Institute of Information Science, Academia Sinica

Various devices such as smartphones and in-car navigating systems are incorporating spoken dialogue systems (SDS) as personal intelligent assistants. Prior studies mostly focused on how to parse the natural language inputs into organized semantic concepts, where such concepts and their structures should be manually created. However, the need of human annotation results in high cost and poor scalability in system development, and the SDS supported domains/topics are limited. Therefore, this talk focuses on improving generalization and scalability of building SDSs by automatically inferring semantic knowledge and learning structures from unlabeled dialogues to reduce human effort. Furthermore, with the automatically acquired knowledge, we develop a spoken language understanding (SLU) component to unify automatically acquired knowledge, incorporate multi-modality, decode semantics, and predict user intents simultaneously through a matrix factorization approach. The experiments show better and deeper understanding performance with consideration of hidden semantics and personalization, presenting the great potential of a self-learning intelligent assistant without human annotations.

Yun-Nung (Vivian) Chen is a current PhD candidate in the Language Technologies Institute of School of Computer Science at Carnegie Mellon University. Her research interests include spoken language understanding, user modeling, speech summarization, information extraction, and machine learning. She received Best Student Paper Awards from IEEE ASRU 2013 and IEEE SLT 2010 and a Student Best Paper Nominee from INTERSPEECH 2012. Chen earned B.S. and M.S. degrees in Computer Science and Information Engineering from National Taiwan University, Taipei, Taiwan in 2009 and 2011 respectively, and an M.S. in Language Technologies from Carnegie Mellon University, Pittsburgh, PA in 2013. (http://vivianchen.idv.tw)