15:07:56 RRSAgent has joined #hcls 15:07:56 logging to http://www.w3.org/2006/04/11-hcls-irc 15:08:09 Zakim, who's on the call? 15:08:09 sorry, vipul, I don't know what conference this is 15:08:12 On IRC I see RRSAgent, Zakim, Don, Davide, vipul, JohnMadden, xwang4, Ken 15:08:34 rrsagent, show actions 15:08:34 I see no action items 15:08:55 action: discuss ontologizing snomed 15:09:10 action: use case discussion continued 15:09:23 rrsagent, show actions 15:09:23 I see 2 open action items saved in http://www.w3.org/2006/04/11-hcls-actions.rdf : 15:09:23 ACTION: discuss ontologizing snomed [1] 15:09:23 recorded in http://www.w3.org/2006/04/11-hcls-irc#T15-08-55 15:09:23 ACTION: use case discussion continued [2] 15:09:23 recorded in http://www.w3.org/2006/04/11-hcls-irc#T15-09-10 15:10:42 Davide maps only first order objects into snomed 15:11:02 viz one relationship in SNOMED = 1 property in OWL 15:11:19 1 concept in SNOMED = 1 resource in OWL 15:11:53 some differences in style between snomed and OWL 15:12:39 (1) what owl sees as relationships do no coincide with OWL properties 15:15:31 alanr has joined #hcls 15:17:38 Snomed as annotation ontology v/s Snomed for reasoning 15:18:08 q+ 15:18:42 Zakim, who's on the call? 15:18:42 sorry, vipul, I don't know what conference this is 15:18:43 On IRC I see alanr, RRSAgent, Zakim, Don, Davide, vipul, JohnMadden, xwang4, Ken 15:19:33 q+ 15:22:00 Vipul: kinds of inferencing that are useful in medicine 15:22:13 (1) transitive closure of isA relations 15:22:24 (2) Propagation of definitional changes 15:23:03 Need to characterize uses of ontologies... 15:23:28 (3) boolean operations on classes (irrespective of relations that cross isA hierarchies??) 15:24:16 Amit: SNOMED x wordnet 15:25:03 Amit: EMR deployed at e.g. heart center for applying RULES 15:25:09 q+ 15:25:24 q- alanr 15:25:55 xiaoshu: earlier made the point that SNOMED should be mapped to OWL using annotation properties 15:26:01 Helen has joined #hcls 15:26:02 john: agrees 15:26:18 john: Some differences between SNOMED and OWL 15:26:29 (1) SNOMED has no concept of instances 15:26:42 (2) SNOMED very sparingly uses binary relations 15:26:58 q- 15:27:01 q 15:27:13 q+ 15:27:30 xiaoshu: there are many ways one could convert and existing taxonomy into OWL 15:27:32 q 15:27:34 q+ 15:27:57 xiaoshu: there are many ways to map legacy taxonomies into OWL 15:28:11 xiaoshu: there is no "better" or "worse" 15:28:12 q+ 15:28:26 q+ 15:28:50 john: (3) SNOMED has very few concepts that are "fully defined" i.e. they lack necessary and sufficient conditions 15:28:55 q 15:29:35 helen: reference ontology would be an abstraction layer above the level of SNOMED 15:29:59 q 15:30:04 q+ 15:31:12 q+ 15:31:50 q- 15:32:57 q+ 15:33:44 vipul: we need to focus on USE CASES 15:34:15 vipul: second layer-- what inferenceing (computational) operations does our use case need to support? 15:34:32 vipul: SNOMED has both pre- and post-coordination 15:34:51 vipul: SNOMED is based on a description logic 15:34:56 q+ 15:35:10 john: in fact, SNOMED did not use negation!!!!! 15:35:57 vipul: it might be sufficient to use transitive closure in some use cases 15:36:15 is SNOMED has a uri for a permant online location? That would be the first step to make it semantic webized 15:36:20 q- 15:36:24 vipul: in other cases, might need to support other features, including composed concepts (post-coordination) 15:36:51 davide: is puzzled. What's this approach of using SNOMED just as a source of identifiers? 15:37:08 davide: but this doesn't sound interesting 15:37:24 davide: on the other hand, this will rule out more interesting efforts 15:37:45 davide: there are some kinds of relationships that SNOMED has which are actually interesting 15:37:55 davide: like for example role-froups 15:37:57 groups 15:38:17 xiaoshu: mapping from whom to whom 15:38:27 I'm leaving IRC to start traveling to a noon meeting but I remain on the phone. 15:40:06 davide: proposed showing how biordf queries can be re-expressed in SNOMED 15:40:09 Action item: BIORDF queries and express them in Snomed... 15:40:20 q 15:40:21 q? 15:40:30 q- 15:40:37 q? 15:40:43 q- 15:41:26 Amit has joined #HCLS 15:41:28 q+ 15:41:34 alan: wants to respond to annotation properties for referenceing SNOMED 15:42:16 alan: instead of using (SNOMED) isA, you create an OWL property hasSubcode 15:42:34 alan: you create a subproperty of hasSubcode that is hasDirectSubcode 15:43:08 alan: can query using query on exists(hasSubcode X) 15:43:40 alan: then add inverse slots so you can climb up the isA hierarcy too 15:44:07 Need a list of relations in Snomed 15:45:24 alan: can use to describe a form 15:45:43 q+ 15:45:44 q-alanr 15:46:35 q+alanr 15:47:22 q+ 15:47:33 xiaoshu: can we get copies of the pristine SNOMED tables? 15:47:39 john: yes, I'm sure 15:47:57 q-xwang 15:48:01 xiaoshu: even if you use annotation properties, you can reason using rules 15:48:03 q-vipul 15:48:21 vipul: concern re alan's suggestion 15:48:33 vipul: isn't this recreating SNOMED? 15:48:36 q+ 15:48:52 vipul: why should we do this? 15:49:15 alanr: so we can compute transitive closure 15:49:43 vipul: again, use cases: what reasoning do we NEED? 15:50:16 vipul: why does Manchester decided to use hasSubcode instead of using isA reasoning? 15:50:31 alan: isA means something different from hasSubcode 15:50:35 q+ 15:51:04 alan: it is better to avoid isA unless you're sure you mean isA 15:51:22 vipul: the trans clos of hasSubcode also has its own semantics 15:51:50 vipul: is hasSubcode transitive? 15:52:10 alan: depends on whether SNOMED terms are modeled as instances or as classes 15:52:27 alan: similar to how partonomy is implemented in GO 15:53:13 alan: if model as instances, then would use direct property and... 15:54:41 action: alan will write up his examples from manchester that show how the "mapping" from SNOMED using hasSubcode would work 15:54:58 action: davide would prepare some examples from his use case world 15:55:20 action: don and ken incorporate into current use case 15:55:56 action: xiaoshu, to look at queries that bioRDF created and see how they would be written in SNOMED 15:56:00 q? 15:56:49 Some early work on evaluaiton of two deep domain ontologies with OBO ontologies and MGED ontology is reported in a WWW2006 paper 15:57:00 there is a link on related info page 15:57:02 amit, can you give reference 15:57:19 I will put up the paper on the wiki today 15:57:23 thanks 15:57:36 http://lsdis.cs.uga.edu/library/download/p1088-sahoo.pdf 15:57:38 some discussion of SPARQL 15:58:07 q? 15:58:10 q-alanr 15:59:29 http://www.snomed.org/2006/06/ 15:59:49 http://www.snomed.org/2006/06# 16:01:15 http://www.snomed.org/2006/06/vocab# 16:01:20 http://www.snomed.org/sct/2006/04# 16:01:43 http://www.snomed.org/2006/06/sct# 16:01:48 agree with alan 16:02:26 I have put Amit's ontology quality paper on the Useful Links page 16:03:14 it is polite to have something at the web at this address, but not necessary 16:03:36 don't confuse uri with url 16:03:54 helen, could you elaborate 16:04:03 q+ 16:04:04 q+ 16:04:29 This argument should be driven best practices suggested by use cases 16:05:46 q 16:08:48 q- 16:09:12 q- 16:09:22 rrsagent, show actions 16:09:22 I see 7 open action items saved in http://www.w3.org/2006/04/11-hcls-actions.rdf : 16:09:22 ACTION: discuss ontologizing snomed [1] 16:09:22 recorded in http://www.w3.org/2006/04/11-hcls-irc#T15-08-55 16:09:22 ACTION: use case discussion continued [2] 16:09:22 recorded in http://www.w3.org/2006/04/11-hcls-irc#T15-09-10 16:09:22 ACTION: item to BIORDF queries and express them in Snomed... [3] 16:09:22 recorded in http://www.w3.org/2006/04/11-hcls-irc#T15-40-09 16:09:22 ACTION: alan will write up his examples from manchester that show how the "mapping" from SNOMED using hasSubcode would work [4] 16:09:22 recorded in http://www.w3.org/2006/04/11-hcls-irc#T15-54-41 16:09:22 ACTION: davide would prepare some examples from his use case world [5] 16:09:22 recorded in http://www.w3.org/2006/04/11-hcls-irc#T15-54-58 16:09:22 ACTION: don and ken incorporate into current use case [6] 16:09:22 recorded in http://www.w3.org/2006/04/11-hcls-irc#T15-55-20 16:09:22 ACTION: xiaoshu, to look at queries that bioRDF created and see how they would be written in SNOMED [7] 16:09:22 recorded in http://www.w3.org/2006/04/11-hcls-irc#T15-55-56 16:17:26 Amit has joined #HCLS 16:22:22 q 16:39:23 quit 18:30:31 Zakim has left #hcls 19:42:34 JohnMadden has joined #hcls 19:42:38 JohnMadden has left #hcls 20:21:40 JohnMadden has joined #hcls