[itsdb] Need help/advise on ItsdbModeling

[halizahb at cs.otago.ac.nz]

Hi,
  I am a PhD student who are doing a research on Statistical Model of
Error Correction in Computer Assisted Language Learning. I have a corpus
of sentences written by English as a second language (ESL) learners. I
have annotated some of the (grammatical) sentences using [incr tsdb()]
tools. So from the annotated sentences, I would like to develop a
probabilistic grammar based on ERG. May I do that with [incr tsdb()] as
well? As for [incr-tsdb], I am a beginner user. I am not quite
understand how to use its features such as Train, Rank and Score.
However, I am able to do annotation, normalization and exporting. My
question is "Do Train, Rank, and Score menus in [incr tsdb()] helps me
in generating probabilistic ERG?"

However, I did also have a try to use the Train menu. I gave a try by
choosing CSLI gold standard and a subset of my annotated sentences
(My-profile) and  then choose Train. Then I get error messages as follows:
/bin/bash: tadm: command not found
train(): unable to read MLM parameters.
I guess here that I need to install TADM (which I will do that later).
However, after that [incr tsdb()] can't be accessed (a cursor becomes a
cross symbol). Then I have to restart it. I noticed that there are two
files created named fc.bdb & fc.mlm created in My-profile folder. I can
only view fc.mlm but I do not quite understand the contents even though I
some of the contents are the feature structures of a sentence.

May anyone can spare time to explain to me about ItsdbModeling whether I
can use it to create a probabilistic grammar. Thank you.

Kind Regards,
Halizah B