[prev in list] [next in list] [prev in thread] [next in thread] 

List:       abiword-dev
Subject:    Re: Text relations in ots;
From:       msevior () physics ! unimelb ! edu ! au
Date:       2003-12-02 10:57:12
[Download RAW message or body]

> I added two new functions to the CVS version of OTS:
>
> int ots_text_relations(
> const unsigned char *text1,const unsigned char *lang_code1,
> const unsigned char *text2,const unsigned char *lang_code2,const int
> topic_num);
>
> char* ots_text_topics(
> const unsigned char *text,const unsigned char *lang_code,int topic_num);
>
>
> These two functions tell you how two blocks of text relate to
> one-another. The score given is the number of topics the two share. One
> example of an application is to have a program automatically give score
> to comments on slashdot based on how they relate to the main article;
> The program can tell if the comment is on or off-topic but cannot tell
> if it is smart/funny/informative etc. Another application is to be able
> to bring up similar documents or the one that relates to our document
> the most from a group of documents. I notified the good people at
> dashboard and storage and got positive feedback. I also added an example
> program to the OTS cvs that gives score to the relations between two
> text files.
>

Hi Nadav,
         This reminds me. I need to contact the guys at Storage and
dashboard. I don;t know if they know about AbiWord's ability to
parse *lots*  of document formats and turn them into plain text
for the purposes you've outlined.

I mean the majority of people have their documents in some kind of Word
Processing format. If they're going to index documents based on content
they have to able to open the docs.

Cheers

Martin

>
> -Nadav Rotem
>
>
>
>





[prev in list] [next in list] [prev in thread] [next in thread]