Cognitive Consistency Models Applied to Data Clustering

Data clustering concerns the discovery of partitions in data such that items from the same groups are as similar to each other as possible, and items from different groups are as dissimilar to each other as possible. The literature presents vast diversity on data clustering approaches, including systems that model the behavior of social individuals from different species. This work proposes a clustering algorithm that is reasoned upon the social theory of cognitive dissonance and the psychological theory of balance. We investigate whether psychological balance aware decision-making capabilities would affect the data clustering task. Partial results revealed the superiority of the proposed approach according to 5 out of 9 clustering quality metrics, when compared to other clustering algorithms.