search

OmarBoudraa / PUNET

GNU General Public License v3.0
3 stars 3 forks source link

evaluation of MAP #1

Open cszjing opened 2 years ago

cszjing commented 2 years ago

I am intrested in your paper, so I have read your paper and code about PUNet. Now, I have some problems about relative codes in calculating the MAP. image I do not understand the following code : "result = 1. - (mean_avg_prec[0] * 1. / mean_avg_prec[1])" Could you please help me solve my confusion. Thank you for your help.

OmarBoudraa commented 2 years ago

Thank you for your interest in my paper, mean_avg_prec[1] : count the number of words mean_avg_prec[0] : sum the number of (1-precision) of each word Best regards

Le ven. 13 mai 2022 à 20:40, zj @.***> a écrit :

I am intrested in your paper, so I have read your paper and code about PUNet. Now, I have some problems about relative codes in calculating the MAP. [image: image] https://user-images.githubusercontent.com/54255578/168345498-1300082c-ab6f-4b21-9591-75577ef55a90.png I do not understand the following code : "result = 1. - (mean_avg_prec[0] * 1. / mean_avg_prec[1])" Could you please help me solve my confusion. Thank you for your help.

— Reply to this email directly, view it on GitHub https://github.com/OmarBoudraa/PUNET/issues/1, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMHUMSFFZMCE5YDJ5VUA4FLVJ2OYNANCNFSM5V4FJDQA . You are receiving this because you are subscribed to this thread.Message ID: @.***>

-- Omar BOUDRAA

cszjing commented 2 years ago

Sorry to bother you again. I still don't understand your above answer, especially about mean_avg_prec[0]. I have computed the mAP using formulation which originates from the task of Information Retrieval in a little test dataset. The formulation is as follows : image But I got the different result of mAP which is lower than yours from your opened codes. Could you please give me an explaination which is easy to understand. Thanks for your help again.