Onomatopoeia Search System Focused on Attributes Based on Sensibility and Various Sounds

Abstract

In recent years, several onomatopoeias have been used in food advertisements as words to express texture. When consumers observe onomatopoeias in advertisements, they link them to food sales by associating them with the willingness to purchase a particular product. It is important to select onomatopoeias that appropriately express the product image. If the most suitable onomatopoeias that represent the advertised product or review text can be searched, it will be easier for consumers to understand the product better, and the advertisement, as a result, will be successful. The purpose of this study was to identify the attributes that characterize onomatopoeias from user reviews and to analyze the relationship between the sounds of onomatopoeias and each of their attributes to construct a system that can search for onomatopoeias by matching attribute images. In the proposed system, onomatopoeia words were extracted by arranging the onomatopoeias on a plane using principal component analysis and then clicking on this plane using a mouse. In addition, we confirmed whether the extracted onomatopoeias matched the images of the attributes through verification experiments.