[lojban-beginners] Re: Lojban parser and graph generator

["John Leuner" <jewel@subvert-the-dominant-paradigm.net>]

Yes, it's much more useful if the user can see the original text somehow
related to the structure.

Were you looking at the full parse tree? With the full parse tree you can
work from the bottom up (where each letter in the original input forms its
own node) and see how things are grouped.

A quick way to do this is to simply add a second line to each label, such
that you have:

gismu
bakni

or

selbri-1
na certu

The problem with this is that you have redundancy, the same lojban text
will be reproduced many times on the graph.

I am still trying to figure out which parts of the grammar are the ones
which users really want to see. If I can make useful set of those, it
makes sense to attach the lojban text to each node.

It might also be possible to generate a box/table structure, something like:

                         text
                       sentence
          sumti                 selbri
           mi                   na certu

(perhaps an english gloss in another row below the lojban text).

in this table each cell would be contained in it's parent cell. Perhaps
nested HTML tables could achieve this?

Also see the graphviz labels (page 8 of
http://www.graphviz.org/Documentation/dotguide.pdf).

John

> [snip]
>
> Pardon me for being dense. I love this idea, but I don't see
> the usefulness of the way it's implemented.
>
> When it creates a graph of nodes, they just have generic labels,
> with no clue what node goes with what node in the original
> sentence.
>
> Am I missing something?
>
>
> Hal
>
>
>
>
>