HCLSIG/SWANSIOC/Meetings/2010-09-27 Conference Call

From W3C Wiki
Jump to: navigation, search

Dial-in & IRC Information

  • Dial-In #: +1.617.761.6200 (Cambridge, MA)
  • Dial-In #: +33.4.89.06.34.99 (Nice, France)
  • Dial-In #: +44.117.370.6152 (Bristol, UK)
  • Participant Access Code: 42572 ("HCLS2")
  • IRC Channel: irc.w3.org port 6665 channel #HCLS2 use IRC direct link or (see W3C IRC page for details, or see Web IRC)


Agenda

1. Housekeeping

2. Tim Clark: brief introduction to task context - Roles of SciDisc Subtasks with respect to each other... also see listing of Scientific Discourse subtasks at http://esw.w3.org/HCLSIG/SWANSIOC

3. Tudor Groza/Paolo Ciccarese: Proposal for the Ontology of Rhetorical Blocks (ORB) - see http://esw.w3.org/HCLSIG/SWANSIOC/Actions/RhetoricalStructure/models/blocksontology

4. Discussion.

5. Adjourn

Minutes

[10:13] TimC: Goal is to provide a strong interface to text-mining for linking web of documents to biomedical ontologies

[10:15] Paolo: http://esw.w3.org/HCLSIG/SWANSIOC/Actions/RhetoricalStructure/models/blocksontology

[10:16] TimC: everyone should go to this: http://www.slideshare.net/tim.clark/scidisc-slides-27-sept-2010-5296765

[10:16] Paolo: okay, Paolo speaking - purpose of ORB is to provide a "menu" of possible blocks for a biomedical article

[10:17] Paolo: http://en.wikipedia.org/wiki/IMRAD

[10:17] Paolo: http://salt.semanticauthoring.org/ontologies/sro

[10:17] TIMC: mixed IMRAD with SALT

[10:19] Paolo: The block ontology is mainly for the body of the article, not for the head and tail. The head and tail are covered by other ontologies

[10:19] Paolo: we understand we cannot force people to use one or the other bibliogrpahic ontology at this point, we will provide the blocks for Head, Body and Tail with guidelines

[10:19] DavidShotton: That is what I have been phoning - now reconnected and listening to Paolo, but not sure how long this call will last.

[10:20] Tudor: these blocks are to be used as blocks within sections - mixed levels between actual paragraphs and entire sections

[10:21] Paolo: this is only the beginning, the Big Blocks, we will work on the details later on

[10:21] Paolo: pretty simple but easily identified in an article

[10:22] Sudeshna: for connecting data with discourse we would like to identify tables - how can we be more explicit about them, also figures?

[10:22] Paolo: I think that is fine grained structure, could be part of results, could be part of methods -

[10:23] Paolo: should we make a stronger recommendation about use of specific ontologies in head and tails?

[10:27] TimC: Schedule a separate session on different ontologies for bibliographic citations,and analyze the pros and cons of each

[10:27] Tudor: let's analyze the different approaches in general first, and then see what we should do

[10:28] DavidShotton: Apologies - call dropped again.  I think it best if I just try to listen.  

[10:29] TIMC: David, what you can do also is type comments into IRC

[10:29] Tudor: David's questions is about rhetorical vs physical blocks

[10:30] Howard: I understand claims and evidence; not so clear that these traditional physical parts of articles really help us get to claims and evidence

[10:30] Paolo: we are avoiding blocks such as "Evaluation" - looking for things almost always present, often titles of sections

[10:30] TIMC: Another way of sying what Tudor said is current patterns are not always present in the physical section of teh article, but some blocks may be

[10:32] TimC: Answer to Howards's questions. These blocks help text-mining processes to identify claims and evidence.

[10:33] TIMC: But also these blocks are to help identify entities

[10:34] DavidShotton: In our draft Document Components Ontology, which takes terms from the National Library of Medicine DTD, I distinguish sections of the main text: {Introduction  [ = Background ], Methods and Materials (or     Methods [ = Experimental Procedure ] and Materials separately), Results  [ = Findings = Data ], Related Work (in Computer journals), Discussion, Conclusions} and Document items additional to text {e.g. Figures and Figure Legends}.

[10:37] TimC: Shall we aim to align the 2 efforts?

[10:38] DavidShotton: we can make an equivalence between classes

[10:38] Tudor: Here is a problem of semantics, relating to "physical" sections of articles in DOCO, while ORB goes a bit further towards rhetorical blocks

[10:39] Tudor: might not be a background section but could be a couple of paragrpahs

[10:40] TimC: what are the use cases?

[10:40] TimC: Are the 2 efforts addressing same or different use cases. If they are addressiong different use cases, they there can be a partioning of functions that can go on. Otherwise we should align

[10:42] David: Started o ontology of citations - CiTo, then ahrmonized with SWAN, and then there was a separation of objectsof citation vs the citation itself

[10:42] David: we started with an ontology that described both Citations and the objects of citations.  As part of our collaboratoin with Harvard, we partitioned out Cito as describing citations from FaBIO for bibliographic records

[10:43] David: then we built more such  as C4O citation counting

[10:44] Sudeshna: Citional network analysis is big use case for this effort

[10:44] David: citation network analysis is a major use case

[10:44] TIMC: seems redundant

[10:44] Sudeshna: ok

[10:45] TimC: Start with overlap what you working on and this task

[10:46] Paolo: Some of the blocks are coming from SALT ontology, we can add references to SALT, DCO etc

[10:46] Paolo: Next step is to work on head and tail - Look at options and their pros and cons

[10:47] TimC: Benefit of working on head and tail vs fine-grained details

[10:48] Paolo: Alignment of body is easy part, head and tail analysis is most work, should also tryo look at BiBO

[10:49] David: Mapping with BiBO to FABIO, using SKOS

[10:50] Paolo: Align elements of body and dive into head+tail

[10:50] David: I have ideas for head and tail

[10:52] TimC: Commet represents comment on web-based papers

[10:53] Tudor: SIOC:Comment is different, so is SWAN:Comment

[10:53] Paolo: Like to keep comment separate

[10:55] TimC: Paolo, Tudor, David to align coarse rhetorical block - mapped or joined etc.

[10:56] Palo: Report after Mon after Nov 1st

[10:59] TimC: Goal is to get a sig note and then put it on NCBO

[10:59] ScottM: Don't b shy and feel free to update versions

[11:00] ScottM: Comment features helps collaboration

[11:01] TimC: Next Meeting, Mon Oct 4: AO and its alignment with text-mining community