Open drdhaval2785 opened 7 years ago
So there is:
Huet's is - generative, paradigm based Amba's - Handwritten Dhaval's - generative, rule based
So we get different types of data. It gives sizeable opportunity to correct. I corrected my algorithm close to 6 months before it became usable.
I have not examine Amba's data at all yet. It is premature. In this part of the pysanskrit work, I am focusing on Dhaval's work.
I can see that you have primarily taken the database of Huet for base.
I would phrase it differently. The elisp/python sanskrit work I am doing is based on (a) the Monier-Williams dictionary and (b) the grammars of Antoine, Kale, and others (e.g., Deshpande) as the need presents itself. I think of it as 'model-based'.
Huet's work was used as a first 'reality check' for the pysanskrit work. I have not changed any code on the basis of differences with Huet, but rather am using Dhaval's work for that purpose.
I amend my ambiguous sentence.
I meant - You have primarily taken Huet's database for testing your output.
I can see that you have primarily taken the database of Huet for base. I would like to draw your attention that Huet's database covers only some common verb and common tenses also. It is not exhaustive. You can benefit from using Amba's data at https://github.com/drdhaval2785/SanskritVerb/blob/master/Data/verbforms_amba.txt And also my database - generatedforms20062016.xml which you mentioned in the mail.
I have done extensive comparisions against Huet's and Amba's and corrected my code output. So my database would be an amalgamation of both the earlier ones.