Association rules mining for culture modeling

Abstract

The difficulty to predict the human’s behavior has caused the need to learn cultural differences between peoples. Although culture is one of the concepts that are difficult to define, best learning these differences depends on the culture modeling quality. In this paper, a new culture modeling is proposed to facilitate the learning process. This modeling allows to generate the frequent cultural characteristics in each region and extract cultural association rules. We create benchmarks of culture, based on a survey accessible from the web. The cultural datasets are analyzed with the Apriori algorithm to extract frequent attributes values. The obtained results show similarities and differences between cultures. Based on these results, we pass to the construction of association rules and their confidence.