Health Care and Life Science Interest Group face-to-face meeting
3-4 October 2006, Amsterdam, The Netherlands

HCLSIG F2F, first day

participants introduction, list of names and afflications to be recorded later

Eric, suggests to Joanthan to give a intro of science commons to the group

espcially in publication, in using leagal tags

Eric N, main goal is to generate public documents about this HCLS's activities, artifacts by the end of the year,

Objectives: focused, direct activities in task areas,

cross-task collabboration, dynamic interaction,

HCLS is focused on best practices, not standards, also identify areas to push to the other Semantic Web areas

put demos, senarios to illustrate issues, also demo at a large scale, how semantic web can work

generate documents to interact with other initiatives, to see what can be done with semantic web.

Important to specify the roadmap for the group in the remaining year, more focused activities

HCLS Timeline (draft),

1/07 timeframe: Draft best practices, (demonstrations - BioRDF, BioOnt, ACPP, DSE),( LS URI - Code Base, Tec Help?)

Collect specific proposals for demostrations,

define areas for best practices definitions

Joanne, want to know the specific charters of the group

Eric N: orginal charter is too big and vague,

Ivan: this is an interest group, not a working group,

working group is more for deliverables, timelines,

IG group is loosely connnected,

at W3C, need to produce working draft regularly, not scienctific publication,

states what we know, what we have for public comments,

Eric N, need to publish the work of this IG group in the near future

Ivan: publish "W3C working draft"
... Next Autumn, by the end of this IG's time, the charter has to be renewed, and may generate enough interests for other groups

Dirk C. suggests identifiy publishing date and task for each TF group at this meeting

Ivan: the deadline of the drafts should be ISWC06
... each group to "elect" editor of these publication

Eric: if any issues are important to healthcare and life science area, but not standarized yet, need to put in sceniaros, and requirements

Kei: SPARQL, on the table?

Chimezie: named graph, is on the discussion yesterday

Don: RDF vs. OWL, heavy weight, light weight semantics, alternatives

advises, what kinds of questions RDF and OWL each best suited for

Vipul: use of rules,

Eric: driving with scenarios,

Alan: between owl and RDF, always use OWL (personal view)

Eric: Day One Agenda

http://esw.w3.org/topic/HCLS/F2F/Agenda

Update on Semantic Web Activities By Ivan

Happening now: core infrastructures: SPARQL, RIF

SPARQL: progress quickly on recommendation stage, however, new technic problems identified, new public documents will be published soon, and goes back to the recommendation candidate phase again

a number of implementations, Jena, Oracle (might have one soon),

RIF group: complicated tasks, reconcile a number of theories.

F2F planed before ISWC

GRDDL group, might be very relavant to HCLS IG,

HTML -> RDF, standard way to abstract information, when annotate data, consider this technology

Former SW Best Practice and Implementation group, published a number of documents worth looking at

<mscottm> ^ standard way to _extract_ information embedded in microformats

last week just published "time ontology",

group reactivated again, last week first Tcon

continue working on modeling problems, patterns,

HCLS important to have bridges with this group,

two relavent areas of this group to HCLS:

RDFa: more vigor micro-format, can be seen as a serialized RDF (?) http://www.w3.org/TR/xhtml-rdfa-primer/

linked to GRDDL

Ivan: Microformat and RDFa have the same goal

New interest grooup education and out-reach, chaired by Susie, educations

Ivan: public perception on SW being complicated, only people with certain background can use, this group is to educate people on using simply SW technology to do their work

Coordination group, weekly Tcon, discuss issues raised in one group, that might be relavent to other groups

also coordinate W3C with other non W3C organizations

Eric is regularly attending this group to make aware HCLS issues to other groups.

an issue: in need of a system tools to bridge Rational DB and RDF

a number of implentations in IBM, Oracle, etc. but might have a workshop on this topic next year

issue: review and extension of RDF specification
... Uncertainty in RDF?

will be a workshop at ISWC, planning an incubator group on this topic

Eric N: lots of confussion around this topic, would be better to get clear use cases,

issue: need a layer between RDF and OWL lite,

so people don't need to have owl reasoner to use SW

W3C resouce constraints, so need to postpone creation of some interest groups

Joanne, Vipul, Eric: HCLS continuously provide use cases for these issues.

Kei: any interest in data-mining, structured data (? intension)

Ivan: W3C is not a research body, need to see if there is possibility to standardize, otherwise, the topic will not be on the agenda

Dirk: any work on workflow?

Ivan: no right now, might be in rule group,

Scott: how about web service?

Ivan: not SW activity, done at web service space,

just published SW annotation of WSDL

Scott: also wants to add semantic data on workflow

no mechanism to state semantic data on port

Short presentation from TFs

BIORDF group, by Susie

Review charter:

participants

Neuroscience focus: Heerogenous data, interfaces, disease docus

Tasks: http://esw.w3.org/topic/HCLSIG_BioRDF_Subgroup/Tasks

Publication: Semantic web meets e-neuroscience: an RDF use case

Report: Converting biological information to the W3C Resouce Description Framework (RDF): Experience with Entrez Gene

Ivan: collect all reports and publications made by IG groups and members on the HCLS page

BioRDF goals:

http://esw.w3.org/topic/BioRDF_Top_Level_Task

Ivan: a huge number of tasks, hard to manage at the IG level

susie: some tasks have been completed, need to make a "completed tasks" list

Eric N: initial stage, many people join in on topics they familiar with

now need to focus on a number of topics that yield results

<Joanne> Would it make sense to narrow to one disease focus? Also, it's hard to see what if any dependencies exist on the list of active tasks.

<Joanne> Focus Focus Focus

Susie: top priority is to generate a URI document, time to start building a demo

Ivan: plus other working documents, reports from the group

BIONT group review (Vipul)

http://esw.w3.org/topic/HCLSIG_BioRDF_Subgroup/

Work so far:

Use case for Parkinson's disease

A seed ontology for Parkinson's Disease based on the cullular and molecular biologist perspective

identification of pre-exisitng ontologies for "cross-link" into the seed ontology

colloaboration opportunities

<Joanne> (backup a bit) - question raised and will be discussed later, is tackling the URI issue something that HCLSIG should take on or something HSLSIG should push back to W3C?

thanks, Joanne

<Joanne> no problem!

BIONT and BIORDF

Collaboration with ACPP task force,

Next steps:

collaborations and real demo

Eric: map data instances to conceptual ontology

<Joanne> ACCP stands for what?

Vipul: two types of mapping: ont - ont (ACPP ont - RIM ont for example), and instnace to high level ont

<Joanne> I mean ACPP

ACPP: Adaptable clinical pathways and protocols

<Joanne> Thanks!

Ivan: ont management would be an issue, need to discuss with deployment group,

if the IG group puts out ont, need to consider how to maintain them

Vipul: if the ont from this group has values, it will be rolled out to other spaces and maintained.

Scott: Any OBO OWL Repository?

Vipul: intension to use OBO to host the ont.

Dirk: wonder if it is the goal for IG to produce ontologies

Vipul, Eric: only for the purpose of demos

emphasis on bridging pieces, no too concern about completeness

Kei: does OBO has data?

Scott: OBO only for ontologies, not for data

<vipul> BIONT Presentation is available at:

<vipul> http://esw.w3.org/topic/HCLS/OntologyTaskForce?action=AttachFile&do=get&target=AmsterdamF2FStatus.ppt

ACPP presentation- Helen Chen...

<Helen> ACPP presentation http://esw.w3.org/topic/HCLS/ACPPTaskForce/Telecons?action=AttachFile&do=get&target=ACPPProgress

scribe: trim down large charter: have a descriptive mechanism for medical knowledge, in order to support reasoning
... also, support personal information of patients
... handle variances in clinical pathways, ala rules, for decision support
... clinical use cases, radiology, coronary artery bypass graft (CABG)
... main constructs: patient states, tasks, preconditions, postconditions, goals, preferences

<Joanne> (for the record.... A clinical pathway is a patient-focused tool, which describes the timeframe and sequencing of routine, predictable multidisciplinary interventions and expected patient outcomes, for a group of patients with similar needs. From http://www.ottawahospital.on.ca/hp/dept/nursing/pathways/index-e.asp

<Joanne> OWL-S http://www.daml.org/services/owl-s/1.0/

Question regarding OWL-S: http://www.daml.org/services/owl-s/

scribe: OWL-S can be used as a process model-- caveat that it may not be supported next year when funding at UMD goes away
... need to include triples that may not yet be factual, but are interpretations
... important issues: exclusion and scoped negation, temporal concepts, weighted (probabilities) conditions, pre-position of knowledge body (evidence)

<Helen> Minor correction: It is SPL (Structured Product Label) not SPE

<Helen> http://www.fda.gov/oc/datacouncil/spl.html

Drug Safety and Efficacy - Eric

scribe: http://esw.w3.org/topic/HCLS/F2F/DrugSafety
... need to follow up on new SPF std (based on HL7 RIM)

URI resolution in ontologies - Alan Ruttenberg

scribe: What goes wrong with URLs
... content disappears, server disappears, content is too slow, too big, what format, info resource or not

scribe: LSID definition
...http-range14: using the result code; type 303 may point to metadata
... content negotiation, server offers list
... short statements from John Barkley, Matthias Samwald, and Phil Lord
... distinguish information from non-information
... Kei: identifiable and gettable, overlap, but not the same
... uri with hash
... how to avoid 'late solutions'; knowing more ahead of time
...alternative: using OWL to represent URI info
... InfoRes vs NotAnInfoRes

Unchanging InfoRes vs EvolveableInfoRes

<Joanne> Information resources can be obtained over the wire (gettable) - so a person isn't an information resource, but their webpage is (may be)

scribe: how to define versioning-- source authority dependent

<Joanne> proposes versioning semantics in an ontology because they are context depentent

<Joanne> argues we don't need named graphs to do this
....RetrievalMethod: Transform, SPARQL, WebService...
... multiple retrieval methods, company specific
... classes of InfoRes's

<Joanne> these examples may not be part of an upper (shared) ontology
....RetrievalMethod: NotAnInfoResto an InfoRes: seeAlso, foaf:homepage, subjectOf_
... do not share "bare URIs'
... how to resolve "clothed URIs" with RESTfulness?

<Joanne> can use "bare" URIs in private

<Joanne> some discussion now is - what is a bare URI? (jonathan rees)

<Joanne> Chimezie had some comments I couldn't capture

<Joanne> Kai - how do we manage/certify trust?

<dturi> I managed to get an account and join. Daniele

<dturi> Please let me know when I can call in.

<Joanne> Discussion about whether to address URIs or not.

<Joanne> Ivan is proposing various options

<Joanne> Susie suggests we create a pros and cons document

<Joanne> Alan (personally) doesn't want to be handed an LSID and doesn't want to support an LISD, suggests poll group and that he suggested another alternative

<Joanne> Advocates we said we see no reason to use LSIDs and leave it at taht

<jar> alan: lsids *are* our problem since we're likely to be handed them.. so we should take a side

<Joanne> Dirk - what makes life sciences so special that we need our own ID

<Joanne> yes, - you don['t get the history and I started scribing before you signed on, I was only suggesting skype so you could see what I'd aldready captured.

<Joanne> continuting .. Scott made note to Dirk that LSIDs are not exclusively for Life Science

<Joanne> Ivan says Henry has a meeting today or tomorrow with Sean Martin (missed some before)

<Joanne> Alan summarizing Eric - get consensus within this group and review before it goes out. although it may be public already

<ivan_> http://www.w3.org/2001/tag/doc/URNsAndRegistries-50

<Joanne> the document is public

<Joanne> Make sure those that are or have been vocal about LSIDs summarizing what we just said and send mail to them and to Henri.

<jar> ivan: argument at tag level is whether lsid's live in a separate uri space

<jar> ivan: and that's separate from the question of whether they should be used at all

<Joanne> Ivan asks how widely they are used today (LSIDs)

<dturi> we use them for all the entities in mygrid

<Joanne> Eric - NCI have had internal battles and have adopted them

<Joanne> Ivan - Interest group needs to take a stand: not really o.k., if possible use through http, etc. there must be a clear statement.

<Joanne> Eric - we've had our own internal disccussion and find it difficult to address all the issues

<Joanne> If we can't get through this at this meeting, then dpn

<jar> Joanne: likes alan's ideas...

<dturi> what's alan's idea?

kei: 80% but we need to be clear on meaning of versions, articulate use cases .

(to dturi: from this morning's presentation - express info about resources in owl / rdf)

<dturi> thank you

alan: action item: describe use case where lsid is too heavyweight

eric: make a draft and circulate it

<Joanne> put together draft of this idea, publish it at as a draft, accept and expect public comments

<Joanne> two issues - articulate cases where the URI ontology is necessary and where it isn't

<Joanne> 2nd - discuss it with LSID community

<Joanne> side benefit - will have a better mechanism for resolution

<Joanne> there are big pieces of what LSIDs do but they have insprired what is in Alan's proposal

<Joanne> Susie -not a big fan of LSIDs, but.... (missed it - check with susie)

<Joanne> Name is always the same in LSIDs - not same with ARK

<Joanne> thanks Daniele - feel free to jump in and scribe if you can hear and capture the arguments. I"m not that fast

<dturi> different names for the same thing in ARK - which might cause problems

<Joanne> LSIDs independent of domain name - an advantage

<Joanne> If Entrez published all their IDs as LSIDs and provided a resolver, then

<Joanne> if they changed (servers? organizatinos?) the resolver would still resolve and it would still work

<Joanne> Matthias - likes his own (more lightweight) proposal best

<Joanne> simple http url (Matthias cont'd)

<Joanne> Ivan - caveat - SPRQL doesn't exist - what he relies on is OWL, a recommendation and considered Stable. SPRQL is a draft. there is a difference,.

<Joanne> Need a draft of how Matthias prposal works

<Joanne> Chimezie - hard to follow discussion - wants to see clear examples

<Joanne> Helen - No need for LSID - URI model with HTTP seems to be enough

helen: preference for generic web architecture, current structure

<Joanne> Dirk - not involved in discussion and no technical opion. Philosophical - Simplist solution

<Joanne> Ray:.. Against LSID because LS isn't special, Also against more infrastructure

<Joanne> Joerg - No Opinion

kei: compare alan's thing to swmantic web services. alan agrees

kirsten like 'eat your own medicine' aspect

scott: lsid is a problem for interoperatbility wsp. w.r.t. owl
... look at alan's email of may 9'th, id's for entrez entries
... does the new proposal address the needs in that email?
... maybe redo these examples in the new context

<Joanne> Jar: not enough representation from those who have 1st hand experience with LSIDs

vipul: we need a separation of concerns - make requirements, then hand them off

<Joanne> Eric - abstains

<dturi> I agree that lightweight is better if possible

<dturi> There are concerns for us about versioning

<dturi> And about the use of foreign authorities

<dturi> I have heard nothing about the latter so far

<dturi> third party authorities

<Joanne> more?

<Joanne> an example?

<dturi> so that you don't need to own the data

<dturi> to say something about it

<dturi> mygrid could add metadata to lsids produced by ncbi

<Joanne> let me know when you're done.

<Joanne> ok done or ok you'll let me know

<dturi> finished

<Joanne> thank you!

<Joanne> Eric neuman's slide...

<Joanne> Depicts and example of his take from what Alan saids

<Joanne> Is Gene concept, Uni Gene ID? Is it a non-information resource?

<Joanne> Example is:

<Joanne> Information Resource (Data Record - (Entrez Gene))

<Joanne> has link labeled "seeAlso" to a Non-information resource (Gene Concept - UniGene ID?)

<Joanne> Data Record (MIPS) has a seeAlso link to Gene Concept UniGene ID? (Same node mentioned above)

<Joanne> Data Record (EBI) has a seeAlso link to Gene Concept UniGene ID? (Same node as mentioned above)

<Joanne> Summary - for nodes with 3 seeAlso relations pointing to Gene Concept UniGene ID? node

<Joanne> the "?" is part of the name of the node in Eric's daiagram

eric: should we (as a group) talk about conceptual models? e.g. what is a gene, or how do gene db's relate to one another

alan: it's fine to talk about "the thing that db record N refers to" without taking a stand on what that thing is

eric: wants to drive preparation of a document, before ISWC (nov 5)

eric is chief editor, with help from others

URL for Ontology presentation (Vipul)

<vipul> http://esw.w3.org/topic/HCLS/F2F/BioRDFBIONT?action=AttachFile&do=get&target=AmsterdamF2F.ppt

<Don> Parkinson's disease ontology

<Don> http://esw.w3.org/topic/HCLS/ParkinsonUseCase

<Don> Non-domain expert building of ontology

<Don> Look at information queries

<Don> Testing using queries potentially show design issues

<Don> Modeling as binary relationship or classes

<Don> Performance is a consideration in approach

<Don> Disease as dynamic process or a static class

<Don> Requirements driven

<Don> modeling constructs depend on the way you use your ontologies

<Don> obo wants to model "reality"

<kei> <Chimezie>it's a matter of perspective

<kei> <Dirk> model disease as a process

<kei> <Alan>disesae and diagnosis are not the same

<kei> <Scott>one needs to know the stage of disease

<Helen> this is no obsolute "reality", it is all about perspective and scope

<kei> <Chimezie>model something one is asking. It also depends on the needs and there is no point of model something one doesn't need

vipul: doesn't want to pay attention to barry smith until we know what real problem he's trying to solve (mistake fixed)

<kei> <kerstin>model common things

<kei> <alan>use case to model a variety of domains?

<Don> <vipul> queries are the use cases

<kei> <vipul>modelin subclasses vs., instances (t-box vs. a-box)

<Don> <alan> nominals break the boundaries between ABoxes and TBoxes

<kei> <vipul>performance is a determining factor

<kei> <Ivan>current reasoners are efficient with t-box (or large a-box and small t-box)

<kei> <vipul>owl-lite is similar to SQL

<kei> <alan>a-box is fine with pellet

<Don> <vipul> what level do you model?

<vipul> how to relate concepts at the right level

<Don> <vipul> trying to come up with principles including level to model relationships

<Don> <vipul> data and queries are important drivers of ontologies

<Don> <vipul> uncertainty plays a central role in science

<Don> <Dirk> there is a way to model uncertainty in OWL

<dirk> we need to consider carefully how to model uncertainty

<Ivan> reification in rdf is problematic

<Ivan> the current rdf semantics ignore reification

<alan> one can make statement about statement that is not true

<Ivan> named graph can be an alternative

<Don> <eric> not only uncertainty, also what are your axiomatic assumptions

<Ivan> one community is bayesian network and the other is fuzzy-logic-based

<vipu> multiple domains and ranges can be model explicitly by derived classes using union or intersection "and" or "or"

<Don> <vipul> handling default values of OWL properties

<Don> <alan> true defaults are not represented in OWL right now

<Helen> no rules are used in the ontology design

<Don> <vipul> we could propose rule based systems to solve these problems

<Ivan> n3 is one set of rules

<Don> <vipul> we want to cross link to other ontologies

<Don> <alan> only way right now is to use owl:Imports

<scott> for modeling purposes protective vs. symptomatic

<eric> mechanism modeling will change over time and then molecular level

<Don> it's causuality

<Don> <scott> how to create an ontology to support your view of your experimental results

<scott> evolution of ontology is an important issue

<vipul> map the ontological design to existing data generated in the biordf group

<Don> <dirk> is there anything that can generalize across domains

<Ivan> identify the problems and communicate with the rules group

<Helen> compare different problems based on what criteria

<Don> <helen> hard to evaluate which choice is best...how to compare

<Don> <alan> how to know when your ontology is good

<Don> <helen> results depend on how tools are optimized

<Don> <kei> evaluation is a topic of research

<Don> <eric> think about if the data are available in RDF and/or OWL, what advantages would that provide

<Don> <eric> try the bridging piece first!

<Don> <ivan> wants dates and names...when will you be done!

<Kei> there is a possbility of submitting an invited paper to a special issue of bmc bioinformatics (we can describe different activities in HCLSIG)

Capture Medical Knowledge (Helen)

<Helen> we can use rules to capture semantics for supporting medical decision making

<Don> <helen> sequence of execution is important in ER

<Helen> describes a thrombolysis use case

<Don> <vipul> are you describing a workflow?

<Don> <helen> no...not a workflow

<Don> <helen> treatment sequence for thrombosis patient

<Don> <helen> decision support system

<vipul> is this a task model driven by knowledge

<Helen> it's an architecture argument

<Don> <helen> ontology should map to SnoMed

<vipul> should patients have more properties

<Don> <helen> not the group's task to exhaustively define patient states

<helen> loinc and snomed concepts can be borrowed

<Helen> patients states include operational states

<Dirk> there is a lack of temporal modeling in ontology

<Don> <vipul> we don't want to confound temporal and normalization issues

<Don> <vipul> OWL-S includes precondition and postcondition concepts

<Don> <vipul> implementing functionality of a planner

<Chimezie> the same things can be implemented using existing rule-based systems

<Chimezie> <Alan>process, planning, production systems?

<alan> planning systems employ pre-conditions and post-conditions to formulate plans

<Alan> is the system working?

<Helen> yes

CABG Coronary Artery Bypass Graft (Chimezie)

<Don> <chimezie> DLP Description Logic Programming

<vipul> any relationship with SWRL?

<Don> <chimezie> SWRL and DLP are nearly the same

<Don> <helen> the query is ultimately a set of rules

<dirk> why not using part-of in the anatomical ontology

<chimezie> HL7 uses a two-level approacn

<Don> <vipul> given this modeling approach, what diseases are you able to cover

<Don> <alan> disease severity is a judgement

<chimezie> FMA is very complex

<chimezie> clinical records need to be considered also

<Don> <chimezie> been developing a patient record ontology for about 3 years

<Don> <chimezie> need to model numeric comparisons

<Don> <chimezie> used rules

<vipul> xml can handle numeric classifications

<vipul> are controlled vocabularies used?

<chimezie> yes, but there are problems with string-based vocabularies

<vipul> how counting is done?

<vipul> hl7 includes the notion of value set

<helen> HL7 is not concerned about qualifiers

<chimezie> controlled vocabulary vs. classifications

<chimezie> use of property values

<Alan> one can define enumerated classes

<Don> <chimezie> what situations are best for controlled vocabularies

<eneumann> ACTION: Assess whether enumerable classes can be used to handle controlled vocabularies ala HL7 qualifiers [recorded in http://www.w3.org/2006/10/03-hcls-minutes.html#action01]

<chimezie> computational issues related to the enumerated class construct

<eneumann> ACTION: Eric+Alan, Draft best practice proposal around URI resolution using ontologies, goal before Nov 6th. [recorded in http://www.w3.org/2006/10/03-hcls-minutes.html#action02]

<Don> <alan> usually there is something behind a string that is of use

<Don> <chimezie> how we go about decision support

<Don> <vipul> this is essentially statistical reasoning

<vipul> very thin on semantic reasoning

HCLSIG F2F, 2nd Day

<jar> eric presented a modal logic called KD45 as a way of talking about provenance (skepticism over truth value of statements)

<jar> jar said fine, another tool for the toolbox

<jar> eric suggested that trotting out KD45 as an epistemological stance might help us when we talk with folks from the AI community

Preparation for the day

<jar> then, discussion of best practices document

<jar> ivan said we need plans to make definite documents with definite editors, and assked what they were

<jar> a laundry list of problem is not good enough; we need to spell out what we tried, what the roadblocks were, what's needed to fix

<jar> jar suggested that everyone's practices are very interesting, and writing about these goes beyond just a problems document

<jar> and it's too early to know which practices are a"best" ones

<jar> eric is typing 'slides' as we talk and these should be included in the minutes

<jar> ivan: for any demo, what really important points do the demo demonstrate? this would be a good document

<jar> eric: until lunch, work on specifying these things (use cases, demos, problems) that we will write about

The IG works in break-out groups most of the day, reports of the groups follow below

Break-out group Reports

ACPP Presentation- Helen Chen...

<scribe> ...new charter: http://esw.w3.org/topic/HCLS/ACPPTaskForce/UpdatedCharter#preview

Helen: Temporal reasoning etc.
...notes: BP Note of SW for Adaptive Clinical Workflow (mid Nov 06)
... Bench to Bedside demo around Parkinsons (BioONT)
... artifacts: ACWF demo, Notes, use-cases: derived ontologies and rules
... Working Draft of Notes, to be public, so will need to receive and respond to comments

Ivan for now is staff contact for adding publications

Helen Chen will be editorial lead for ACWP Notes

<ivan> -> pointer to the W3C Publication rules: http://www.w3.org/2005/07/pubrules

BioRDF+BioONT: Vipul presentation...

scribe: USe case nailed: "Show me the location of receptors that bind..."
... spans anatomy, neurosci, mol bio, pharma, clincial/medical

/spans DBS; HI DB (OWL), Neuron DB, BioRDF

scribe: sources Cocodat...
... Demo Tasks: specify SPARQL forms, annotations, merge warehouses...
... start as one merged warehouse, then demonstrate distributed DBs
... user interface? choices?
... Architectural Alternatives (in Note): Data Warhouse , Mediator, etc
...demo: april 2007
... Report: use-cases, arch, value SW prop, UI, editors Kei-Cheung, Don Doherty, Susie Stephens -- Deadline Nov 06

<ivan> -> Found a reference to a system doing federated queries (but I have *no* experience with it): http://darq.sourceforge.net/

also...http: //www.w3.org/2004/10/04-pharmaFederate/

and .. http://www.w3.org/2004/04/30-RDF-RDB-access/

scribe: Don Doherty editorial lead for BioRDF-BioONT report

Helen Chen will be editorial lead for ACWP Notes

scribe: BioRDF-BioONT next steps: joint TC/wiki, sessions map; where to have a demo machine(s) located?
... possible inclusion of guidelines for Parkinson's ala ACPP

BioONT report, first draft Nov 08 ontology modeling and mapping

Alan's demo...

scribe: graph tool 'Prefuse' for graphing classes and instances
... ontology instances for instance relations; use punning later Owl1.1

<jar> hybrow is a great example of a potential consumer of semweb stuff

<jar> it's going to be very sensitive to the wild type/mutant distinction (i.e. what's true in nature versus what's been created in the laboratory)

<jar> and many knowledge bases don't make this distinction ==> BAD semweb practice.

<Joanne> Alan Ruttenberg gave demo - Alan, would you summarize?

<jar> http://www.hybrow.org/

<Joanne> is there a download site for alan's sw and/or presentations?

<jar> location of alan's software is on the hcls wiki: http://esw.w3.org/topic/AlanRuttenberg?highlight=%28lsw%29

<jar> for today's work we'll have to ask him

DSE Group (Eric)

<Susie> Notes from Eric Neuman’s presentation.

<Susie> Identify strong, feasible focus areas.

<Susie> This in the clinical space, but connects to many other activities within HCLS.

<Susie> The theme is translational research, which is cell cultures, animals and into clinical.

<Susie> Share information and knowledge across and within projects.

<Susie> Important to have state thinking about people and animals.

<Susie> Are reactions to drugs favorable or not.

<Susie> Lot of state information from biomarkers.

<Susie> Biomarkers have strong link to the data sets that BioRDF has been working with.

<Susie> Important to map from animal to human.

<Susie> Lot of people focus on genomics and functional genomics.

<Susie> Put data into repository and do mining.

<Susie> Don’t just do this for 1 drug.

<Susie> Are looking to learn from the experience, so can apply knowledge to future work.

<Susie> In mechanism space we are utilizing BioPAX.

<Susie> Domain semantics and flow of them have been defined. (slide 3).

<Susie> Relationship between HL7, CDISC, FDA. (slide 4)

<Susie> Overview of ontologies that relate to HL7, CDISC, etc.

<Susie> It’s all about exchanging data sets. (slide 5)

<Susie> The FDA has large pool of clinical data, but can’t be analyzed because it’s not available. (slide 6)

<Susie> Proposed notes and activities.

<Susie> SDTM exchange model has limitations.

<Susie> Make it available in a Semantic Web representation.

<Susie> Goal is to create a note by November.

<Susie> Eric commits to getting it complete.

<Susie> Want to have demo by early 2007.

<Susie> The demo would work with some set of SDTM data in the scenario of a simple experiment.

<Susie> Latter action items are scheduled for mid to late 2007 (slide 7).

Publication Plan

ACPP: 2 Notes: ACPP Ontology (Mid Nov. 06), Temporal Reasoning (Dec, 06)

BioONT+BioRDF: Parkinsons Use-Case: WorkingDraft Parkinson's Ontology

BioONT: Ontology Modeling and Mapping Note (Vipul)

DSE: CDISC's STDM from a SW Perspective

DSE: (2) Retrospective DBs (JANUS) and SW based annotations and links
... (3) Provenance and trust (non-reputability) and security

Alan Ruttenberg: URI resolution with ontologies (Jan 2007)

scribe: Jonathan Reese lead editor for URIs...
... scheduled for Jan 2007 (maybe lucky sooner ; ) )

Demonstrations: ACPP Demo of adaptable clinical workflows
... Bio+Ont+BioRDF Parkinson's Collab Demo (April 2007)

DSE Demo: STDM Table and XML models ala RDF (early 2007)

scribe: (2) DEMO: of retrospective DB (JANUS) using URI wrapping and annotations