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

Re: Concept-based Search

Expand Messages
  • Flo
    Thanks to everybody who has contributed to this topic so far! @hsavage wrote... ... Wheresoever one might find it, I would never consider NT5 as an instrument
    Message 1 of 34 , Jan 13, 2008
    • 0 Attachment
      Thanks to everybody who has contributed to this topic so far!

      @hsavage wrote...

      > Flo,
      >
      > Back to synonyms, where would one ever find a complete .txt
      > file of synonyms in a form that would be searchable by NoteTab?
      >
      > Such a list would be extremely important in this process,
      > trying to complie your own would be difficult to say the least.

      Wheresoever one might find it, I would never consider NT5 as an
      instrument to deal with a file like that -- possibly some millions of
      words (lemmata plus word forms).

      Text Corpora, Machine Intelligence etc -- very interesting topics,
      indeed. But may I get back to the initial idea of a simple clip that,
      to some humble extent, is able to search for A and will find A plus B
      and C which have a defined semantic relation to A. Synonymy is just
      one of many possible relations like that -- others have been
      mentioned before.

      Let's look at a more manageable job. For me, a practical use would be
      to find stop words or non-descriptors in a text database. Think of a
      medical database that contains the words "Aspirin"
      and "acetylsalicylic acid". In order to use a standard
      terminology, "Aspirin" is not accepted for describing that active
      substance. Consequently, a clip is needed that searches the database
      (exported into a TXT file) for "Aspirin" and -- in some way -- points
      to "acetylsalicylic acid" (for certain reasons I wouldn't simply
      replace it).

      To do that job with NT5, we have to create a list of words which have
      a semantic relation like that. Normally, there won't exist a ready-
      made list -- we have to create it ourselves. On these preconditions,
      we have to look at some details: How to format that word list, how to
      get access to it, how to design an output of search results etc...?

      My intention was to discuss details like that, and possibly get some
      proposals from users who are more experienced in this field than me
      or even have a working model to be used with NT5 -- if possible a
      a "NoteTab-only solution" :-)

      Regards,
      Flo
    • dracorat
      I should also point out that including the pluras es form is sometimes not desired. For example, if we had car|automobile we would get: Sally cares
      Message 34 of 34 , Jan 18, 2008
      • 0 Attachment
        I should also point out that including the pluras "es" form is
        sometimes not desired. For example, if we had car|automobile we would get:

        Sally cares {automobile} about her dog.

        But for that matter, the "s" form has the same issue, just in lower
        quantity. Thus, it's a question of what's better - smaller dictionary,
        or dictionary with every valid permutation.

        (Or even, just do only "s")

        His cat let out a loud hiss.

        his|owner

        His {owner} cat let out a loud hiss {owner}.

        The "s" case would be pretty rare, however. (The "es" not so rare.)

        --Keith

        --- In ntb-clips@yahoogroups.com, "dracorat" <dracorat@...> wrote:
        >
        > I forgot to include the trailing questionmark. Sorry bout that.
        > (Because it's optional to be plural)
        >
        > If you leave off the $2, the plural form will be changed to the
        > singular. The second capture is what plural form we found.
        >
        > --Keith
        > (Happy ho help. - I LOVE regular expressions)
        >
        >
      Your message has been successfully submitted and would be delivered to recipients shortly.