CambridgeSemanticWebGatherings/Meeting/2007-03-13 Gathering

From W3C Wiki

Chairs: Benjamin Grosof

Date: March 13, 2007

Venue: Star Conference Room, Dreyfoos tower, room 463, 4th floor, Stata Center, MIT; and the R&D pub.

Registration List:

  • Benjamin Grosof
  • Susie Stephens
  • Kei Cheung
  • Huajun Chen
  • Tim Clark
  • Paolo Ciccarese
  • Joshua Aresty
  • Lee Feigenbaum
  • Kevin Richard
  • S. A. Batla
  • Michael DiLascio
  • Gwen Wong
  • Eric Neumann
  • Ralph Swick
  • David Booth (tentative)
  • MacKenzie Smith
  • Frank Manola
  • Melli Annamalai
  • Murali Mani
  • Webster Mudge
  • Elizabeth Wu
  • Kingsley Idehen
  • Elgar Pichler
  • Jordana Riley
  • Sandro Hawke
  • Aykut Firat
  • Walter Perry
  • Mihai Lupu
  • Wing Yung
  • François Proulx
  • Francesco Calabrese
  • Jeremy Zucker
  • ... (please add your name Wiki-style via a Wiki topic if you plan to attend).

Demos:

Notes:

% notes from 3/13/07 eve W3C SWIG Cambridge meeting % notes by William Hayes and Benjamin Grosof, % merged and edited by Benjamin Grosof; % informational links provided afterwards by the Demo presenters

Disclaimer: The notes below are not guaranteed to be accurate, and are certainly not complete, but rather are a best effort.

%%%%

Chair of Meeting: Benjamin Grosof of MIT Sloan,

                         Assistant Professor of Information Technology 

Notetaker: William Hayes of Biogen Idec, Research Informatics Library and Info Systems, - william.hayes@biogenidec.com

Benjamin Grosof as auxiliary notetaker

approx. 41 attendees

%%%%

Demo 1: OTN Semantic Web Beta

semantic web based navigation for Oracle Technology Network by S.A. Battla (presenter) and Michael (Mike) Dilascio of Siderean,

      plus Susie Stephens of Oracle 

o Michael Dilascio's info provided within the day after the Demo:

The Seamark relational navigation application is in production and can be seen at http://otnsemanticweb.oracle.com and another production Seamark application at Oracle.com is the "PressRoom" which we did not show. It is at: http://pressroom.oracle.com

Susie will be posting a SWEO use case on OTN application this week with additional information.

o William Hayes' notes on Demo 1 presentation and discussion:

Integrated ~39 silos of information on OTN (the forums and some of the information sources) using RDF.

This system provides faceted views of the content - example SOA Middleware suite facet returns All Articles, Podcasts, Downloads, etc. You can also add meta tag requirements to the faceted view to filter even further.

Showed a pre-set facet of Blogs and the associated tag cloud, with a graphical date filter allowing easy restriction using the date meta-tags.

Showed a pre-set Discussion Forums with a tag cloud but with a more forum-like presentation of the forum articles as opposed to a blog-like presentation above.

Another view is a subset of Forum posters that are defined to be important, you can show each VIP postings using rollovers (across a set of symbolic postings).

Q: Is this database accessible as an SparQL archive? Yes, but it's not opened up to outside of Oracle.

Q: Is it RSS enabled? Yes.

Q: What version of RSS? Don't know

Q: Have you built an ontology on top of the RDF? Yes - Cmark (sp?) was the application that did it. They used the vocabulary provided by Oracle for this. Using OWL-Lite.

Q: Did you categorize the content automatically? No - manually categorized.

o Benjamin Grosof's notes on Demo 1 presentation and discussion:

39 total feeds in a variety of formats: RSS, XML, proprietary content mgmt

aggregating this in a RDF graph in their product Seamark

navigate by "facts" -- a pane of facets on LHS of screen - dynamic in that properties constit'ing facets are difft as go down the graph - flexible wrt what to pivot on

  • topic, products w/in product categories, dates, authors

- difft visualiz's eg: forums use row by row; blogs cf. below - in blogs:

  • use a slider for date
  • use font size to show who blogs more vs. less frequently

Discussion:

Kingsley Idehen: it would be good to be able to query the RDF fm elsewhere A incl. by Susie Stephens: yes, good sugg!

Oracle gave Siderean/Seamark an ontology to use for OTN

use UIMA architecturally to help do categorization of unstructured content

 coming in on pipelines
  - if you have annotators that are UIMA compliant 

have some RDFS (lack domains and ranges) and some OWL-Lite support

%%

Demo 2:

SWAN: Semantic Web Applications in Neuromedicine

Tim Clark of Harvard/MGH presenting ; there's info on him at http://iic.harvard.edu

o Tim Clark's info provided within a day after the Demo:

website: http://swan.mindinformatics.org ; see also: http://alzforum.org ; http://mindinformatics.org

  • ***PDF of the Demo presentation slides is available,

look on the W3C SWIG Cambridge website to find it

o William Hayes' notes of Demo 2 presentation and discussion:

A common digital framework for shared knowledge annotation and scientific discourse. A robust community process for sharing knowledge via the framework. A scalable environment for use to collect, share, browse hypotheses.

Not a model of biology - or correct view of scientific theory - or a model of the internal structure of discourse

Based on an Ontology of Discourse

Knowledge Ecosystem - Interlocking Cycles of Research

SWAN will be deployed to the Alzheimer's Research Forum (Alzforum website) this year - which is a moderated scientific web community for AD with over 4000 registered members - a significant portion of AD research community is part of it.

(Refer to SWAN website for Ontology structure)

Q: How are you getting the concepts incorporated into SWAN? We use an application to upload the information in an easy format that is then incorporated into the semantic knowledgebase.

Q: Will the information be available publicly? No not right now - it will be available publicly after release. These are highly curated triples that are slowly being generated - will take a long time to flesh out.

o Benjamin Grosof's notes of Demo 2 presentation and discussion:

goals: serve bench scientists shared K annot and org of sci discourse community process for K sharing scalab env

coarse-grained semantics of how documents / experiments relate to each other revolves around struct of hypotheses help curate review articles

will be deployed on Alzheimer Research Forum (Alzforum) soon (2007)

several 10's of percent of the active res'ers in the field are registered members of Alzforum

Research Questions, Research Statements, cites to Articles, Hypotheses relnsh: refutes, supports, discusses, alternativeTo

can drill down navigationally, e.g., from Res Statem's to Hypotheses to Claims to Articles

compare to Pathogenic Narratives in time sequence of disease etiology can help allocate resources by funders wrt upstream vs. downstream

 in such seq's

enthusiastic users of IBM SLRP (Boca) from the IBM Cambridge SW group

Discussion:

A to Q by _ Astra Zeneca (woman): yes, would be good to add inferencing about the hypotheses as well

A to Q by Kingsley Idehen: the instance data will be public, but isn't yet; but is relatively small: they are hard-won triples

%%%%

Demo 3:

DartGrid: Building Semantic Web on Relational Databases

Huajun Chen visitor at Yale, from Zhejiang Univ. in China

  working w/ Kei Cheung of Yale Medical School

- in town this week fortuitously

o Huajun Chen's info provided within a day after the Demo:

The link to DartGrid is : http://ccnt.zju.edu.cn/projects/dartgrid, everything about DartGrid is on the website.

o William Hayes' notes of Demo 3 presentation and discussion

[[[DartGrid]]: Building Semantic Web on Relational Databases] using a programming framework together with set of semantic tools that support relational databases

User layer - semantic browser, query, search, navigation Semantic Web Layer - defines vocabulary, formal model, and semantic relations between databases

Demo shows a visualized semantic mapping tool with mappings are defined as rules. It aims to deal with complex mappings.

DartGrid DartMapping

Q: How does it compare to Cleo? Don't know

Q: How does it handle views? just like tables

Q: What dbs are mappable? mysql, Oracle, others

Q: What is the diff between this and Object relational mapping tools? They are similar but

Q: What have you done with it? We have integrated several datbases, SenseLap project, neuroscience databases, query gateway for those projects

Dartmap is also a W3C HCLS WWWW2007 Demo project IBM SLRP (Boca)

AOB

IBM Workshop on RDF and ... info

o Benjamin Grosof's notes of Demo 3 presentation and discussion

semantic wrappering of DBs

visualized sem. mapping tool

aims to deal w/ complex mappings: physical properties and logical structuring

dynamic query and search tool

rewrite SPARQL to SQL, cross DB boundaries

import RDF/OWL ontol's

look at DB's tables, properties, entities

mapping panel -- drag the above into it to create mappings - assoc DB (table,column) to a ontological property

semantic registry -- of mappings etc.

then maps SPARQL to SQL, retrieve it there, map back to SPARQL

(didn't have time to demo the query rewriting and exec part of the sys)

Discussion:

Q: how relates to Clio relational to XML mapping A: similar, but SW-based

Q: can handle views and derived tables A: yes, handle it same way

A to Q: use JDBC or Oracle to mySQL

Q: what's different from relational-to-relational DB-DB mappings A: pretty similar; but then can query by concepts in SPARQL rather than

     having to do it in terms of tables and columns

has been used to integrate 70 difft DBs for a Chinese portal

 [Susie Stephens observation]

have been applying to neuroscience, for example

Q: how would you store new derived RDF back? into a relational DB?

   - esp. to avoid redundancy 

A: could do it in a SPARQL store, but we don't "do storage" Observation by Eric Neumann: ditto for annotation


If you have any questions please contact Benjamin Grosof

(If the Wiki page is immutable, you will need to login before being able to add your name.)