RE: question about including DCCI information in MMI events (fwd)

If an event carries sufficient evidence to identify the device (though not necessarily the particular instance), the approach being adoped by the Device Description WG may provide the access to "a priori" information regarding that device. In the most simple example, a User Agent header value may be sufficient evidence to enable a Device Description Repository to look up the relevant properties (as listed in one or more vocabularies, which in turn reference the common ontology for the purpose of semantic definition and disambiguation). This approach solves part of the delivery context requirement and has been the focus of the DDWG since its inception.
 
However, the general need for delivery context information extends beyond the "a priori" knowledge contained within a DDR. One also requires the dynamic information pertaining to the current device configuration, which may include user preferences. There may also be environmental parameters, such as the maximum available bandwidth, that are accessible only via interrogation of intermediary components of the communications delivery path.
 
In essence, there are many sources of delivery context information and these are being addressed by several technologies, including DCCI, the DDR, UAProf, JAIN, OMA DPE etc. It is regrettable that these technologies have not harmonized to a common access mechanism, but at least the introduction of a common ontology offers some hope.
 
Timing is a bit of a problem. The DCCI exists, but without an agreed vocabulary. The DDWG has almost completed a core vocabulary (of a priori knowledge) but cannot normatively reference an ontology because that ontology is still in flux. The OMA DPE is advancing while being aware of the DCCI and DDR API, but the latter is still some months away. Even so, the first version of the DDR API is likely to be a simplified derivative of the more complex interface under development. That complex interface could potentially evolve to encompass dynamic information (thus taking it beyond the scope of the responsible group's charter).
 
Access to delivery context information, especially in an increasingly diverse Web, is a big issue for the Web in general and is no longer confined to the specialist domains of mobile and embedded devices. While the work of groups like UWA, MMI, DDWG etc is contributing to a resolution of several aspects of the problem, I feel it is perhaps time for a pan-W3C approach to this issue.
 
---Rotan

________________________________

From: public-uwa-request@w3.org on behalf of Dave Raggett
Sent: Tue 05/02/2008 16:22
To: public-uwa@w3.org
Subject: question about including DCCI information in MMI events (fwd)




The MMI WG is interested in conveying delivery context information
as part of the events used in the MMI Architecture. I have asked for
more details on what they are looking for from the ontology and
think that this will be something for us to consider at our upcoming
face to face.

p.s. MMI and VB WGs are holding their f2f meetings at the same time
as us, but in Orlando, Florida as opposed to Seoul.

  Dave Raggett <dsr@w3.org> http://www.w3.org/People/Raggett

---------- Forwarded message ----------
Date: Tue, 5 Feb 2008 11:05:50 -0500
From: Deborah Dahl <dahl@conversational-technologies.com>
To: Dave Raggett <dsr@w3.org>
Subject: question about including DCCI information in MMI events

Hi Dave,

Multimodal interaction managers need to know the properties of the
devices they are communicating with. In a typical scenario, at the
beginning of a session a device might tell the interaction manager
its screen size, audio output properties, battery level and other
properties that would be relevant to the application. The MMIWG has
defined a set of life cycle events for communicating between
interaction managers and modality components in the MMI Architecture
specification [1]. We are talking about including device information
in relevant life cycle events, but we were wondering if the UWA WG
has some advice about how it might be represented. One option might
be to include an XML tree based on the Delivery Context Ontology
[2], maybe augmented with some user preference information. If there
was some kind of a device registry where device information is
stored, there could also be a pointer to that.

regards,

Debbie Dahl

[1] MMI Architecture: http://www.w3.org/TR/2006/WD-mmi-arch-20061211/
[2] Delivery Context Ontology: http://www.w3.org/TR/dcontology/

Received on Tuesday, 5 February 2008 17:31:22 UTC