A cultural knowledge-based method to support the formation of homophilous online communities

We propose a three-step method to identify people in social networks sites (SNS) who are talking about the same topics, even though they may be from different cultural backgrounds. Our method uses a cultural knowledge base from the OMCS-Br project to normalize cultural differences and find common interest among users based on statements they make various topics in a SNS. We evaluated three initial phrases that were used to search for sentences in a large social network using the cultural translation; we found that 81% of the retrieved sentences were judged to be related to the initial phrases. Thus, we have evidence that cultural normalization can support finding people talking about the same topic in a SNS even when they have different ways of saying the same thing. We believe that these culturally translated similarities can be used in a recommender system to contribute to the formation of homophilous online communities.