The paper describes an agent-based model of research collaboration in collaborative tagging system in order to develop an effective collaboration support system for scientific research communities. Based on a survey on social software such as collaborative tagging for the purpose of collaboration on the web, at first we analyze requirements for the social software as a collaboration support system for scientific researchers. Next we apply agent-based methodology to develop an agent-based model of research collaboration in collaborative tagging system that satisfies the requirements. Then we validate the model by comparing its functionalities to the existing collaborative tagging systems. Finally we conclude the paper and mention our future work.