Semantic Sensor Network Incubator Group Teleconference

02 Mar 2010

See also: IRC log


+26431aaaa, Holger, kelsey, Payam, michael, laurent_oz, kerry, +1.716.688.aadd, +1.937.775.aaee, krzysztof_j
Manfred, Arthur, Kevin


<trackbot> Date: 02 March 2010

<Danh_DERI> zkim,+25431aaa is me

<krzysztof_j> http://www.w3.org/2005/Incubator/ssn/wiki/Generate_Meeting_Minutes#Scribe_duties

<kelsey> http://www.w3.org/2008/04/scribe.html

<Holger> http://www.w3.org/2005/Incubator/ssn/wiki/Generate_Meeting_Minutes#During_the_meeting

<Holger> ScribeNick: kerry

<Holger> http://www.w3.org/2005/Incubator/ssn/wiki/images/8/8f/SSN-XG-Semantic-Markup-3March2010.pdf

<scox> Some interesting (housekeeping?) news - pre-DIS review of ISO 19156 (Observations and Measurements v2.0) has been completed. Will be passed from TC 211 to ISO Geneva any day

semantic markup technique options: presentation by Laurent

Laurent: definition -- extended to suit rdfa: points to any type of thing that can be an element in an ontology
... rdfa can do this -- not just tbox material
... e.g . xml, service descriptions (wsdl), xsd, REST services, html, even rdf itself
... 3 types of techniques: 1. link to content managed elsewhere, (as xlink in ogc) : "normal" annnotation that might be "semantic"

<scox> Note that OGC is moving from URN to http URIs for system identifiers. e.g. see http://www.opengis.net/def/ from where you can navigate to parallel GML and RDF representations of 'definitions' or 'concepts'

laurent: 2. annotate a service description e.g. sawsdl or hrest, lifts into rdf
... 3: process the content to add annotation for rdf in html or ontology

<scox> The RDF was generated from the GML using a simple XSLT.

laurent_oz: awkward mixture of xml, html,rdf: confusing and difficult for validation tools

<scox> The RDF was generated from the GML using a simple XSLT.

laurent_oz: example of lifting data:(method 2) . Annotation refers to a "lifting script" usually xslt, a W3c recommendation?. Can be done throughout the fil, but needs user scritps to transalte throughout to rdfa
... in some places is easy

scox: butt in -- refer to chat comments

laurent_oz: e.g annotation using sawsdl liftng script (does something different - i missed it). The annotation links the wsdl or xml to ontolgy class as a "model reference".
... there is also a lowering script. And you can translate the xml itself into rdf. could move *all* the wsdl content into rdf this way

laurent_oz if you want to transalte a resful service (html) its not the same: e.g. 2 ways -1 microformats or 2 rdfa

laurent_oz (aside) yes , the "hrest" approach is used by SA-REST

laurent_oz: if you move all the service definition into the rdf form then you use an existing service ontology such as sa-rest or wsmo-lite

amit: sa-rest has evolved an can be used for any resources, not just services -- should also be considered. small differences with h-rest.

lamit: sa-rest is alive and is an altertantive to sawsdl

<scox> How can I stop the IRC from kicking me off?

laurent_oz: (accelerating_) lifting from html to rdf : has defined annotation for just about everything
... lifting xml to "semantic urns" : use xlink ogc-like to produce the rdf result, implies conventions for use of xlink

<Holger> Simon, error message is 'Ping timeout' - probably a network problem?!

laurent_oz: refer to AOW paper : rdfa has set the bar higher because i can cover all of OWL and can be used by developers in advance: not required scripts by users
... xlink in ogc is used by every other schema: but rdfa can be in many more places than xlink
... see short comparison in aow paper of xlink vs rdfa (rdfa is more powerful)
... use of xlink on ogc: can only be used in leaf property elements - something like an include mechanism -- tells you to go elsewhere for more
... matching the ogc xlink convention and semantic web needs is tricky as you can't tell the pupose of the xlink
... ogc usually uses element and attribute simply -- but attributes are hard

<scox> I don't understand if the comment that xlink can only be on props is meant to highlight a limitation?

laurent_oz: example:xlink usually points to an ontology instance (a urn) . kyzystof has a varaint using xlink and sawsdl together

<scox> In fact the correct comparison should be with RDF - OGC XML (GML) documents are isomorphic with RDF by design; xlink:href == rdf:resource

laurent_oz: this completes summary of techniques:

scox: (faded out) ..was developed as an xml rep of rdf as it was in 2000. whether links should point to classes or individuals is not clear in rdf either

laurent_oz: agreed. using xml and ontolgy languages together may be breaking new ground here
... what should we deliver for semantic annotation?
... focus on ogc swe standards: xg needs to liase with other groups w3c web service activity) as is not a striclty ssn issue. take it to the TAG?
... may be better new ways inspired by rdfa -- beyond the scope of the xg.
... want to use OGC and provide guidance on using the ontology. Hard to write a w3c doc addressed to OGC community. should be an ogc discussion paper?
... lackof semantic markup examples out there and in the ssn-xg

<scox> SHould be OGC 'Best Practice' (policy) not DIscussion Paper (only FYI)

laurent_oz: wish-list:if you use sawsl model ref and xlink at the same time: need help. do we need to (lost it..), refer krz's xomments last week about sapience

Krzysztof_j: how would a new ogc dis paper differ from the one laready out there?

laurent_oz: more detatil: e.g what is allowed and what is not. the w3c side brings a different view.

<scox> There is a big lack of xlink implementations in the world

<scox> xlink is a failed standard

<scox> The problem is: even though it has failed, it is needed!

krz: would add to confusion -- maue and cory's versions/appers. do not want another paper -- just focus on sensors and build some experience. we don't know how to use them from experience

laurent_oz: we have to ensure that our output reaches the ogc community

<scox> But remember that OGC is rooted more in structured-data, so XML Schema validation is seen as an important tool.

laurent_oz: wrapup : more examples of semantic markup (as in the draft final report - cory) sawsdl and model reference at same time. are there other examples like this? is this a new use unlike current ogc and sawsdl approaches
... is it possible to build validators for such a standard
... do we need model ref and domain ref annotation ? need more examples of needs to drive this.

holger: ask scox to summarise his interjections

scox: can say why ogc is done this way.: gml and followers were developed as rdf before rdf in xml stabilised. also lean towards object modleeing and validation becuase work fom structured

data unlike semantic web community (e.g. open world assumptions)

scox: own work is encouraging ogc to adopt more sw technologies. will move away frm gml and xlink for "semantic " things -- see my link to ogc website

<krzysztof_j> +q

laurent_oz: hard to find doco on where ogc is going with semantics

<scox> I'm pushing OGC to be more flexible, and use RDF technologies where appropriate (e.g. defs)

laurent_oz: 2 approaches to updating ogc tech 1. update nothing but do it outside e.g.grddl. 2. new ogc services to do the semantic stuff

<scox> but stick with XMl technologies for structured data

<scox> There are some interesting grey areas in between - e.g. CRS definitions, which play the role of semantic resources but are highly structured in terms of the information they need to provide

laurent_oz: acknowledge change in the way things are done now -- uing rdfa like the rest of the world is going

scox: agree guidance is needed: through compelling examples

<scox> isn't rdfa starting from the premise that the base document is HTML?

krzystof_j: in 3 years we will be using sw stuff all the way? maybe "annotation" will be an outdated idea

<scox> What if the base documents are _data_ (XML, not web pages)

laurent_oz: agrees, maybe. evolution for all standards orgnaisations , not just ogc

<scox> do you see rdfa replace (failed) xlink?

<Payam> +q

laurent_oz: yes rdfa is mainy for html but it changes the level of automation : has really superseded all the previous methods based on many use cases

<scox> q

krzysztof_j: thin if linked data as everything: semantics plays a much more major role, but don't want to buy in to the ontology of he person who built the data: argument for spearating the semantics from the the data

payam: confused -- are we trying to annotate sensor data? descriptions and o&m. we should start there. integraqtion with others comes later

<scox> Linked data will not make structured data go away

laurent_oz: both annotate data and annotate a sevice to produce semantic-rich data. these are mixed.

<scox> it just makes it visible in the web

payam: anotatng a service is not our business -- only annotating the data as i see it

laurent_oz: we need to clarify this objective..... both uses cases have been discussed in this group

<krzysztof_j> isn"t linked data structured data?

payam: wants to annotate data: services are out of scope

holger: thanks to laurent for presentation.

<scox> My apologies for next week - I'll be at OGC TC in Rome

<scox> Linked data is usually interpreted as RDF

<scox> and while RDF is 'structured' it is not easily validatable

kerry: has put phenoet architeture on the wiki under "communities" link
... plan for meeting times - will go in email

holger: close and thansk.


Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2010/03/02 21:16:39 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.135  of Date: 2009/03/02 03:52:20  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Found ScribeNick: kerry
Inferring Scribes: kerry
Default Present: +26431aaaa, Holger, kelsey, Payam, michael, laurent_oz, kerry, +1.716.688.aadd, +1.937.775.aaee, krzysztof_j
Present: +26431aaaa Holger kelsey Payam michael laurent_oz kerry +1.716.688.aadd +1.937.775.aaee krzysztof_j
Regrets: Manfred Arthur Kevin
Found Date: 02 Mar 2010
Guessing minutes URL: http://www.w3.org/2010/03/02-ssn-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

[End of scribe.perl diagnostic output]