[sword-devel] search idea

Trevor Jenkins sword-devel@crosswire.org
Tue, 04 Jan 2000 02:20:31 +0000


On Monday, 3 January, 2000 21:04:20, Matthias Ansorg <aNsis@gmx.de> wrote:


>>My view is that text is text and annotation is annotation. The default would
>>be to search only the text. The search syntax I'm proposing includes "field
>>names" so that items like annotations could be dealt with sensibily.
>
> OK, Agreed. Bible is Bible, annotation is addition. Anyway, with your new
> search system one will get only hits on the bible text by default. Good.

We aim to please. :-)

>>> 3. Names: Imagine the situation you have forgotten the name of a single >>
>>The GBF markup scheme does not appear to have the same feature of
>>distinuishing names.
>
> Might be useful to be implemented in GBF2 during the current redesign.

I'm compliling a list of comments about GBF1 that might be useful for
incorporation into GBF2.

>>(By the way, where is a specification of ThML?)
>
> Try http://www.ccel.org/ThML/ThML1.0.htm

Thanks. At first glance this is a real heavy weight markup scheme in the
sense that all the element names are very long. I'd like to see some
minimisation there. But comments upon ThML will have to wait until I've got
more time.

>>> 5. Meta information:
>>> ... FIND meta.author(Darby) IN modules.books
>>
>>... I do not see a need to distinguish this "meta" data from other
>>associated data provided with a text. To me "meta data" implies the
>>structure of the database itself, i.e. the schema. What mght be useful is a
>>search like FIND modules.meta=(annotation and names and ThML).
>
> "meta information" is an inapropriate expression I used. I tried to describe
> that kind of data that does not belong to any specific part of the text
> itself, such as copyright info or author or title or translator or ISBN of
> printed edition or ... .

Ah understand. I took a database-centric view of meta data. Perhaps
publication data would be a better description of what you describe? Or are
you thinking of other supplementary information being added.

> I am, however, glad that you will already integrate corresponding search
> features in your specifications :-)

I'm bogged down in getting a grammar for the search language that includes
all the features I'm proposing. It's been a few years since I hacked a yacc
grammar together. The last one was for a customer project where I needed to
parser this same search syntax; sadly I didn't keep a copy for my own
reference. :-( I can write an LL(1) parser to deal with this language; even
write it standing on my head but writing that grammar down in a concise
format is taking some time.

> You see my little understanding of what "full-text-retrieval" is.

I think you understand fully what it is. :-)

Regards, Trevor

British Sign Language is not inarticulate handwaving; it's a living
language. So recognise it now.

--

<>< Re: deemed!