[topicmapmail] Dictionary
Bernard Vatant
bernard.vatant@mondeca.com
Tue, 5 Mar 2002 18:40:46 +0100
Bob
Lars Marius is definitely a good commercial for PubSubj TC ;-) I just wanted to second him
and invite you to look at what is going on in the TC, and even join the gang if you feel
like it. Dictionaries are definitely the kind of stuff that should be made available as
Published Subjects, and you provide a full-scale use case where we could sort out some
practical issues (such as synonymy and polysemy) ...
Organizing the dictionary to provide on-line users a topic map navigation experience is a
somehow different issue, and I suggest you make clearly distinct those two objectives, and
figure which one you consider as a priority. I tend to suggest that delivering Published
Subjects is a more important task ... but I'm biased somehow :))
Bernard
----- Message d'origine -----
De : "Lars Marius Garshol" <larsga@garshol.priv.no>
À : "Bob Parks" <bobp@lightlink.com>
Cc : <topicmapmail@infoloom.com>
Envoyé : mardi 5 mars 2002 11:16
Objet : Re: [topicmapmail] Dictionary
>
> Hi Bob,
>
> * Lars Marius Garshol
> |
> | Are you asking for advice on how to turn your dictionary into a
> | topic map? How to represent it as a topic map? How to implement
> | searching in it once it has become a topic map? Or how to integrate
> | the dictionary with other data that is in topic map form?
>
> * Bob Parks
> |
> | The answer is "yes" ... ;-) We would like to do all those
> | things. The most important are (1) representing the dictionary as a
> | topic map; and
>
> This need not be very difficult, actually. Judging from your web site
> you could turn each headword into a topic, and each different meaning
> of that headword into a topic. You could then use associations to
> connect the the meaning with the headword. The "part of speech"
> information could be captured by making the headwords instances of
> "noun", "adjective", etc, all of which would be subtypes of "word".
> Phrases would also merit topics of their own.
>
> Definitions, pronunciations, and examples would become internal
> occurrences of those types. (That is: "definition", "pronunciation",
> and "example".)
>
> The links between different headwords you could capture with
> associations of different types.
>
> Probably there is a bit more to your information, but it should be
> easy to represent it as a topic map.
>
> At one point I converted WordNet into a topic map. This was a very
> direct translation of the WordNet structure using our autogeneration
> toolkit, but if you're interested I could dig it up and make the topic
> map available to you.
>
> Your web site could actually quite easily be reimplemented on top of a
> topic map using a topic map web publishing solution and a full-text
> search integration. It wouldn't give you much that you don't already
> seem to have, though, unless you want to expand your service with more
> content.
>
> | The basic perspective behind my interest is to use the dictionary
> | definitions/concepts as a "starter kit" for indexers or creators of
> | topic maps, where the concepts/topics they index may or may not be
> | represented in the dictionary.
>
> Then I understand where you are headed. Yes, this makes a lot of
> sense. Essentially, what you want to do is to create a set of
> so-called published subjects. These are well-defined topics with
> stable identifiers, which can be reused by other topic map creators to
> a) save ontology design time, and b) ensure that their topic maps are
> compatible with those created by others.
>
> The stable identifiers ensure that if two different topic map creators
> both create a topic for the subject "head" (the body part), they can
> both refer to your identifier for that subject (your published
> subject, essentially), and their topic maps will merge correctly.
>
> You may be interested to learn that there is an OASIS technical
> committee that is working on creating general guidelines for how to
> publish sets of published subjects. The work of this TC might make
> your own work a lot easier.
>
> You can find a description of the TC, as well as some of their
> work-in-progress on the page below. Note that you can also quite
> easily join the TC, should you want to.
>
> <URL: http://www.oasis-open.org/committees/tm-pubsubj/index.shtml >
>
> | Hope this helps.... though I may only have communicated my lack of
> | understanding.
>
> Definitely not. You seem very well clued-in to me.
>
> --
> Lars Marius Garshol, Ontopian <URL: http://www.ontopia.net >
> ISO SC34/WG3, OASIS GeoLang TC <URL: http://www.garshol.priv.no >
>
> _______________________________________________
> topicmapmail mailing list
> topicmapmail@infoloom.com
> http://www.infoloom.com/mailman/listinfo/topicmapmail
>