A Study on Judging Phonetic Similarities for Trademark Searches by Focusing on Enhancing the Search Precision of Korean and English Character Trademarks

Trademarks are marks (symbols, letters, figures, sounds, etc.) used to distinguish one’s goods from those of others. Trademark rights are a type of intellectual property, that protect trademarks and maintain reliability, thereby safeguarding consumers’ interests, and contributing to industrial development. To register a trademark, assessing its identity or similarity with previously applied trademarks is necessary to examine its distinctiveness, particularly since the phonetic similarity of trademarks plays a crucial role. The importance of a textual trademark pronunciation search system, that determines phonetic similarity has been emphasized to maintain the trademark system’s efficacy. However, as trademarks are composed of both Korean and English characters, difficulties arise in converting the pronunciation of one language into another for assessing similarities in trademark pronunciation search systems. This study proposes a search methodology to address these issues. It focuses on determining the phonetic similarities of textual trademarks composed of Korean and English characters. It utilizes AI-based transliteration and phonetic transcription models on trademarks written in Korean or English, to propose a methodology that unifies Korean and English into a single pronunciation, thereby converting them to closely match the phonetic similarity criteria for textual trademarks. Additionally, it verifies the performance of this methodology through actual cases of trademark phonetic similarities for confirming its effectiveness.