Data-driven student homophily pattern analysis of online discussion in a social network learning environment

Yen An Shih, Ben Chang, Jonathan Y. Chin

Research output: Contribution to journalArticlepeer-review

3 Scopus citations


On social network sites, students establish relationships with peers and teachers, and they prefer to foster friendships with people who exhibit similar characteristics to them. However, the rapid changes and complications of information sharing and social networks creates difficulty in understanding students’ homophily in social learning activities. To fill this research gap, this study explored student homophily in student–student and student–teacher relationships in online discussion activities on a social network site named The research involved an analysis 494 users (5 teachers and 489 learners). We adopted text-mining approaches to extract keywords from discussion content and conducted social network analysis to explore students’ relationships and interactions. Discussion keyword similarity and interaction frequency were evaluated as factors of student homophily in each social tie. We further applied two-step cluster analysis to reveal patterns of student homophily in student–student and student–teacher relationships. The results indicated the feasibility of two student homophily factors for use in homophily evaluation. Three student homophily patterns were identified in both student–student and student–teacher relationships. Finally, a discussion was provided on the implications of student homophily phenomena, and implications were detailed for each homophily pattern from the perspectives of origins, development, students, teachers, and learning collaboration.

Original languageEnglish
Pages (from-to)373-394
Number of pages22
JournalJournal of Computers in Education
Issue number3
StatePublished - 1 Sep 2020


  • Clustering analysis
  • Social network analysis
  • Student homophily
  • Text mining


Dive into the research topics of 'Data-driven student homophily pattern analysis of online discussion in a social network learning environment'. Together they form a unique fingerprint.

Cite this