Output Format

[turchmo]

Dear All,
I have run the InvitationModel on my data, but it is not clear to me:

1. the output format (several files have been produced, but it is not clear which one should be considered as final output)
2. where the final output will be produced.
   Please, can you give some details about both issues?
   Thanks a lot!
   Marco

[amirkamran]

Dear Marco,

Sorry, I should have added this detail in the Readme file. So each iteration of the EM produces a output_x.txt with three columns:
setence_number, sentence score for being in in_domain, language module score using in_domain language model

the result is sorted by the in_domain sentence score (second column) and in case of equal scores we sort it using the in_domain lm scores (third column)

You can take the first column of the output_x.txt and retrieve the sentences from mix_domain corpus and select the top n sentences that will be your final output.

[turchmo]

Dear Amir,
thanks a lot for your clarification!
Just a simple doubt, does the sentence number start from 0 or 1?

Thanks a lot!
Marco

[amirkamran]

Hi Marco,
The sentence numbers start from 1.

[turchmo]

Hi Amir,
thanks a lot and sorry for the silly question!
Marco

[amirkamran]

thats a very valid question :) .. and its really nice that you are asking all this. This will help me to improve the documentation.