Methods for Quantitative Evaluation of Individual Differences in Tactile Perception: Euclidean Distance, Cosine Similarity, and Their Hybrid Indices

Abstract

As tactile evaluation is an important assessment indicator for consumer product developments, the importance of objective tactile evaluation is increasing. However, individual differences among subjects in tactile research has not been sufficiently investigated. This paper focuses on individual differences in tactile perception and proposes a method for quantifying the similarity with a developed index. We compared three representative indicators, Euclidean distance, cosine similarity, and their hybrid index, for measuring the similarity in sensory evaluation data for tactile sensation of aluminum plates. Comparison of the conventional 2 indices revealed that opposite tactile sensations were observed, leading to the misjudgment of similarity between subjects. While, the proposed hybrid index provided better clustering of the subjects, resulting in extracting more accurate tactile factors than those obtained with the conventional indicators.