Knowledge Sharing in Peer-to-Peer Online Communities: The Effects of Recommendation Agents and Community Characteristics

Knowledge sharing (KS) is generally performed in a centralized fashion, through a knowledge repository. However, the centralized knowledge management systems may sometimes cause problems in sharing knowledge. One potential solution is to conduct KS in a decentralized network supported by peer-to-peer technology. While extensive research has been done to address the flaws of current reputation-based feedback mechanisms for promoting knowledge diffusion in peer-to-peer communities, some serious drawbacks of feedback mechanisms, namely dimensionality and interaction-dependency effects, have been overlooked. Using an agent-based simulation, this paper examines these effects while also considering characteristics of the community. Results of our simulation show interesting findings, for example, that negligence of knowledge dimensionality and interaction-dependency in designing reputation mechanisms has a major negative effect on community knowledge performance. Further, more complex knowledge adversely affects community-level diffusion outcomes whereas greater network density improves them.