HCLSIG BioRDF Subgroup/Meetings/2009/09-28 Conference Call
Conference Details
- Date of Call: Monday September 28, 2009
- Time of Call: 10:00 am Eastern Time
- 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: 4257 ("HCLS")
- IRC Channel: irc.w3.org port 6665 channel #hcls (see W3C IRC page for details, or see Web IRC)
- Duration: ~1 hour
- Frequency: bi-weekly
- Convener: Kei Cheung
- Scribe: Eric Prud'hommeaux
Attendees
Kei Cheung, Eric Prud'hommeaux, Matthias Samwald, Adrian Paschke, Scott Marshall, Satya Sahoo, Huajun Chen, Jun Zhao
Regrets
Rob Frost
Agenda
- Introduction [Kei]
- HCLS KB upate [Matthias, Adrian]
- F2F meeting -- biordf [All]
- Semantic Web Development for Traditional Chinese Medicine: a Report for Current Status [Huajun] (slides)
Minutes
<ericP> topic: KB update
<mscottm> I just skyped the new code to Matthias.
<ericP> kei: matthias migrating to a new server and virtuoso instance
<mscottm> Matthias is trying to dialin.
<ericP> adrianpaschke: we're migrating to a new server and new version of allegrograph
<ericP> ... should be more stable
<jun> Adrian is also updating his server in Berlin, and software. will perform much better
<ericP> ... (enterprise edition)
<ericP> topic: F2F 2-3Nov
<ericP> kei: attending: kei, mscottm, ericP, others?
<ericP> ... we'll be doing a joint activity with LODD
<ericP> ... would could work on traditional chinese medicine
<mscottm> Zakim, who is here?
<Zakim> apparently Team_(HCLS)13:49Z has ended, mscottm
<Zakim> On IRC I see adrianpaschke, jun, ssahoo2, mscottm, RRSAgent, kei, huajun, Cloud, AlexPassant, Zakim, ericP
<ericP> ... we invite other ideas
<ericP> topic: KB update
<ericP> matthias: we're using virtuoso six
<ericP> ... the amount of data is large (too many named graphs)
<ericP> mscottm: ~200 named graphs
<ericP> matthias: there's no dedicated mechanism for dumping a graph?
<Zakim> sorry, adrianpaschke, I don't know what conference this is ericP adrianpaschke, you need to join #hcls3 to talk to zakim
<mscottm> you have to issue that command on #hcls3 to get results
<mscottm> http://esw.w3.org/topic/HCLSIG_BioRDF_Subgroup?action=AttachFile&do=get&target=W3C-Group.pdf 16:12 kei http://esw.w3.org/topic/HCLSIG_BioRDF_Subgroup?action=AttachFile&do=get&target=W3C-Group.ppt
<ericP> topic: TCM slides
<ericP> -> http://esw.w3.org/topic/HCLSIG_BioRDF_Subgroup?action=AttachFile&do=get&target=W3C-Group.ppt TCM slides
<ericP> [slide 7: The ultimate vision of the TCM Semantic Web]
<kei> cheung: chinese academy of tcm is located in Beijing
<ericP> Huajun: we've helped developed many TCM databases
<ericP> ... there are many which have been developed by local communities
<ericP> ... will be very useful to make a connection between TCM knowledge and western medicine
<kei> cheung: great interest in linking TCM and western medicine
<ericP> ... chinese medicine has its own logic derived from traditional chinese philosophy
<kei> cheung: many technical/cultural challenges
<ericP> ... there is a wide gulf between western medical culture and chinese philosophy
<jun> cheung: also does data mining and analysis
<jun> cheung: tcm ontology engineering started in 2001
<ericP> ... dartgrid uses semantic tech to integrate diverse databases
<jun> cheung: currently focused on building a semantic query engine
<ericP> [slide 10: TCM Ontology Engineering]
<ericP> kei: [re slide 12: TCM ontologies] who created these?
<ericP> Huajun: collaborative project between my univ and China Academy of TCM
<ericP> ... we provide technical support
<ericP> ... they have a committee
<ericP> ... a small percentage of these have corresponding english names
<ericP> ericP: so you have translations to English *and* correspondances with western medical terms
<jun> cheung: it's hard to translate, because the concept is different
<ericP> Huajun: mostly we have the english translation of the chinese concept
<ericP> mscottm: e.g. chi-related terms like "dampness" and "heat"
<ericP> kei: LODD folks attacking this from the chemical ingredient angle
<ericP> s/re slide 12:/re slide 11:/
<ericP> [slide 12: Ontology visualization and query engine]
<ericP> Huajun: allows you to graphically browse semantic connections
<ericP> [slide 14: System Architecture]
<ericP> Huajun: provide a semantic mapping from db to high-level concepts
<ericP> ... compiles into an SQL query plan
<ericP> ... doesn't do materialized views, uses query mapping instead
<jun> huajun: transformation by query, using query rewritter
<mscottm> It is here: http://sourceforge.net/projects/swobjects/
<mscottm> (That is Eric's approach to federated query)
<ericP> Huajun: offer full-text-search in all DB's
<ericP> ... have a new version which integrates more data
<ericP> [slide 15: Visualized Mapper]
<ericP> Huajun: makes user's job easier
<ericP> [slide 16: Semantic Search Portal Version 1]
<ericP> huajun: user enters terms
<ericP> ... gets mapped to semantic terms
<ericP> ... TCM ontology provides search suggestions based on connections
<ericP> ... combining new resources
<ericP> ... major feature: uses the ontology as a sort of search index
<ericP> ... easier for user to retrieve data sets and related items
<ericP> kei: is this released to the public or to a consortium?
<ericP> huajun: rules are changing
<ericP> ... traditionally didn't want to make data public
<ericP> ... they're providing open access to some of this data
<ericP> ... e.g. data coming from public web pages
<ericP> ... we're working on semantic integration of these DBs
<ericP> ... asking what can we do with this huge web of data
<ericP> [slide 19: Graph vs Semantic Graph]
<ericP> huajun: in graph, the basic data element is a node. in RDF, it's an arc
<ericP> ... we can view a graph from two different perspectives
<matthias_samwald> (need to leave now, bye!)
<kei> bye, matthias, thanks for attending
<ericP> ... migrating by looking for most common patterns
<huajun> hello
<huajun> am i lost
<ericP> yep
<ssahoo2> kei I need to leave
<kei> bye sahoo
<mscottm> Huajun - can you dial in again?
<kei> huajun can you deal in again
<mscottm> bye sahoo
<huajun> I am trying
<huajun> but it is restricted
<mscottm> Huajun - we will have to get another code.
<jun> me
<huajun> what do we do?
<jun> we all hang up now
<huajun> I am about to be finished
<mscottm> We will get another code. Wait a minute
<ericP> Zakim, space for 7?
<Zakim> sorry, ericP; could not schedule an adhoc conference; passcode overlap; if you do not have a fixed code you may try again
<huajun> It has past 11, is that ok?
<kei> it's ok ...
<kei> if it's ok to you
<ericP> 22631
<ericP> sorry
<kei> i'm dialing in now ...
<mscottm> New code is: 26631
<ericP> 25531
<mscottm> New code is: 25531
<ericP> s/5/6/g
<mscottm> New code is: 26631
<jun> i am in now
<ericP> [slide 21: An example]
<ericP> huajun: we're try mining approaches
<ericP> [slide 22: Semantic data analysis]
<ericP> [slide 23: DartSpora: a interactive mining engine for TCM]
<ericP> huajun: allows user to combine [and present] different data
<ericP> huajun: connecting TCM and western medicine at the data level will help both communities find and use patterns
<ericP> mscottm: is someone working on a map from western disease names and conditions to TCM symptoms?
<ericP> huajun: we couldn't find many direct correlations
<ericP> ... we're using a statistical approach
<ericP> ... e.g. identify a set of associated symptoms in TCM and find a corresponding [symptoms for a] disease
<ericP> ... tells you that there are relationships, but not that they are equal
<ericP> kei: e.g. TCM "hot" maps to {fever, inflamation, }
<ericP> ... getting deeper, the mapping probably becomes more difficult
<ericP> ericP: the heuristic mappings should meet most of the use cases, so long as the interface indicates that the association was derived heuristically
<ericP> kei: example: alzheimer's is well known in western med, but you need to look by it's synonym "dimentia" to find mappings in TCM
<mscottm> I have to go. Thank you Huajun.
<mscottm> Bye.
<ericP> huajun: [mentions plan for TCM XG]
<kei> bye scott
<kei> eric and huajun explore the query mapping