Research On Personalized Electronic Document Recommendation Algorithm Based On Community Detection Of Scientific Research Paper Co-authoring Network

Nowadays,scientific and technological progress are new every day and different every month,with all kinds of information resources emerging in endlessly,overwhelming in people's vision everywhere.For scientific research personnel,selecting effective electronic document resources from massive information is time-consuming and laborious.How to recommend valuable electronic documents effectively from massive information for scientific research users based on personalized recommendation technologies become a problem to be solved urgently.Generally,after researchers' publishing papers,Chinese and English database such as CNKI,Wanfang,SCI,EI can retrieve the paper information and author information.Based on published research paper information,build scientific research paper co-authoring network,find each scientific research community by community detection algorithm.Members of the community belong to the same research area,or belong to the same course or research team.Based on the scientific research community,having the research on personalized electronic document recommendation can improve the recommendation efficiency and accuracy.The paper mainly studied personalized electronic document recommendation algorithm based on the community detection of scientific research co-authoring network,combining the community detection stage with personalized electronic document recommendation stage.At the stage of community detection,it could be subdivided into the construction of scientific research co-authoring network and the detection of scientific research community.First of all,through the data analysis of published papers of scientific research user of the certain college,got the paper co-authoring relationship,using Pajek to build the scientific research co-authoring network.Then,by constantly searching for the maximum node in the network,adding the neighbor nodes and comparison operation of connecting degree,and constantly expanding the community,and finally got all the scientific research community structure of scientific research co-authoring network.In the scientific research community,when having a the personalized electronic document recommendation for a useriU,firstly,extracted the key words from published papers ofiU,and constructed the user preference model.Then,all the users in the same community as the user iU were regarded as a set U,and published papers of each user as the first author in U were put together as the recommended document resource P.When theiU was recommended,calculated the similarity between the similarity calculation vector of the useriU 's interest and the similarity calculation vector of each document in the recommended document resource P,and recommended the target user with the top-N documents of maximal similarity degree.Finally,had performance analysis evaluation on the scientific research community detection algorithm by using the test data set,and obtained that the scientific research community division'seffect of the paper was pretty good,and had community division on experimental data set to obtain scientific research community structure.On the basis of scientific research community,had the research on personalized electronic document recommendation algorithm,and had performance analysis evaluation on the personalized electronic document recommendation algorithm using the accuracy rate,recall rate and F value index,and finally verified the accuracy and effectiveness of the personalized electronic document recommendation algorithm under the scientific research community of the paper.