Finding Interdisciplinary Users On Social Media And Its Application

Cross-links and penetrations between different disciplines are essential features in the modern era.With the further investigations over the natural phenomena,scientific problems are becoming more and more complex,and require more cooperation between multiple disciplines and domains.Such disciplinary knowledge integration can not only solve massive problems,but also promote progress in emerging and new fields.However,due to the lack of interdisciplinary researchers,some interdisciplinary projects are difficult to been financed,as they may be assigned to the panel or reviewers who are ill-equipped to look through all parts of the project.This is extremely unfavorable for the development of interdisciplinary research.Therefore,this paper aims to seek users who have experience in research spanning multiple fields.With the development of Web 2.0 technology,social media,such as ResearchGate,Mendeley and ScienceNet,has gradually become the platforms for scientific communication among scholars.On these social media platforms,teams or individuals coming from different disciplines often communicate with others on topics of common interest(such as publishing blogs,adding friends,making comments and recommendations).This brings new opportunities for the discovery of interdisciplinary users on social media,so this dissertation researches on seeking users who span multiple fields.ScienceNet is a Chinese-scientific social media co-sponsored by the Chinese Academy of Sciences and the National Natural Science Foundation of China.Because ScienceNet is rich in academic information,we use it to find interdisciplinary users in the five disciplines,namely,Library and Information Science,Computer Science,News and Media,Higher Education,and Biology.There are two types of scholarly communication on ScienceNet,including the structure of blogger's social networks and the content of the blog,so we seek interdisciplinary users from these two perspectives respectively.Focusing on interdisciplinary user discovery,this paper includes the following four aspects:(1)From the perspective of the topic to find interdisciplinary users,this paper firstly constructed domain corpus with abstract and keywords of academic literatures crawled from CNKI.Then,we classified blogs according to Labeled-LDA.Finally,we determined the subject of blogs based on the naive Bayesian classification results,and then the subject attributes of bloggers were divided and interdisciplinary users were found.(2)In the process of applying community discovery algorithm,we find that multi-label propagation algorithm(COPRA,Community Overlap PRopagation Algorithm)is unstable.When a node belongs to multiple neighbor communities and its belonging coefficients are equal and less than the set threshold parameter,multi-label propagation algorithm may randomly choose one from these neighbor communities,which seriously affects the algorithm performance.Hence,this paper proposes an optimization of multi-label propagation algorithm based on contribution degree,namely COPRA_CD(COPRA based on Connecting Degree).The tests on real-world networks and synthetic networks showed that COPRA_CD is validity and stability.Then COPRA_CD is applied to the friend social network of science bloggers in the field of Library and Information Science,and we found out the overlapping communities effectively.(3)From the perspective of the structure of social networks to find interdisciplinary users,this paper firstly uses COPRA_CD algorithm to find overlapping users on the blogger's friend network,and then constructs discipline phylogenetic tree for each blogger according to their distribution of friends' disciplines,and finally determined interdisciplinary users by interdisciplinary distance and the number of friends in the field.(4)In order to effectively integrate interdisciplinary users discovered from the perspective of topic and relationship,we design an interdisciplinary user recommender system.This system can simultaneously recommend interdisciplinary users based on discipline orientation and discipline knowledge.In addition,this paper also designs a directional questionnaire to verify the reliability of interdisciplinary users we found,and also testes the accuracy of the recommended users by manual test.It proves that our method has a good credibility.In short,this paper not only enriches the methods of interdisciplinary research in theory,but also benefits the supplement of interdisciplinary user database.