In online discussion platforms, human facilitators are introduced in order to facilitate the discussions to proceed smoothly and build consensus efficiently. However, problems such as human bias and scalability are becoming critical with increasing sophistication of these online discussion platforms. In order to address these problems, online discussion facilitation support becomes more and more essential. Towards this end, in this paper, a novel case-based reasoning (CBR) based online discussion facilitation support approach, which consists of a case definition method and a case retrieval algorithm, is proposed to support online facilitation in large-scale discussion environments. The proposed approach models the online discussions using the issue based information system (IBIS) discussion style, where complex problems are modelled as a conversation amongst several stockholders. In the proposed approach, discussion cases are generated and retrieved based upon the structure features of their discussions. The experimental results show the proposed discussion case generation approach is able to reflect more precise discussion features than those approaches that are based only on the quantitative features, and the ability of the proposed case retrieval algorithm to retrieve the most similar case from the case base.