Proposal of an Onomatopoeia Feature Table Using Cosmetic Reviews

Onomatopoeia is a general term for imitative and mimetic words. A situation or event can be expressed in different ways using onomatopoeic words. However, the expresser and reader may interpret the words differently because the usage of onomatopoeic words differs from person to person, especially in merchandise reviews. Therefore, it is necessary to analyze the features of onomatopoeia for interpreting these words accurately. This study proposes a method to characterize onomatopoeic words in merchandise reviews by analyzing their nature and features. The proposed method combines the following analysis techniques: cluster analysis and Latent Dirichlet Allocation for data categorization, correspondence analysis for determining the relationship between the merchandise category and onomatopoeia, and N-gram for searching for the co-occurrence relation between the characteristic words and onomatopoeia. Consequently, we find the features and differences for each onomatopoeia and construct an “onomatopoeia feature table”. Using it, a user can determine the difference in the nature of the use of onomatopoeia. The nature of use of the onomatopoeic word can also be correctly interpreted. Some characteristic words in the constructed onomatopoeia feature table were verified.