Meeting Record

There are some format problems with the chatlog. Please correct them and reload this page. They are labeled on this page in a red box, like this message.

It may be helpful to

Scribe problem: the name 'dbooth' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown dbooth: dbooth has changed the topic to: OWL and OWL-S Talks by Bijan Parsia ←

Zakim IRC Bot: I do not see a conference matching that name scheduled near this time, dbooth ←

Zakim IRC Bot: ok, dbooth; I see WS_(owl)9:00AM scheduled to start 25 minutes ago ←

Scribe problem: the name 'ht' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Scribe problem: the name 'RalphS' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Scribe problem: the name 'slides' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Scribe problem: the name 'em' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Scribe problem: the name 'plh-owl' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Zakim IRC Bot: On IRC I see timbl, em, caribou, ht, plh-owl, Yves, MattBOS, RalphS, Liam, RRSAgent, Zakim, dbooth ←

Unknown plh-owl: zakim, MIT351 holds Simon, Bijan, John, Ralph, SteveMiles, Tim, David, Philippe ←

Unknown plh-owl: zakim, MIT531 holds Simon, Bijan, John, Ralph, SteveMiles, Tim, David, Philippe ←

Zakim IRC Bot: +Simon, Bijan, John, Ralph, SteveMiles, Tim, David, Philippe; got it ←

Scribe problem: the name 'Liam' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown dbooth: Hmm, slide 4 shows how an HTML link can be viewed as an RDF triple, but neglects the text inside <a>...</a>. I think it would be good to mention that also. ←

Scribe problem: the name 'Bijan' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown Bijan: "SW representation says you have open world assumption and use URIs" [ Scribe Assist by Unknown dbooth ] ←

Scribe problem: the name 'metacrap' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown dbooth: "ontology should be: 1. for people; 2. for programs (sensibly computable), i.e., programmers will know what to do with it" ←

Scribe problem: the name 'timbl' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown Bijan: It seems tha tformalizeing ghe TAG docuemnt realu is our most important task, and the most basic way to eat our dogfood. [ Scribe Assist by Unknown timbl ] ←

Scribe problem: the name 'Tim' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown Tim: in Web Architecture whenever you are given a URI it is reasonable to go dereference it [ Scribe Assist by Unknown RalphS ] ←

Unknown Bijan: but there's nothing that says what you must get when you do that operation [ Scribe Assist by Unknown RalphS ] ←

Unknown RalphS: [re: "owl:imports being the only way external meanings get into your ontology"] ←

Scribe problem: the name 'Q' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown Q: What are some of the ontologies that have been built? [ Scribe Assist by Unknown dbooth ] ←

Scribe problem: the name 'BIjan' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown BIjan: THis afternoon i'll talk about ome for Web services -- planning, matchmaking, etc. [ Scribe Assist by Unknown dbooth ] ←

Unknown dbooth: ... MOst popular now is FOAF, using inverseFunctional property. (FOAF = Friend Of A Friend) ←

Unknown dbooth: ... Galan has been migrated to OWL. National Cancer INstitute has one that is migrating to OWL. ←

Unknown Q: Taxonomy versus ontology. In my domain we have lots of taxonomies, and there seems to be benefit in converting to ongologies. [ Scribe Assist by Unknown dbooth ] ←

Unknown BIjan: Ask: What's my application? THen ask: What inferences do I need? [ Scribe Assist by Unknown dbooth ] ←

Unknown dbooth: ... Being open world surprises people -- most people work with databases, where if something is missing from your knowledge base, it's false. You get used to working like that. ←

Unknown dbooth: ... Some of my students ran into problems with a pizza ontology: olives and onions might be the same thing for all the reasoner knows! ←

Unknown dbooth: ... Re: KR for the Web, when do you do it? People sit down to write a Web page. Do they sit down to write a KR? ←

Unknown dbooth: ... Best advice: Get someone who has done it who can sit down with you to ask you what you want to get out of it, and proceed from there. ←

Unknown Q: What about automated ontology building? [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: Induction from instances -- not that bad. Translating from schemas. If you have an ontology that's pretty rich already, andyou have DB schemas, you can give a few things and it can be partially automated, but not flip a switch. [ Scribe Assist by Unknown dbooth ] ←

Zakim IRC Bot: Liam, you wanted to ask about relationship between OWL and ISO topic maps (if any) ←

Unknown Q: What about relationship between OWL and ISO TopicMaps? [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: My feeble understanding: there might be a happy marriage, but that's a hand-wavy answer. [ Scribe Assist by Unknown dbooth ] ←

Unknown Q: In open world assumption, how do you deal with mistakes? [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: When I say open world assumption, I don't mean modifying the KB. I mean when you reason, you don't assume that inability to prove something means that it's false. [ Scribe Assist by Unknown dbooth ] ←

Unknown dbooth: ... After you fix one thing, you find unsatisfiable constants, then fix that, and then discover that what you're trying to model is a mess. ←

Scribe problem: the name 'TimBL' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown TimBL: ANy agent is not trying to model the entire Web. It will look at a small piece and do a limited amount of stuff. If you're trying to index it all, you're probably not trying to use it all. The reason you use trust in a specific thing are very app dependent. [ Scribe Assist by Unknown dbooth ] ←

Unknown Q: No way to encode provinence -- where the assertion came from. [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: One kind of problem you can't do anything about: Modeling errors. FOr example, assumption of two sexes. What about hermaphrodites? [ Scribe Assist by Unknown dbooth ] ←

Unknown dbooth: ... There are ways of dealing with inconsistencies without destrying the knowledge. ←

Unknown dbooth: ... Talk about only part of your KB, or only part of a class that I care about. ←

Scribe problem: the name 'Carine' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Scribe problem: the name 'Ht' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Scribe problem: the name 'EMiller' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Scribe problem: the name 'Slides' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown timbl: Semantci web to support web services or vice-versa..... but is thi sall self-referntial? ←

Unknown Bijan: it is hard to represent dynamics without some heavyweight vocabulary [ Scribe Assist by Unknown RalphS ] ←

Unknown Bijan: Encoding in OWL is messy, for one thing because of the OWA, which you don't normally have in BNF! [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: Think of Service as a service description [ Scribe Assist by Unknown dbooth ] ←

Unknown dbooth: ... Profile is the outer face to the world, for discovery, etc. ←

Unknown dbooth: ... Grounding is how to invoke the service -- connects to WSDL. Groundign will probably disappear when the mapping from WSDL to RDF comes into being. ←

Scribe problem: the name 'Ralph' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown Ralph: You may not need the Grounding for planning, but you may for other purposes. [ Scribe Assist by Unknown dbooth ] ←

Unknown RalphS: i.e. the wire protocol details aren't so critical for planning, but other parts of the system sure care about having that detail available ←

Unknown Q: What about equivalent class capability? I could imagine it to be useful. [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: Wait til you see the problems with it! [ Scribe Assist by Unknown dbooth ] ←

Unknown TimBL: Can model the message as a publication (to an individual), like in the REST model [ Scribe Assist by Unknown dbooth ] ←

Unknown TimBL: Should people be shipping around RDF collections all the time? Could say: My children set is: .... [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: Not in OWL. Can't represent lists in OWL. You can only represent things that LOOK like lists. There's nothing that says that this list (of my children) in my DB is not necessarily a list in in the REAL world of 3 members. [ Scribe Assist by Unknown dbooth ] ←

Unknown TimBL: You require as input an invoice with 3 fields. I say this is my invoice, with the rest, the rest, the rest, nil. [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: If you give me a KB with 4 individual and they're related by the hasChild, you have a contradiction. [ Scribe Assist by Unknown dbooth ] ←

Unknown dbooth: ... Part of validation is to detect MISSING information, and because of the OWA, you cannot detect that in OWL. ←

Unknown dbooth: ... E.g., if I say someone has 3 kids, but I don't tell you who they are, then you have no contradiction. ←

Unknown dbooth: ... But if I list 4 specific children, then you have a contradiction. ←

Unknown dbooth: ... Another thing not in owl: Define a person as a person with a parent. Then feed in a geneology. Might ask about my great-great-grandparents, for example. How much of the geneology do I provide? ←

Unknown Bijan: Need to be careful saying that the input is a person, because in OWL it's a REAL person! Need to be clear that the input is a representation of a person. [ Scribe Assist by Unknown dbooth ] ←

Unknown dbooth: ... Even if you say that the input is a message *about* a person, it still doesn't get you the data validation that you want. ←

Unknown dbooth: "Successes in software configuration world -- McGuinness's(?) work" ←

Scribe problem: the name 'VivienSimon' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown Bijan: slide 30, bullet 1 "Given a state of the world..." belongs on slide 29 [ Scribe Assist by Unknown RalphS ] ←

Unknown RalphS: [I wonder what is the significant difference between 'goal' and 'task'] ←

Unknown Q: Is the ServiceProfile a static document? Could it change depending on the load on a machine? [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: Yes. You could update your KB. [ Scribe Assist by Unknown dbooth ] ←

Scribe problem: the name 'Stuart' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown Q: Could negotiation occur through the Profile? [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: Profiles don't inherently have negotiation -- they're just descriptions. Usually you'd haggle over points of the profile. I could give one profile to preferred customers and another to others. [ Scribe Assist by Unknown dbooth ] ←

Unknown Q: You said no data validation, what are the issues? [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: Defining your ontologies with millions of instances. Spreading: for a web of ontologies, you don't want a kabal of ontology writers. You may want teachers to develop an ontology through a class with the students -- more bottom up and popular. Not sure it's possible to be popular. [ Scribe Assist by Unknown dbooth ] ←

Unknown dbooth: ... And issue of imports: is it slurping in the whole thing, or is there a way to share only part? ←

Unknown dbooth: ... Might allow distinct ontologies from different sources with some interaction between them. ←

Unknown TimBL: Like creating The Big Ontology -- when it's done, the world stops [ Scribe Assist by Unknown dbooth ] ←

Scribe problem: the name 'marja' does not match any of the 3 active names. Either change the name used, or request the list of names be altered.Active names: Zakim IRC Bot Trackbot IRC Bot RRSAgent IRC Bot

Unknown marja: (people always have different views of the world and those also evolve) ←

Unknown Bijan: I still think of ontologis as the place where people and programs interact. Interesting challenge: write ontologies and code that does interesting things with them. [ Scribe Assist by Unknown dbooth ] ←

Zakim IRC Bot: leaving. As of this point the attendees were Liam, Ht, Carine, EMiller, Simon, Bijan, John, Ralph, SteveMiles, Tim, David, Philippe, 19_people+Bijan, +1.617.395.aaaa, Marja, ←

Unknown dbooth: dbooth has changed the topic to: OWL-S Hands-On starting 1:30pm ET ←

Unknown dbooth: Useful test ontology: Galen (Medical Surgical ontology) opengalen.org ←

Unknown Q: Best way to apply SW tech to WS? [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: Network Inference is supposed to provide big ontologies to help with matching. [ Scribe Assist by Unknown dbooth ] ←

Unknown Q: Amount of detail: Too tedious? [ Scribe Assist by Unknown dbooth ] ←

Unknown Bijan: I hope the WSDL->RDF mapping will allow us to say interesting stuff. Understanding about how WSDL documents are related. [ Scribe Assist by Unknown dbooth ] ←

Unknown dbooth: "Use of Classic (a DL) at AT&T for configuration was a big win." ←

Owl Intro by Bijan Parsia

Minutes of 30 April 2004

1. Owl Intro by Bijan Parsia

2. OWL-S Talk by Bijan Parsia