From: Shawn Mikula <brainmaps@gm...>  20060905 05:24:43
Attachments:
Message as HTML

can someone please explain why DAG's are necessary or desirable in ontologies? Why can't recurrent relations be used, or why are they discouraged (besides the fact that they generate cycles)? For example, say I have a set of brain areas, A and B, that are distinct entities and I want to denote anatomical connections, A>B and B>A; I can't do this directly with a DAG unless I specify classes, A>B and B>A, which seems undesirable since it involves needless multiplicity. Can anyone help me out with this? Thanks. Shawn  Shawn Mikula, Ph.D., Postdoctoral Scholar Center for Neuroscience University of CaliforniaDavis, 1544 Newton Court, Davis, CA 95618, Phone: 5307549209 Fax: 5307549136 mail: samikula@... web: http://brainmaps.org 
From: Smith, Barry <phismith@bu...>  20060905 13:41:34

Shawn I think you may be right, for those relations which admit of symmetric cases (note that not every case where anatomical entity of type A is connected to anatomical entity of type B will be such that the connection relation holds also from B to A). However for binary symmetric cases we can if necessary define a new relation of symmetric connectedness. More interesting would be a cycle with three terms A>B, B>C, C>A, if you have any examples. Barry At 04:25 AM 9/5/2006, you wrote: >can someone please explain why DAG's are necessary or desirable in >ontologies? Why can't recurrent relations be used, or why are they >discouraged (besides the fact that they generate cycles)? For >example, say I have a set of brain areas, A and B, that are distinct >entities and I want to denote anatomical connections, A>B and B>A; >I can't do this directly with a DAG unless I specify classes, A>B >and B>A, which seems undesirable since it involves needless >multiplicity. Can anyone help me out with this? >Thanks. >Shawn > > > > >Shawn Mikula, Ph.D., >Postdoctoral Scholar >Center for Neuroscience >University of CaliforniaDavis, >1544 Newton Court, >Davis, CA 95618, >Phone: 5307549209 >Fax: 5307549136 >mail: <mailto:samikula@...>samikula@... >web: <http://brainmaps.org>http://brainmaps.org > > > >Using Tomcat but need to do more? Need to support web services, security? >Get stuff done quickly with preintegrated technology to make your job easier >Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo >http://sel.asus.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 >_______________________________________________ >Obodiscuss mailing list >Obodiscuss@... >https://lists.sourceforge.net/lists/listinfo/obodiscuss 
From: Melissa Haendel <mhaendel@uo...>  20060905 14:37:59

Hi Barry and Shawn, I believe this is similar to a symmetric case we've been discussing that of homology, which hasn't yet been defined. In this case though, A>B necessarily means B>A. This type of relation might be represented between two terms in the same ontology, or between two ontologies. We've also been discussing another 'horizontal' relation between ontologies, that of stage and anatomical structure. This type of relation does not have reciprocal meaning. I am not an ontologist, but your problem brought up some similar examples. More food for thought. Melissa Haendel > Shawn > I think you may be right, for those relations which admit of > symmetric cases (note that not every case where anatomical entity of > type A is connected to anatomical entity of type B will be such that > the connection relation holds also from B to A). However for binary > symmetric cases we can if necessary define a new relation of > symmetric connectedness. More interesting would be a cycle with three > terms A>B, B>C, C>A, if you have any examples. > Barry > > > At 04:25 AM 9/5/2006, you wrote: >>can someone please explain why DAG's are necessary or desirable in >>ontologies? Why can't recurrent relations be used, or why are they >>discouraged (besides the fact that they generate cycles)? For >>example, say I have a set of brain areas, A and B, that are distinct >>entities and I want to denote anatomical connections, A>B and B>A; >>I can't do this directly with a DAG unless I specify classes, A>B >>and B>A, which seems undesirable since it involves needless >>multiplicity. Can anyone help me out with this? >>Thanks. >>Shawn >> >> >> >> >>Shawn Mikula, Ph.D., >>Postdoctoral Scholar >>Center for Neuroscience >>University of CaliforniaDavis, >>1544 Newton Court, >>Davis, CA 95618, >>Phone: 5307549209 >>Fax: 5307549136 >>mail: <mailto:samikula@...>samikula@... >>web: <http://brainmaps.org>http://brainmaps.org >> >> >> >>Using Tomcat but need to do more? Need to support web services, security? >>Get stuff done quickly with preintegrated technology to make your job >> easier >>Download IBM WebSphere Application Server v.1.0.1 based on Apache >> Geronimo >>http://sel.asus.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 >>_______________________________________________ >>Obodiscuss mailing list >>Obodiscuss@... >>https://lists.sourceforge.net/lists/listinfo/obodiscuss > > > >  > Using Tomcat but need to do more? Need to support web services, security? > Get stuff done quickly with preintegrated technology to make your job > easier > Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo > http://sel.asus.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 > _______________________________________________ > Obodiscuss mailing list > Obodiscuss@... > https://lists.sourceforge.net/lists/listinfo/obodiscuss > 
From: chris mungall <cjm@fr...>  20060905 15:47:05

Hi Shawn A DAG is a useful abstraction when the ontology consists of acyclic, nonsymmetric relations. Even in the cases where one has cycles and symmetric relations, a directed graph is still a useful abstraction  you could simply have a rule such that if R is symmetric, then X R Y implies Y R X. There is nothing forcing you to visualise or abstract the ontology as a graph  although almost all formalisations come down to entities and predicates/relations which map to nodes and edges. Can you provide an alternate formalisation? I'm not sure how recurrence relations solve the problem, insofar as there is a problem. Cheers Chris On Sep 4, 2006, at 10:24 PM, Shawn Mikula wrote: > can someone please explain why DAG's are necessary or desirable in > ontologies? Why can't recurrent relations be used, or why are they > discouraged (besides the fact that they generate cycles)? For > example, say I have a set of brain areas, A and B, that are > distinct entities and I want to denote anatomical connections, A>B > and B>A; I can't do this directly with a DAG unless I specify > classes, A>B and B>A, which seems undesirable since it involves > needless multiplicity. Can anyone help me out with this? > Thanks. > Shawn > > > >  > Shawn Mikula, Ph.D., > Postdoctoral Scholar > Center for Neuroscience > University of CaliforniaDavis, > 1544 Newton Court, > Davis, CA 95618, > Phone: 5307549209 > Fax: 5307549136 > mail: samikula@... > web: http://brainmaps.org > > >  >  > Using Tomcat but need to do more? Need to support web services, > security? > Get stuff done quickly with preintegrated technology to make your > job easier > Download IBM WebSphere Application Server v.1.0.1 based on Apache > Geronimo > http://sel.asus.falkag.net/sel? > cmd=lnk&kid=120709&bid=263057&dat=121642______________________________ > _________________ > Obodiscuss mailing list > Obodiscuss@... > https://lists.sourceforge.net/lists/listinfo/obodiscuss 