Digital Bionics and Heritage: Achieving Text Usage Habit Replication through Semantic Recognition Technology and Large Language Models

Abstract

With the growing number of individuals living alone in Taiwan, loneliness has increasingly become a significant social issue. Among people aged 20-44, over 50% are unmarried, and single-person households now exceed 3 million. The pain of losing a loved one is often difficult to overcome, and many people long to maintain an emotional connection with their departed loved ones as a source of strength for facing the future. In response to this, the "Digital Bionics and Legacy" system was developed to help users preserve and revisit memories of lost loved ones. This conversational AI tool allows users to upload chat records with the departed, enabling the system to train a model that simulates the communication style of the lost loved one. Through dialogue generation technology and sentiment analysis, users can “converse” with them, alleviating emotional pain and supporting the healing process. This paper introduces the system as a WebApp, allowing users to upload chat data to the backend. The system then uses machine learning, natural language processing, and sentiment analysis, to replicate the linguistic style, emotional nuances, and conversational habits of the target person, creating a highly realistic interactive experience. This system not only helps users reminisce about loved ones but also has potential applications in fields such as psychotherapy and literary creation, providing comfort and companionship as users revisit past conversations.