Intelligent human-computer interaction
OVERVIEW
As artificial intelligence technology rapidly evolves, we aim to leverage AI solutions to help users solve problems efficiently and optimize their workflows. Our institute specializes in developing intelligent human-computer interaction technologies by integrating AI-driven voice, text, and generative AI capabilities. These innovations have led to solutions such as smart customer service chatbots, AI-powered document review, multimodal retrieval, and embodied AI companion robots, effectively enhancing user satisfaction and work efficiency.Intelligent human-computer interaction
CORE TECHNOLOGY
- AI Customer Service Chatbot
- AI document review
- Multimodal Retrieval Technique
- Embodied AI Companion Robot
AI Customer Service Chatbot / AI document review / Multimodal Retrieval Technique / Embodied AI Companion Robot
Intelligent human-computer interaction
APPLICATION STATUS
Intelligent Customer Service Chatbot: We have developed a multi-turn dialogue chatbot powered by AI question-answering and knowledge management technologies, delivering rapid response services across multiple channels. The solution reduces customer waiting times while enhancing customer service agent productivity. Our comprehensive solutions include text-based chatbots, IVR (Interactive Voice Response) call-steering systems, and outbound survey chatbots, which have been successfully implemented in Chunghwa Telecom's customer service operations and numerous external projects. This innovative technology has earned the Best Product Award of the 19th National Brand Yushan Awards, the TCCDA-CSEA Best Intelligent Customer Service System Application Awards (2019-2023), and the Patent Gold/Silver Award of the Taiwan Innovative Technology Exhibition (2020, 2022-2023).
AI Document Review: By integrating multimodal large language models and retrieval-augmented generation technologies, we provide AI-powered solutions for reviewing documents and contracts. These solutions are applicable to financial regulatory and legal documentation, significantly improving review efficiency and reducing manual review time. A proven application of this solution is the AI-assisted review of annual reports for publicly listed companies.
Multimodal Retrieval Technique: By leveraging multimodal retrieval technologies, this solution analyzes diverse content types, such as text, charts, and images, to provide highly accurate question‑answering and search capabilities. It can be applied to accounting knowledge retrieval, as well as retrieval-based Q&A for surveillance video and key traffic event segments, thereby significantly enhancing knowledge utilization efficiency and overall operational performance.
Embodied AI Companion Robot: Built upon large language models integrated with emotion recognition and long-term memory modeling, this solution enhances contextual understanding and interaction accuracy through real-time feature extraction from images and speech combined with vector-based retrieval. It supports diverse companionship and monitoring needs in intelligent care centers.