Parser for new language

[assamXpress]

I want to develop a parser for one of the Indian language using Link
grammar. I go through the manuals and pages provided by link grammar
site. Can you please give me the starting point or the road map or
flowchart to develop the parser. It will be extremely helpful for me.
Thanking you.

[Linas Vepstas]

The first step would be to download, compile, install, run the parser.
Play with it long enough to get familiar. Be sure to try to !help and
the !var commands at the parser command line.

Next, create a new data directory, e.g. data/hindi -- this would contain
the dictionary files. You can specify the location of this when starting up
the link-parser: so, for example:

link-parser data/en

It should print something stating that it found (or failed to find) your
data files.

For the actual data files, you need to create a 4.0.dict, 4.0.affix and
a few others. See for example data/lt which is an experimental dictionary
for lithuanian-- it has maybe a few dozen words from the lithuanian language,
and an assortment of link types for joining them together into valid sentences.

You should probably copy the contents of data/lt to data/hindu, and start
hacking from there ...

Let me know when you get to this point; we can have a more detailed
discussion then.

--linas