely used,and its performance is relatively mature.It only needs user's scoring data to recommend it,the advantage is that it can tap the potential interest points of users,but there are some shortcomings,such as cold start problems.In the early stage of the system,because the data is sparse,this will affect its recommendation effect.Therefore,in order to solve the cold start problem,this paper introduces the Slope one algorithm to deal with it,filling the sparse data,so as to improve the accuracy of collaborative filtering recommendation based on user,the experimental results show that,Slope one algorithm can alleviate the influence of data sparseness on recommendation accuracy to a certain extent.Recommendation based on social tagging is to establish user's contact with target project through user's tag behavior,so as to excavate user's interest,this article will use the social tag recommendation algorithm based on text resource recommendation,by setting the social tagging of text resources,then use text similarity algorithm to calculate the target resource and the most similar learning resources,and recommend it to the learners.(4)This paper will design and implement a set of learning resources recommendation system,and describes the overall framework of the system,requirements analysis,database design,sub module design,etc..