Pervasive Artificial Intelligence Research (PAIR) Labs

Natural Language Application

Pervasive Artificial Intelligence Research (PAIR) Labs

Natural Language Application

If you are interested in the technical cooperation, please feel free to contact us.

Professor Ching-Han Chen

NCU

Core Technologies

CDNA, a natural language based dialog navigation agent , which use a cell-inspired efficient software framework, is implemented for blind navigation:(1) conversational navigation service, (2) real-time path planning and path modification, and (3) location guide and place evaluation. CDNA integrate a deep learning NLP engine that use Google speech recognition engine, real-time text inference tasks, including topic generation, topic hints, semantic disambiguation and classification, analog production, and blind-guided situation-oriented reasoning. The chinese POST and segmentation of dialog system use Stanford Parser, and the concept representation mechanism of dialog use common-sense database data from ConceptNet and E-HowNet.

Professor Hsin-Min Wang

Academia Sinica

Core Technologies

Our research goal is to develop methods for analyzing, extracting, recognizing, indexing, and retrieving information from audio data, with special emphasis on speech and music. Currently, we focus on automatic speech recognition, voice conversion, text-to-speech synthesis, speaker recognition, speaker diarization, and spoken langauge identification.

Professor Wen-Lian Hsu

Academia Sinica

Core Technologies

Developed components including the “Facebook of Words” knowledge base, sentence reduction method, statistical principle-based approach (SPBA), and natural language scripting engine. We can simultaneously handle word segmentation, parsing, named entity recognition, dialogue generation and other important natural language processing tasks. It has been successfully used in dialogue agents for primary school math problem solving and financial information or other professional domains. The effectiveness of our systems are verified by users from industry, academia, and the general public to be outstanding.