Search This Blog

Thursday, March 18, 2010

New Models of Human Language to Support Mobile Conversational Systems

W3C has announced a Workshop on Conversational Applications: Use Casesand Requirements for New Models of Human Language to Support MobileConversational Systems. The workshop will be held June 18-19, 2010in New Jersey, US, hosted by Openstream. The main outcome of theworkshop will be the publication of a document that will serve as aguide for improving the W3C language model. W3C membership is notrequired to participate in this workshop. The current program committeeconsists of: Paolo Baggia (Loquendo), Daniel C. Burnett (Voxeo),Deborah Dahl (W3C Invited Expert), Kurt Fuqua (Cambridge Mobile),Richard Ishida (W3C), Michael Johnston (AT&T), James A. Larson (W3CInvited Expert), Sol Lerner (Nuance), David Nahamoo (IBM), Dave Raggett(W3C), Henry Thompson (W3C/University of Edinburgh), and Raj Tumuluri(Openstream).
"A number of developers of conversational voice applications feel thatthe model of human language currently supported by W3C standards suchas SRGS, SISR and PLS is not adequate and that developers need newcapabilities in order to support more sophisticated conversationalapplications. The goal of the workshop therefore is to understand thelimitations of the current W3C language model in order to develop amore comprehensive model. We plan to collect and analyze use cases andprioritize requirements that ultimately will be used to identifyimprovements to the W3C language model. Just as W3C developed SSML 1.1to broaden the languages for which SSML is useful, this effort willresult in improved support for language capabilities that areunsupported today.
Suggested Workshop topics for position papers include: (1) Use casesand requirements for grammar formalisms more powerful than SRGS'scontext free grammars that are needed to implement tomorrow'sapplications (2) What are the common aspects of human language modelsfor different languages that can be factored into reusable modules?(3) Use cases and requirements for realigning/extending SRGS, PLS andSISR to support more powerful human language models (4) Use cases andrequirements for sharing grammars among concurrent applications (5) Usecases that illustrate requirements for natural language capabilitiesfor conversational dialog systems that cannot easily be implementedusing the current W3C conversational language model. (6) Use cases andrequirements for speech-enabled applications that can be used acrossmultiple languages (English, German, Spanish, ...) with only minormodifications. (7) Use cases and requirements for composing thebehaviors of multiple speech-enabled applications that were developedindependently without requiring changes to the applications. (8) Usecases and requirements motivating the need to resolve ellipses andanaphoric references to previous utterances.
Position papers, due April 2, 2010, must describe requirements and usecases for improving W3C standards for conversational interaction andhow the use cases justify one or more of these topics: Formal notationsfor representing grammar in: Syntax, Morphology, Phonology, Prosodics;Engine standards for improvement in processing: Syntax, Morphology,Phonology, Lexicography; Lexicography standards for: parts-of-speech,grammatical features and polysemy; Formal semantic representation ofhuman language including: verbal tense, aspect, valency, plurality,pronouns, adverbs; Efficient data structures for binary representationand passing of: parse trees, alternate lexical/morphologic analysis,alternate phonologic analysis; Other suggested areas or improvementsfor standards based conversational systems development..." also W3C Workshops:

No comments: