Loading ...
Sorry, an error occurred while loading the content.

Re: [TaxoCoP] Incorporating uncertainties into ontologies...

Expand Messages
  • Keipat Patkei
    Thanks, Bob.  Much appreciated. Keith ________________________________ From: Bob DuCharme To: TaxoCoP@yahoogroups.com Sent:
    Message 1 of 11 , Jun 4, 2013
    • 0 Attachment
      Thanks, Bob.  Much appreciated.

      Keith


      From: Bob DuCharme <bducharme@...>
      To: TaxoCoP@yahoogroups.com
      Sent: Friday, May 31, 2013 5:24 PM
      Subject: Re: [TaxoCoP] Incorporating uncertainties into ontologies...

       
      One more standard to think about (and note that PR-OWL is not a standard--I'm pretty familiar with the OWL landscape and this is the first time I've heard of it): The W3C's SKOS, in addition to allowing for storing things like scope notes and information about related terms, includes properties for storing exact match relationships but also broad match and close match relationships. I've never heard of any attempts to quantify how close a non-exact match is and to then use those quantities in application logic (the way that fuzzy logic is used in other application domains) but this does give you some added flexibility in adding data that can be used to enhance searches without asserting precise relationships.

      Bob DuCharme



      On Thu, May 30, 2013 at 1:01 PM, Keipat Patkei <keipat1962@...> wrote:
       
      Thanks for reading this. I hope members of this community might be able to point me to some online responses related to how one goes about "incorporating uncertainties" into ontologies when domain experts are in disagreement regarding the terms, relationships, assertions, and so on that should be part of them.  How does one deal with this and does it lend itself to building "probabilistic ontologies?"  

      I'm sort of stuck in thinking that the general approach to handling this is through consensus building, governance, ontology mapping, use of thesauri, and, perhaps, leveraging PR-OWL (I have very limited knowledge of it, and I suspect it isn't applicable.)  Any thoughts on this?  I don't mind being told I'm way off base here.  Please advise and thanks to all.

      Keith DeWeese




    • Ahren
      Keith, Very good question and equally good responses on this thread. Ontologies are potentially very complicated. It s easy to find yourself treading down a
      Message 2 of 11 , Jun 14, 2013
      • 0 Attachment
        Keith,

        Very good question and equally good responses on this thread.

        Ontologies are potentially very complicated. It's easy to find yourself treading down a path of attempting to categorize all things to all people completely.

        Taking a step back, I have to agree with the opinion that you must carefully scope your vocabulary and then use designated sources (Library of Congress, subject matter expert sources, even Merriam-Webster for straightforward concepts) to define your terms with some consistency. Ontologies are part of the larger definition of "controlled vocabularies" after all, so keep it controlled. Adding too much ambiguity and potential for uncertainties makes the results uncertain, in my opinion.

        You might want to consider your corpus to which this vocabulary will be applied (if it is to be applied to content). Try using some of the fancy text mining tools with good semantic capabilities and attempt to discover where these ambiguities exist and then define or redefine your terms as necessary. This is no small task, but it might be worth the effort.

        I think I'm simplifying a potentially hugely philosophical and complicated issue, but, at the end of the day, you probably have a job to do and need a more practical answer!

        Good luck!

        Ahren

        --- In TaxoCoP@yahoogroups.com, Keipat Patkei <keipat1962@...> wrote:
        >
        > Thanks for reading this. I hope members of this community might be able to point me to some online responses related to how one goes about "incorporating uncertainties" into ontologies when domain experts are in disagreement regarding the terms, relationships, assertions, and so on that should be part of them.  How does one deal with this and does it lend itself to building "probabilistic ontologies?"  
        >
        > I'm sort of stuck in thinking that the general approach to handling this is through consensus building, governance, ontology mapping, use of thesauri, and, perhaps, leveraging PR-OWL (I have very limited knowledge of it, and I suspect it isn't applicable.)  Any thoughts on this?  I don't mind being told I'm way off base here.  Please advise and thanks to all.
        >
        > Keith DeWeese
        >
      Your message has been successfully submitted and would be delivered to recipients shortly.