Evidence for backing statements

Thread: Evidence for backing statements

Search this list only Search all lists
Evidence for backing statements
United Kingdom	2007-05-18 03:21:14
I've been lurking & reading the discussion with interest. It might be worth pointing out that there is an ongoing attempt to classify/ represent evidential links/ weight/ etc. started in the legal domain by people such as Wigmore and continued by people such as David Schum & William Twining. There's currently a Leverhulme-sponsored research programme on "Evidence Science", centered at UCL, bond. Such efforts don't seem to easily map to rdf (they're often based on Bayesian models), but might provide some inspiration, although some of the legal niceties may be unnecessary. I also know that the issue has been looked at by some in the argumentation community, where the "source" of the rules that make the argument need to be defined. If anyone is interested in these, please contact me off-list. Thanks, Matt -- http://acl.icnet.uk/~mw http://adhominem.blogs ome.com/ +44 (0)7834 899570

Re: Evidence for backing statements
United Kingdom	2007-05-18 14:55:58
Matt Williams wrote: > > I've been lurking & reading the discussion with interest. > > It might be worth pointing out that there is an ongoing attempt to > classify/ represent evidential links/ weight/ etc. started in the legal > domain by people such as Wigmore and continued by people such as David > Schum & William Twining. There's currently a Leverhulme-sponsored > research programme on "Evidence Science", centered at UCL, bond. > > Such efforts don't seem to easily map to rdf (they're often based on > Bayesian models), but might provide some inspiration, although some of > the legal niceties may be unnecessary. Interesting! There's also another related group here in W3Cland - an incubator on Uncertainty Reasoning for the Web, see http://www.w3. org/2005/Incubator/urw3/ ...though things are only just starting up. The only mailing list traffic so far is about scheduling the first telecon. The charter has some notes on what they're doing: http:// www.w3.org/2005/Incubator/urw3/charter ...and Bayes gets a few mentions there. My inclination with RDF and evidence/probability, ... is that without reinventing the RDF graph model, it is likely easier to attach probability and other annotations to collections of statements, rather than to individual triples. This can be done for example by making assertions about an RDF/XML document, ... and is somehow related to the ability in SPARQL to associate a graph with a URI. For example see h ttp://www.w3.org/TR/rdf-sparql-query/#queryDataset cheers, Dan

Re: Evidence for backing statements
United Kingdom	2007-05-18 15:16:51
Dear Dan, That is interesting. My thesis (3/4 done) is on integrating Ontologies (using OWL-DL or OWL 1.1 as an exemplar) with Argumentation (which is an extension of previous work in NM logics) to represent conflict. I'm using Breast Cancer therapy choice as a domain. I'll have a look at the links you sent and probably be in touch.... Matt Dan Brickley wrote: > Matt Williams wrote: >> >> I've been lurking & reading the discussion with interest. >> >> It might be worth pointing out that there is an ongoing attempt to >> classify/ represent evidential links/ weight/ etc. started in the >> legal domain by people such as Wigmore and continued by people such as >> David Schum & William Twining. There's currently a >> Leverhulme-sponsored research programme on "Evidence Science", >> centered at UCL, bond. >> >> Such efforts don't seem to easily map to rdf (they're often based on >> Bayesian models), but might provide some inspiration, although some of >> the legal niceties may be unnecessary. > > Interesting! There's also another related group here in W3Cland - an > incubator on Uncertainty Reasoning for the Web, see > http://www.w3. org/2005/Incubator/urw3/ ...though things are only just > starting up. The only mailing list traffic so far is about scheduling > the first telecon. > The charter has some notes on what they're doing: > http:// www.w3.org/2005/Incubator/urw3/charter ...and Bayes gets a few > mentions there. > > My inclination with RDF and evidence/probability, ... is that without > reinventing the RDF graph model, it is likely easier to attach > probability and other annotations to collections of statements, rather > than to individual triples. This can be done for example by making > assertions about an RDF/XML document, ... and is somehow related to the > ability in SPARQL to associate a graph with a URI. For example see > h ttp://www.w3.org/TR/rdf-sparql-query/#queryDataset > > cheers, > > Dan > > > -- http://acl.icnet.uk/~mw http://adhominem.blogs ome.com/ +44 (0)7834 899570

RE: Evidence for backing statements
United States	2007-05-18 18:40:21

This would be a good time to begin collecting use-cases from HCLS for the URW (Uncertainty Reasoning for the Web, http://www.w3.org/2005/Incubator/urw3/) incubator group involving uncertainty, beliefs and probabilities... I see several potential categories of uncertainty, some around the current reification discussions; here's a starter list that is by no means exhaustive: 1. Annotations and belief statements: these may always be uncertain (though over time their uncertainty may reduce), yet they may never have an globally agreed probability associated with them. e.g., ; It would depend on how much a particular consumer values a particular annotation coming form a particular research group-- let each consumer determine there own certainty weighting... 2. A data set from clinical findings or microarrays that is analytically processed using a statistical package, and the resulting relations may have confidence intervals and p-values associated with them (student pair test, Z-scores, non-parametric ests, F-statsitic cluster scores, etc). 3. OWL-based classifications that are proposed but not fully-tested or validated yet, e.g., whether patient Z's disease symptom W are indicative of a new disease class or not. This will be closely aligned with evolving ontologies. 4. ??? Again, I'll mention my personal view that within HCLS, most RDF statements (possibly all) are really beliefs that may eventually be proven false. Hence we would need to explicitly state that such RDF statements follow KD45 modal logic constraints (one can say something that may be later proven false, but for now the author honestly does not know this), rather than S5 (only truthful statements can be made). In other words, am I correct to assume that within HCLS, all RDF statements we are considering are not facts, but assertions, that may in the future be proven false, but never proven true? Eric
Re: Evidence for backing statements
United States	2007-05-19 06:11:32

On May 18, 2007, at 7:40 PM, Eric Neumann wrote: am I correct to assume that within HCLS, all RDF statements we are considering are not facts, but assertions, that may in the future be proven false, but never proven true? I am very excited to hear this will be a W3C focused activity. Statistical techniques of all sorts - Bayesian especially - are critical data reduction and analysis tools driving interpretation in all areas of biomedical science and clearly there needs to be some way for representational techniques to interoperate with the derived probabilistic analyses. My tendency is toward the sort of link Chris Mungall mentioned earlier, whereby the statistics is linked indirectly, as opposed to being intrinsic to the represented assertions. I agree with your proposal, except for the closing statement. It's not clear to me under what circumstances the following assertions given as an OWL "defining" relations would be useful to consider as something other than fact: "All viable eukaryotic cells have functional mitochondria located inside them" "Presynaptic vesicle fusion in neurons leads to release of small molecule and neuropeptide neurotransmitters into the extracellular space." Establishing facts such as this obviously have more than a pedagogical purpose. They are an important part of the network of assertions used to drive inference, and - as facts - they need to be accorded a different role in the inferencing process than assertions that cannot be established as "universals". I'm probably being overly naive in stating this view. If that is so, I welcome others with greater in depth knowledge of the logical formalisms and implemented tools to explain why one would not want to make this distinction between the superset of RDF assertions and that subset expressing established fact. This is a subset whose absolute size continues to grow - though its relative size compared to all assertions is clearly decreasing at a much faster rate given all the high-throughput experimental techniques introduced in the last 30 years. Having said this, I do agree it would be very much mutually beneficial for members of the HCLS IG to provide use cases for this uncertainty reasoning group to examine. Cheers, Bill Bill Bug Senior Research Analyst/Ontological Engineer Laboratory for Bioimaging & Anatomical Informatics www.neuroterrain.org Department of Neurobiology & Anatomy Drexel University College of Medicine 2900 Queen Lane Philadelphia, PA 19129 215 991 8430 (ph) 610 457 0443 (mobile) 215 843 9367 (fax) Please Note: I now have a new email - William.BugDrexelMed.edu">William.BugDrexelMed.edu
Re: Evidence for backing statements
	2007-05-19 14:55:23

Eric -- You wrote...   I see several potential categories of uncertainty, some around the current reification discussions; here's a starter list that is by no means exhaustive: ........ 4. ??? Here's my 2 cents worth for item 4. I'd say that an application or a reasoner that uses RDF or other data can supply natural language phrases to indicate the real world meaning and (un)certainty of what it has found.  Such as --       "Arif Durrani could have met with the following person(s) near LAX from 200610060900 to 200610061945 : Mohammed Al-Bisri , Samir Hakim";       "as far as is known at 200705119 1529 the contributors to the discussion are Matt, Adrian, Dan, Chris, Eric" ,   "based on past experience, it is likely that at least one further W3C member will add to the discussion"   "according to the popular literature, red wine is good for you, but the refereed literature mostly regards this as unproven" and so on. It just so happens that this kind of thing works rather well in the online system [1]. In particular, English explanations from the system allow one to drill down into the support for a hedged result, in English, to see in detail if one agrees with the hedging. The key requirement is to have a completely automated round trip linkage between English and some form of executable logic.  In general this is regarded as an "AI-complete" problem, but there are automated work-arounds that are useful in practice, such as in [1]. Some examples you can view, run and change:   www.reengineeringllc.com/demo_agents/ArmsDealerMeeting1.agent   www.reengineeringllc.com/demo_agents/BlackScholes1.agent   www.reengineeringllc.com/demo_agents/MedMine2.agent     www.reengineeringllc.com/demo_agents/Pc-Os-Ontology1.agent ; www.reengineeringllc.com/demo_agents/RDFQueryLangComparison1.agent     www.reengineeringllc.com/demo_agents/SemanticResolution2.agent   www.reengineeringllc.com/demo_agents/Oil-IndustrySupplyChain1Sqlmysql.agent Hope this helps.       ;         ;         ;         ;     Cheers,  - Adrian [1] Internet Business Logic A Wiki for Executable Open Vocabulary English Online at www.reengineeringllc.com   Shared use is free Adrian Walker Reengineering On 5/18/07, Eric Neumann < eneumannteranode.com">eneumannteranode.com> wrote: This would be a good time to begin collecting use-cases from HCLS for the URW (Uncertainty Reasoning for the Web, http://www.w3.org/2005/Incubator/urw3/) incubator group involving uncertainty, beliefs and probabilities... I see several potential categories of uncertainty, some around the current reification discussions; here's a starter list that is by no means exhaustive: 1. Annotations and belief statements: these may always be uncertain (though over time their uncertainty may reduce), yet they may never have an globally agreed probability associated with them. e.g., ; It would depend on how much a particular consumer values a particular annotation coming form a particular research group-- let each consumer determine there own certainty weighting... 2. A data set from clinical findings or microarrays that is analytically processed using a statistical package, and the resulting relations may have confidence intervals and p-values associated with them (student pair test, Z-scores, non-parametric ests, F-statsitic cluster scores, etc). 3. OWL-based classifications that are proposed but not fully-tested or validated yet, e.g., whether patient Z's disease symptom W are indicative of a new disease class or not. This will be closely aligned with evolving ontologies. 4. ??? Again, I'll mention my personal view that within HCLS, most RDF statements (possibly all) are really beliefs that may eventually be proven false. Hence we would need to explicitly state that such RDF statements follow KD45 modal logic constraints (one can say something that may be later proven false, but for now the author honestly does not know this), rather than S5 (only truthful statements can be made). In other words, am I correct to assume that within HCLS, all RDF statements we are considering are not facts, but assertions, that may in the future be proven false, but never proven true? Eric
RE: Evidence for backing statements/Wiki page for HCLS Uncertainty Use Cases
United States	2007-05-23 11:47:02

Hi, I participated on the URW3 Telcon today and volunteered to collect the HCLS uncertainty use cases. Have set up a wiki page for HCLS Uncertatinty use cases is available at: http://esw.w3.org/topic/HCLS/UncertaintyUseCases Please add more use cases and annotate existing ones with your suggestions and feedback. An interesting exercise being proposed in URW3 is to create an “Uncertainty Ontology” and my suggestion was that HCLS use cases could provide the motivations and driving force. This could be an interesting collaboration opportunity across the two groups. Cheers, ---Vipul ======================================= Vipul Kashyap, Ph.D. Senior Medical Informatician Clinical Informatics R&D, Partners HealthCare System Phone: (781)416-9254 Cell: (617)943-7120 http://www.partners.org/cird/AboutUs.asp?cBox=Staff&stAb=vik To keep up you need the right answers; to get ahead you need the right questions ---John Browning and Spencer Reiss, Wired 6.04.95 From: public-semweb-lifesci-requestw3.org [mailto:public-semweb-lifesci-requestw3.org] On Behalf Of Eric Neumann Sent: Friday, May 18, 2007 7:40 PM To: Chris Mungall; Dan Brickley Cc: Matt Williams; public-semweb-lifesci hcls Subject: RE: Evidence for backing statements This would be a good time to begin collecting use-cases from HCLS for the URW (Uncertainty Reasoning for the Web, http://www.w3.org/2005/Incubator/urw3/) incubator group involving uncertainty, beliefs and probabilities... I see several potential categories of uncertainty, some around the current reification discussions; here's a starter list that is by no means exhaustive: 1. Annotations and belief statements: these may always be uncertain (though over time their uncertainty may reduce), yet they may never have an globally agreed probability associated with them. e.g., ; It would depend on how much a particular consumer values a particular annotation coming form a particular research group-- let each consumer determine there own certainty weighting... 2. A data set from clinical findings or microarrays that is analytically processed using a statistical package, and the resulting relations may have confidence intervals and p-values associated with them (student pair test, Z-scores, non-parametric ests, F-statsitic cluster scores, etc). 3. OWL-based classifications that are proposed but not fully-tested or validated yet, e.g., whether patient Z's disease symptom W are indicative of a new disease class or not. This will be closely aligned with evolving ontologies. 4. ??? Again, I'll mention my personal view that within HCLS, most RDF statements (possibly all) are really beliefs that may eventually be proven false. Hence we would need to explicitly state that such RDF statements follow KD45 modal logic constraints (one can say something that may be later proven false, but for now the author honestly does not know this), rather than S5 (only truthful statements can be made). In other words, am I correct to assume that within HCLS, all RDF statements we are considering are not facts, but assertions, that may in the future be proven false, but never proven true? Eric The information transmitted in this electronic communication is intended only for the person or entity to whom it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination or other use of or taking of any action in reliance upon this information by persons or entities other than the intended recipient is prohibited. If you received this information in error, please contact the Compliance HelpLine at 800-856-1983 and properly dispose of this information.

[1-7]

about | contact Other archives ( Real Estate discussion Medical topics )