Research On Scholar Recommendation Based On Dynamic Features In Virtual Academic Community

The rapid development of Internet technology and the continuous increase of Internet penetration rate have made social networks play an important role in people’s life and learning.The virtual academic community fully relies on Internet technology and social networks to connect scientific researchers at home and abroad,providing them with a platform for resource sharing and academic exchange that is not restricted by time and space.However,with the increasing frequency of academic sharing and communication behaviors of users in the virtual academic community,the academic information in the community has grown massively,and the problem of information overload has become more and more serious.The rapid increase in the number of scientific research results and the uneven academic quality have seriously affected the use of virtual academic communities by scholars.How to help scholars quickly establish contact with potential collaborators with similar research interests in an open knowledge exchange environment,and efficiently obtain academic resources has become a key issue for virtual academic communities to improve services.When constructing a virtual academic community scholar recommendation system,it is necessary to integrate the multi-dimensional characteristics of scholars for mining and analysis,for example,based on the characteristics of scholars’ scientific research interests,and integrate scholars’ natural attributes,ability attributes,and social attributes.However,the research interests of scholars are changing in stages,and recent research results can better represent the current research interests of scholars.Most of the current recommendations of scholars are based on the fact that the research interests of scholars are static,and the dynamics of research interests of scholars are less considered.Variety.Based on this,this paper proposes a scholar recommendation model based on dynamic interest characteristics.The model is based on the academic research results of scholars in the virtual academic community.At the same time,a time-weighted function is introduced to extract the dynamic interest of scholars.The scholars are clustered,and on the basis of clustering,the ability and social attributes of the scholars are integrated to carry out the research on the scholars’ personalized recommendation.When modeling scholars’ dynamic interest features,this paper first uses the LDA topic model to extract scholars’ scientific research interest features from papers in virtual academic communities,establishes a scholar’s interest feature vector,and introduces a time weighting function to weight the scholar’s interest features to form a scholar’s dynamics.Interest feature vector,constructing scholar’s dynamic interest model.Then,based on the scholar’s dynamic interest model,the scholar-dynamic interest matrix is constructed,and the K-means clustering method is used to cluster the scholars with similar research interests.Then,based on the scholar clustering model with dynamic characteristics,a scholar recommendation model was constructed by integrating the two dimensions of scholars’ scientific research ability and social attributes.The scholar’s scientific research ability attribute selected the scholar’s h index and academic achievements as measurement indicators,and the social attributes were selected It is the frequency of academic cooperation as a measurement indicator.Finally,the Comb MNZ score integration strategy is used to integrate the scholar’s ability attribute and social attribute measurement results to obtain the scholar’s final recommendation value.And verify the model proposed in this paper with the real data set of "Baidu Academic" to prove the feasibility and effectiveness of the model.