Graph Neural Network for social recommendaiton about qrec HOT 8 CLOSED

Okay, I have found his code before, but that code is poorly written and runs so slow. So I decided to reimplement my own version of GraphRec which performs bad. I also asked the authors for the whole datasets using in that paper but received no reply. According to what you say, I guess it probably doesn't work as reported in his paper. Maybe I should stop wasting time on GraphRec. Thanks for your help.

from qrec.

Well...

I have also tried to run his released code of Graph Neural Network for Social Recommendation. I guess it needs several days to train GraphRec with the whole dataset used in his/her paper...

And I have requested for the code of Deep Adversarial Social Recommendation more than twice but received no reply. BTW, is this the work you said that didn't work for you?

from qrec.

Thanks for you guys.

from qrec.

Well... actually I have implemented the released code of GraphRec and I must say that the results are almost as same as their reported ones.

from qrec.

I am questioning the IJCAI'19 paper. The results of the www'19 paper might be reproducible. Even so, I found the author closed the issue tab of the repo, which is interesting. As for other papers, you would never receive any replies from him if you try to ask for the implementations via email. Personally, I hold a very negative view of his work.

from qrec.

I see~ also one thing is really interested me is that the author published 1, DSCF(RecSys19), 2, GraphRec(WWW19),3, DeepSoR(WWW2018) at ALMOST the same period, its quite fast, fast as abnormal.. the second thing is, these three paper use almost same baseline models, the only change is they are trying to compare the model developed by themselves(i mean the same author), I think its easy and tricky. BTW, only the code of GraphRec is released, I guess the author is confident about graphRec(although other people wont think so), but less confident about neither DSCF and DeepSoR...

from qrec.

The fast pace is not the problem. Refusing to release the codes is. Anyway, I would not be serious to his work.

from qrec.

The fast pace is not the problem. Refusing to release the codes is. Anyway, I would not be serious to his work.

you are right

from qrec.