---------- Forwarded message ----------
From: Parya Razmdide <[email protected]>
Date: Mon, Dec 5, 2016 at 1:00 PM
Subject: Fwd: To solve writing interrogative transfer rules
To: Felipe Sánchez Martínez <[email protected]>



---------- Forwarded message ----------
From: Parya Razmdide <[email protected]>
Date: Fri, Dec 2, 2016 at 4:30 AM
Subject: Re: To solve writing interrogative transfer rules
To: "Víctor M. Sánchez-Cartagena" <[email protected]>


To whom it may consider,
Actually we couldn't run ruLean and if possible i send you our files or
share our files in your repository to help us to solve this problem. We
tries a lot but the exact reason its not clear.
The next matter is that we decided to use another way parallel to inferring
transfer rules from our hybrid system via ruLearn, that is using Active
Leaning theory on Apertium. At first we started from our dictionaries and
based on the criterion of Frequency, we extracted all missing words and i
am giving these words to my dictionaries frequently. Then  the dev-file is
evaluated each time by PER and WER metrics. After the lexicon finished we
are going to test Active learning for transfer rules. But at present we are
not sure how the missed and essential how transfer rules should be added or
edited. At present we are not sure how they are extracted, may be by
considering their PoS pattern of the missing rules.
Please tell us your idea and direction to solve our first problem to run
ruLearn? Then we are pleased to know your suggestion on using Active
Learning theory on Apertium?
Also Ive already sent this email to Víctor, unfortunately i didnt receive
any answer from him.
I hope to hear from you so soon.
Best wishes,
Parya Razmdideh
------------------------------------------------------------------------------
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff

Reply via email to