Detection of opinion leaders: Static vs. dynamic evaluation in online learning communities

Abstract

Opinion leaders play a critical role in maintaining the functioning of online communities. This study aims to detect opinion leaders in online learning communities by evaluating the influence of users within the community. We use Baidu Post Bar’s Python learning community as an example and employ the catastrophe progression method to statically evaluate the influence of users in three dimensions: user creativity, user posting quality, and user online interaction. Based on this, we introduce the dual-incentive control line to dynamically evaluate users' influence from 2016 to 2020 regarding speed change characteristics, thus scientifically detecting opinion leaders in online learning communities. Compared to the static evaluation method, the results show that our proposed dynamic evaluation method can more effectively reveal the dynamic development trend of users' influence, thus accurately detecting opinion leaders. Moreover, this “invisible” development trend is fully reflected in the setting of the dual-incentive control line.

Keywords

• Online learning communities
• Opinion leaders
• Influence evaluation
• Social network analysis
• Catastrophe progression method
• Dual-incentive control line