Corporations in Japan can theoretically be identified by their organizational names and addresses. However, when users input these strings online using Japanese, variations in the representation of names and addresses can make it challenging to determine their uniqueness. In such cases, manual reconciliation, often referred to as “record linkage” or “name resolution”, is commonly employed in practice. For. JP domain names, however, a method involving string normalization of organizational names and addresses has been implemented, enabling high-accuracy and real-time identification of organizational uniqueness, yielding practical results. In this study, we extended this approach using machine learning to enhance its general applicability and obtain promising outcomes. Specifically, we propose a method for identifying the same organization from string data, particularly for multi-script languages where character/word variations exist, such as Japanese and some other Asian languages.