See also: IRC log
ok, TIm
<TimC> joanne
<TimC> rubin
<TimC> m roos
<Paolo> http://code.google.com/p/annotation-ontology/
<scribe> scribenick: mscottm
Tim: We want to discuss the
Annotation Ontology (link above from Paolo), which is an open
project.
... We also want to discuss the Annotation Framework, which
enables you to curate, modify, and respond to text mined
results in RDF.
... Over to Paolo now.
Paolo: AO an
ontology/vocabulary/data model that we use to annotate online
documents.
... AO was made to be used in the Annotation Framework. We
tried to make it as big as possible to cover annotations for
journal articles or documents on websites that we don't
control.
... AO could also be used for medical reports.
... Terms can be attached to tags as well as fragments and
images.
... You can use the ontology to connect all relevant
information to a 'claim', especially 'evidence'.
<scribe missed the last bit due to an interruption - feel free to add in>
<anita> THere are three use cases for our rhetorical document task at http://esw.w3.org/HCLSIG/SWANSIOC/Actions/RhetoricalStructure - I'd be grateful if we could include these, thanks.
Rob: Would you like to talk about
it in terms of particular use cases.
... Would you like to talk about it in terms of particular use
cases?
... It would be great to talk about how to capture from full
text and express in ontological terms.
Daniel Rubin: We in the Terminology task force have radiology and pathology reports that would serve well as a use case here.
<anita> Hey Cartic Ramakrishnan from ISI!
Cartic Ramakrishnan from ISI: This approach could be useful for biocuration.
<anita> Here is more on SciKnowMine: https://wiki.birncommunity.org:8443/display/NEWBIRNCC/SciKnowMine
<jodi> http://www.informatics.jax.org/
<jodi> So, triage, and subsequent annotation using text mining. With the potential for human correction in AO.
scribe: problem is 'document triage' for Mouse Genomics Informatics, whether they contain a novel gene, gene expression, or other subject of interest
<anita> Cartic: in SciKnowMine we do triage of documents
Tim: Let's add these things to the wiki.
What Daniel Rubin was referring to (we've considered using text mining in this use case): http://esw.w3.org/HCLSIG/Terminology/PathRadCorrelation
<anita> Anita: let's link all the use cases to a specific page; Paolo: yes, somewhere on the wiki.
Tim: Interested in a common
format for sharing annotations produced from text mining.
... Also, want to provide the annotation software as a set of
services
Caric: Doing this with Jackson Labs, Univ of Colorado, and Univ of Utah
Cartic: Doing this with Jackson
Labs, Univ of Colorado, and Univ of Utah
... Created RDF triples from text mining and is available.
Wondering if this community is interested in that data?
Daniel: Would you provide a link?
Tim: Please also provide a link to the annotation schema that you used.
Daniel: How does it relate to the Annotation Ontology?
Cartic: Hoping to find that out. Could possible transform / map to it.
Tim: That's what the Annotation
Framework is for
... The first service that we connected to is the NCBO
Annotator and found some issues. Who should we contact?
Daniel: Nigam Shah
Paolo: There are some truncated term URI's
<Trish_> Read Codes, Clinical Terms Version 3
<mroos> FYI Some time soon we will talk about the nano-publication format for the Concept Web again (work on some proof-of-principle examples). Initial concept web content will be partially based on text mining, hence I think the format shall be a subset of a text mining annotation model. Seems obvious to me that we will find common ground. NB I'm only following the chat (trouble calling in).
Trish: Unless there is an error in the original source file ontology, they are supposed to be snippets.
Paolo: Just tried and they resolved. Will look at this again.
<Trish_> The URIs from the Read Codes includes "."*n at the end of the URI
<Trish_> this is how the data is in the source ontology file
<Trish_> E.g. http://purl.bioontology.org/ontology/RCD/B7...
<TimC> Hi Marco - we can chat about that in deatil when I visit Manchester
<Trish_> will resolve, but extra "." may not be automatically included in hyperlink
<jodi> the second one: http://esw.w3.org/HCLSIG/SWANSIOC/Actions/RhetoricalStructure/UseCases/1
<jodi> the three sets of Rhetorical Structures use cases are linked from here: http://esw.w3.org/HCLSIG/SWANSIOC/Actions/RhetoricalStructure
Scott: <elaborates on use case
suggested by Daniel Rubin from the Terminology task force>
We started with manually crafted RDF representing salient
features of radiology and pathology reports. Long term goal is
more automation using something like NCBO Annotator or ODIE
from Univ. of Pittsburgh, Rebecca Crowley, which creates
ontology annotations for Surgical Pathology Reports.
... Also mentions that colleague Marco Roos couldn't get into
the call (call is full). Explains the relation of text mining
and Concept Web Alliance to SciDisc. CWA based initial
modelling of discourse and assertions on SWAN/SIOC ontology.
There is an abundance of material available from CWA.
<scribe has fallen behind! any help welcome - just write into the irc please>
<mroos> Has IsaTab been discussed in these calls? (Sorry I missed so many :-[) This seems to be very popular in bioinformatics.
Tim: Suppose that we had a way to share text mining results through the use of common services like those based on AO. Use feedback from the community to improve the approach, reach for consensus, check interop of results.
Cartic: This sounds like a workable approach. We would be willing to share some data.
<mroos> I think IsaTab also uses NCBO for annotation. But not (yet) in a very RDF kind of way (as far as I could see).
<mroos> NB Please ignore my chats if they are out of sync ;-)
<mroos> http://isatab.sourceforge.net/
<rfrost> using AO to build a gold standard for training NER classifiers: set of NER classifiers process a set of documents and map to AO annotations, these pooled results are reviewed by human curators using AO-based tools to create gold standard (w/ the additional detail supported by AO), this gold standard is used to train a new NER classifier or to enhance the input classifiers. Explore whether the superior NER performance can be achieve via the use of an AO-based tr
<TimC> Hi - ISATAB has been discussed on the Discourse-Data-Experiment thread of this call but not in this particular context
<Trish_> The ISATAB tool does use ontologies from NCBO
<mroos> Re: Isatab. Thanks. Would you have a link that I can forward to some people using ISATAB? They may not be aware of your experiences (assuming there was a report of some kind of that).
Scott: <long preamble> Speaking for both Terminology task force and Concept Web Alliance, both would be interested, in slightly different ways. Reaching toward interoperable RDF from text mining for publishing in the Linked Open Data cloud is an excellent goal. Pistoia Alliance SESL project (Ian Harrow, Dietrich Rebholz, et al) also has relevant output and potential interest.
Tim: Next week is Columbus Day in the U.S. Next call will be in two weeks Bibliographic records and citations, David Shotton and also Elsevier.
<Trish_> http://isatab.sourceforge.net/index.html
Scott: Let's add the list of text mined RDF resources to the wiki.
Tim: I'm excited about the way this is developing. Talk to you all in two weeks.
<Tudor> leave
<johnM> Hi scott, are you there?
<matthias_samwald> the biordf call just ended.
<johnM> Hi matthias
I am here now.
but not on the phone.
<johnM> Hi xiaoshu
<xiaoshu> hi John
<xiaoshu> calling in now
<scribe> scribenick: mscottm
John: I'm here at a Hyatt with several people from an HL7 meeting, including Eric Prud'hommeaux
Francois: I work for a government agency, co-chair of XXXX and YYYY.
<ericP> victor brotsky: asst director of informatics at ....
<ericP> ... liason to IHE pathology group
<ericP> mary kennedy: CAP
Mary Kennedy at NCI part of the CAP
<ericP> crytal: @@paris
Christel
<ericP> ... working on content integration profile: manage terminology attached to pathlex
<ericP> geoffrey carp: CAP assiting dr. booker and dr. brotsky in running the AP group
<ericP> david booker: liason from from HL7
tx
<ericP> daniel rubin: standford -- radiology
<ericP> harold solbrieg: mayo clinic
<ericP> ... lots of terminolgy work
<ericP> ... curator of lexgrid
<ericP> matthias_samwald: at DERI galway
<ericP> ... working on improving interop in life sci, translational med via ontologies, data standards...
<ericP> mscottm: co-chair of hcls. making the most of data via ontology and related standards
<ericP> xiaoshu: @@1 in UNC challel hill
ericP: <tells about HCLS to
people in hotel room>
... HCLS creates demonstrators that make use of public data (so
they can be demonstrated).
... Clinical Observation Interoperability: Look at how Semantic
Web can ease data integration, mapping, and queries across the
information systems of different hospital systems and clinical
trials.
..Terminology: How do we use RDF versions of medical
terminologies in the clinical context?
BioRDF: How do we assemble data from different sources and databases to answer scientific questions?
Linked Open Drug Data: Pushing data into the Linked Open Data cloud that is relevant to HCLS, such as Allen Brain Atlas (latest endeavor).
Translational Medicine Ontology: Building a patient-centric ontology for translational medicine.
Scott: Scientific Discourse - ontologies for representing discourse, including Annotation Ontology and Annotation Framework.
EricP: describes how HCLS approaches make it possible to create personalized medicine and better leverage data in health care and life sciences.
John: PathLex is being created (named in homage to RadLex). Considering whether to use OWL/RDF, SKOS/RDF or other.
Christel: We will face the issue of querying. Having a content profile could help with querying.
Harold: Translating between pre-coordinated and post-coordinated would benefit, example table of people with broken femurs vs. femur broken boolean and laterality vs. tags
MaryK: Problem using post-coordination and looking for pre-coordinated examples that fit it.
John: One problem is that there is no clear semantics for SNOMED that enable you to test if a SNOMED assertion is valid.
Harold: ITSIDO is using ECL+ reasoning on SNOMED in OWL (?)
John: Disjoint expressions..
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: mscottm Found ScribeNick: mscottm Inferring Scribes: mscottm WARNING: No "Topic:" lines found. Default Present: +1.302.598.aaaa, jodi, +1.706.206.aabb, +1.617.768.aacc, Scott_Marshall, +1.619.252.aadd, +1.603.868.aaee, Joanne_Luciano, +1.650.331.aaff, matthias_samwald, +1.619.252.aagg Present: +1.302.598.aaaa jodi +1.706.206.aabb +1.617.768.aacc Scott_Marshall +1.619.252.aadd +1.603.868.aaee Joanne_Luciano +1.650.331.aaff matthias_samwald +1.619.252.aagg WARNING: No meeting title found! You should specify the meeting title like this: <dbooth> Meeting: Weekly Baking Club Meeting WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth Got date from IRC log name: 04 Oct 2010 Guessing minutes URL: http://www.w3.org/2010/10/04-hcls2-minutes.html People with action items: WARNING: No "Topic: ..." lines found! Resulting HTML may have an empty (invalid) <ol>...</ol>. Explanation: "Topic: ..." lines are used to indicate the start of new discussion topics or agenda items, such as: <dbooth> Topic: Review of Amy's report[End of scribe.perl diagnostic output]