|
From: Samuel L. B. <sa...@li...> - 2001-04-19 20:17:32
|
All - Here are my initial thoughts on the relationship between W3CVB and the Galaxy Communicator infrastructure. Over the years, I've personally reviewed many of the W3CVB documents, and none of them have alarmed me from the point of view of the Galaxy Communicator effort. It's important to remember that W3CVB and GC are trying to do very different things on a number of dimensions: - First, the W3CVB proposals are all proposals for data description. None of them address the issues of infrastructure connectivity or the issues of service definitions. So it's perfectly conceivable that a Communicator-compliant server might also be completely compliant with the relevant W3CVB standards proposals. - Second, the W3CVB proposals are attempting to standardize current technology for the purpose of current market exploitation, while GC is attempting to support a development infrastructure for the purpose of dialogue research and future market insertion and exploitation. So we can expect that there are capabilities which W3CVB won't envision (such as mixed initiative dialogue) which are very important from the point of view of GC. With that in mind, I'll try to address Jim's three questions. Some of these questions are program-level questions, rather than technology-level questions; with apologies to the non-Communicator participants listening in, I'll address those dimensions as well. 1. Is the proposed solution helpful or hurtful for Galaxy Communicator work? I believe that some of the W3CVB standards are very relevant to Communicator, such as the speech synthesis markup standards. Others, like VoiceXML, aren't compatible with GC goals like mixed initiative dialogue, but shouldn't be expected to be, since W3CVB is a near-term effort. Additionally, since the work in mixed-initiative dialogue isn't ready to be standardized yet, the appropriate thing to do in my mind is to use these near-term proposals as a baseline so that we can be more responsive to the established community when it comes time to propose a more flexible standard. 2. Should DARPA through MITRE (and MIT, IBM, et. al.) have an input? Everyone on this mailing list ought to be reading the W3CVB documents. The W3CVB effort is a broad effort with a good deal of industry support. For MITRE's part, we've provided considerable feedback to the W3CVB committee on their proposals already; I'm hoping that others in the program have as well. 3. What issues are key for the Communicator program to follow? For a dialogue research program, the importance of standards is unclear. Plug and play at the message level is useful, but any well-documented message set ought to be readily digestible by a Communicator developer. Nevertheless, I think we ought to start a conversation about these things for more well-understood services, since these efforts are complementary to W3CVB. Plug and play at the data level can also be useful, but given the cutting-edge nature of the Communicator work program, I think it's more important for us to digest the appropriate proposed standards rather than proposing standards ourselves. If there's interest, I'll attempt to summarize the points of similarity and contrast between the current W3CVB documents and the Galaxy Communicator effort in a subsequent message. Sam |