Question Answering on Smartphones
Description of the Invention
Voice based assistance on smartphones such as road directions; entertainment information, making reservations, and various phone functions can all be naturally enabled by a general purpose Question Answering (QA) search engine that answers all questions. In order to do this, we face several major difficulties: (a) Improve voice recognition for diversity of speakers in noisy environments; (b) Design a general purpose
QA search engine. UW team has implemented (a) and (b). Any smartphone voice personal assistance systems which bypassing these two steps are building on shaky grounds.
UW system prototype has already been built as a proof of concept. It works and can be demoed. The technology involves a new technology for correcting user speeches for QA, so that it overcomes the difficulties of speech variations and (not too) noisy environment – this is a key for the technology to succeed. The system also involves new technologies of understanding the queries, and methods of searching the internet data and databases so that queries can be correctly answered, and a new translation technology for cross language search.
Advantages
Ø Improvement of current speech recognition technologies by 30% for non-native speakers, and 10% for native speakers, in the QA domain.
Ø Advantage over iPhone 4S Siri:
(a) Our system do QA, Siri does not.
(b) We significantly improve voice recognition for QA, Siri does not.
(c) We do cross language search. Nobody else does. Better search technologies using a general theory of information distance which invented by the researcher (and accepted as a standard measure of information in the world).
Ø Convenience for many applications on smartphones.
Ø First cross-language QA search technology: There is no QA system, such as Wolfram Alpha, in all languages. Our technology solves the problem by allowing QA search in all languages (demo in Chinese is available.)
Potential applications
Ø Personal assistance: calendar, appointment, location, any general questions, all by voice.
Ø Children games (Talking Tom, R2-D2)
Ø Q & A service, internet search
Ø Cross language, multi-language search, for Chinese market.
Development status
Ø Proof-of-principle is done with product demo
Ø Seeking industrial partners with mobile search and voice activated applications
Ø Studies for additional applications for different market sectors