See also: IRC log
<jeff__> scribe: Jeff Waters
<jeff__> scribenick:jeff
<jeff__> Hello and Welcome to the 9th meeting of the Decisions and Decision-Making Incubator. This is an interesting set of topics on today's agenda. Thanks for participating. As a reminder, we're here to explore how best to represent decisions with our semantic standards. In past meetings, we've defined a number of use cases and began an eXtreme Design approach to modeling utilizing the Neon Toolkit. We've had a number of good discussions along the way to ensure
<piotrnowara> Hello... do you still want me to say something about may Decision Ontology?
<jeff__> Hi, Piotr
<jeff__> Yes, that would be great if you can join us and talk about your work, since we missed that last time
<jeff__> Are you able to call in?
<jeff__> Hi, Eva
<piotrnowara> OK. I'll try to make a call
<jeff__> One of our goals is to ensure we learn from the other related activities and one of those is the Provenance incubator group. I've recently reviewed a number of the materials on their website and it has been very educational. I'd like to summarize a bit of what I think I've learned and we can follow up some more. First, let me say that the work being done by the Provenenace Incubator group is truly outstanding and impressive.
<jeff__> They are a great model for us. Their goal is to develop use cases which drive requirements, and they are reporting on both of those, and then from there to develop a state-of-the-art report and understanding of the provenance issues and a roadmap for semantic web technologies, development and possible standardization.
<jeff__> They have three major example use cases. The first is a Blog News Aggregator where provenance (how "things" came to be, i.e. the derivation of a thing) is needed to confirm validity of a source, to determine the license of content, e.g. photo, has the phonto been modified, etc.
<jeff__> The second use case is a Disease Outbreak where trust and validation of scientific research across domains is key. The third use case is a Business Contract where proof that work was performed in conformance with requirements needs to be determined
<jeff__> The second use case is a Disease Outbreak where trust and validation of scientific research across domains is key. The third use case is a Business Contract where proof that work was performed in conformance with requirements needs to be determined
<jeff__> marion: do we have references on this topic?
<jeff__> jeff: yes, in fact the provenance has a wiki
<jeff__> jeff: http://www.w3.org/consortium/activities.html
<jeff__> jeff: There you can find a list of all the working groups and incbutors, and links to everything you need
<jeff__> jeff: let's go to Piotr
<jeff__> piotrnowara: from the beginning, i had two use cases, one is modeling decision schemes describing decisions from formal legal, industry standards
<jeff__> pitornwara: the second use scenario is creating decisions based on experience gained from actual decision-making, individual decisions or generalized schemes
<jeff__> piotrnowara: we tend to use decisions relating to decision-making sometimes to the result, sometimes to mixture of the two. My own definition is to understand it as a purpose-oriented attitude
<jeff__> piotrnowara: I have prepared a brief outline of the decision ontology and I have share it on my website
<jeff__> piotrnowara: I will paste that in shortly and you can review that, my ontology would include a decider, purpose, motivation e.g. knowledge of facts of patient health, blood test results
<jeff__> piotrnowara: then there is realization conditions, what is needed for the specified actions such as patient agreement, possible use case for my ontology is say we have a health insureance company
<jeff__> piotrnowara: that wants to make services available to patients and hospitals, a patient could check whether entitled to specific treatment and provider could provide or not, and whether
<jeff__> piotrnowara: that is right or wrong. I am involved in projects and looking for collaborators. Do you have any questions?
<piotrnowara> http://code.google.com/p/decision-ontology/downloads/list
<jeff__> jeff: is this something you're working on with your company?
<jeff__> piotrnowara: no, on my own, but they are getting interested, but I'm looking for collaborators, it offers interesting opportunities
<jeff__> piotrnowara: I work on this after hours,
<jeff__> marion: I am working on an ontology of cognition, the interaction of language, culture and cognition, to identify when someone belongs to a terrorist group and to develop the components of cognition
<jeff__> marion: the decision work would be a subset of the ontology of cognition, and we have a research project to look at this and its impact on decision-making would be a follow-on
<jeff__> jeff: (jeff summarized that collaboration opportunities with piotr as a future member, or invited expert, are possible.)
<jeff__> marion: general concept is recursive subsets, without actually showing the ontological structure, we show what is a subset of what and this hints at the structure.
<jeff__> Eva: no specific comments, but interesting to look at marion's paper and learn what we can
<jeff__> marion: provenance is origin or source and I gather it has something to do with pedigree metadata
<jeff__> jeff: The use cases drive the provenance requirements, which come under three headings: Content, Management and Uses. Under content, the object of provenance needs to be modelled as well as the attribution (origin and responsibility), the process followed to derive the ojbect, the evolution and versioning, and justification of decisions. Under management, publication of provenance info, accessibility, dissemination control and scaleability are issues. Und
<jeff__> jeff: The group is developing the two major reports, one of which is completed in at least an initial version (requirements and uses) and the other (technologies) is underway. (Jeff continued to note how decisions can be seen as a subset of provenance, but there are some unique aspects of decisions as well)
<jeff__> eva: not too much progress on modelling side, I had some things I would investigate, I posted something about SPARQL constructs and it could be relevant and I added some links
<eblomqvi> http://www.w3.org/2005/Incubator/decision/wiki/Decision_Incubator_Tools/Collections
<jeff__> eva: just recently, I put info on modelling a collection. I promised I would look into best ways to model that. I will print the link in the chat.
<jeff__> eva: it is a short passage of text with a couple of links, I looked for other examples and found the SWAN ontologies that model these collections, lists, sequences and this is an alternative
<jeff__> eva: I also put a link to an RDF tutorial. However, I suppose we should talk more about how we want to use this. We should model in an ontology as opposed to the RDF functionality.
<jeff__> Eva: That's what I have to report
<jeff__> marion: I suggest you look at whatever allows you to search in the most efficient way
<jeff__> eva: I agree, if you want to refer and talk about the collections, better to model them in the vocabulary
<jeff__> eva: not a big deal, we can model it and not just one ontology we reuse
<jeff__> jeff: recommend Named Graphs interesting solution for how to handle reification
<jeff__> jeff: we've run out of time, I'd like to thank everyone for calling in. Please feel free to send to me any agenda items for next time and we'll continue our modeling efforts and addressing any issues we encounter along the way.
<jeff__> Here are a few links and items discussed at today's meeting:
<jeff__> jeff: Here are some interesting links to info on Named Graphs (basically assigning a URI to a set of triples, maybe even to a single triple, which is useful when you want to say something about the triple. Named Graphs appear to be a bit more elegant and appropriate than RDF reification). Also on the topic of Open Provenance Model (OPM) which is attractive primarily because of its elegance, in that it is simple in concept but especially with subclassing t
<jeff__> jeff: it appears capable of representing complex and meaningful derivations.
<jeff__> jeff: The thought is that Named Graphs are perhaps a good solution wherever reification is needed, for example when we want to say a certain triple represents the "question" of a "decision" (rather than have the "question" be merely plain text). Also interesting to see if OPM can be used to represent some of our decision use cases. OPM is intended to represent the derivation of basically anything, including derivation of decisions.
<jeff__> http://eprints.ecs.soton.ac.uk/18332/1/opm.pdf -- The OPM Core Spec
<jeff__> http://openprovenance.org/examples/opm-overview-for-xg-prov.pdf -- Slide presentation on OPM
<jeff__> http://www.ldodds.com/blog/2009/11/managing-rdf-using-named-graphs/ -- Summary of Named Graphs in context of RESTful access
<jeff__> https://www.usenix.org/events/tapp10/tech/full_papers/chapman.pdf -- Building on OPM for determining trust for decision-making (by some MITRE folks)
<jeff__> http://openprovenance.org/ -- Java toolkit for OPM
<jeff__> http://github.com/lucmoreau/OpenProvenanceModel/blob/master/opm/src/main/resources/opm.1_1.xsd -- OPM XML Schema
<jeff__> http://www.w3.org/2005/Incubator/prov/wiki/User_Requirements -- Good Summary Writeup on Provenance Incubator's 3 main use cases and the overall requirements
<jeff__> http://www.w3.org/2005/Incubator/prov/wiki/What_Is_Provenance -- Provenance Incubator's Page on Defining Provenance
<jeff__> (Here are a few discussion points that the scribe wasn't able to capture live, since he was talking.)
<jeff__> jeff: Regarding the agenda item "emergency information as linked data", Don and I are both interested in emergency management, including data represented in OASIS emergency management standards like
<jeff__> jeff: the Common Alert Protocol (CAP), Hospital Availiability (HAVE), the Distribution Element (DE), and emerging situation reporting and patient tracking.
<jeff__> jeff: These standards are a significant step forward for effective management of emergency information. As part of our effort to show the benefit of this work and extend the
<jeff__> jeff: interoperability, we're looking at enabling them also as linked data utilizing RDF. This will help to make this information easier to merge and query, and also be a linked dataset
<jeff__> jeff: which we can leverage, just like the other linked data that we're discussing in our open linked data use case. In other words, as we develop our decision model
<jeff__> jeff: to utilize open linked data for making decisions, we'll be able to apply the decision model to this emergency information because we'll have it in the same RDF format.
<jeff__> jeff: So this will be an excellent opportunity to extend our effort into the emergency management domain.
<jeff__> jeff: Regarding modelling progress, I need to pause from time to time to think about some of these modelling issues, like reification, collections, etc. and now that we have discussed and looked into best practices for handling these,
<jeff__> jeff: i feel more prepared to proceed apace. So I will have more to report on my modelling of my use case(s) next time.
<jeff__> Re best practices for ontology modeling, on the reification issue, as mentioned I found the Named Graphs interesting. The basic concept is to have a URI that can be a unique reference for any subgrah, even down to an individual triple.
<jeff__> jeff: With that URI, one can then make statements about that subgraph. This now creates a "quad" format, which adds the graph link to the traditional triple of subject-predicate-object. prov
<jeff__> jeff: An example where this become important for decision modeling is where we want to represent the decision "question", but we want to do it as a triple rather than the traditional string, so we have something machine understandable and represented as open linked data.
<jeff__> jeff: Now the "question" can point to the subgraph resource link in a simpler, more direct way than other mechanisms. We'll continue to explore this, but it looks like a potential best practice for reification.
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 Scribe: Jeff Waters Found ScribeNick: jeff WARNING: No scribe lines found matching ScribeNick pattern: <jeff> ... WARNING: 0 scribe lines found (out of 101 total lines.) Are you sure you specified a correct ScribeNick? WARNING: No "Present: ... " found! Possibly Present: Eva eblomqvi https inserted jeff jeff__ marion piotrnowara pitornwara scribenick You can indicate people for the Present list like this: <dbooth> Present: dbooth jonathan mary <dbooth> Present+ amy Agenda: http://www.w3.org/2005/Incubator/decision/wiki/Decision_Mtg_9_Agenda Got date from IRC log name: 22 Jul 2010 Guessing minutes URL: http://www.w3.org/2010/07/22-decision-xg-minutes.html People with action items:[End of scribe.perl diagnostic output]