lancopku / aapr

Automatic Academic Paper Rating: Data and Model (ACL 2018)

AAPR

• This code is based on: Automatic Academic Paper Rating Based on Modularized Hierarchical Convolutional Neural Network
• The dataset for the AAPR task is available at https://drive.google.com/open?id=1ttKdXUjaFi3eS6zeHITuZjsTGrh6M4ij
• For the AAPR dataset, you can process the latex source data in your own way (e.g. considering the number of formulas or charts). The script clean_latex.py provides an optional way of initial rough processing.
• If there is any problem, please contact Pengcheng Yang ([email protected])

Citation

If you use the code or the dataset for your research, please cite the paper:

@inproceedings{YangACL2018,
  author    = {Pengcheng Yang and
               Xu Sun and
               Wei Li and
               Shuming Ma},
  title     = {Automatic Academic Paper Rating Based on Modularized Hierarchical
               Convolutional Neural Network},
  booktitle = {Proceedings of the 56th Annual Meeting of the Association for Computational
               Linguistics, {ACL} 2018, Melbourne, Australia, July 15-20, 2018, Volume
               2: Short Papers},
  pages     = {496--502},
  year      = {2018}
}

aapr's Issues

Error while running Dict.py

While running Dict.py i am getting the error:
label = self.idxToLabel[i]

KeyError: 0
How can i solve it

The dataset

Hi. There are 37464 papers downloaded in your dataset link, but the dataset used for experiments in your article is 19218 papers. How did you select these 19218 papers?

License

Hi, thanks for the great work! I am wondering what license the dataset is released under. Thank you again!

Code for attentive pooling for CNN

Dear Authors,

Thanks for your work. It was really interesting to learn something new :)

I have a doubt regarding attentive pooling for CNN. There are two classes class Attention and class Conv_attention. I couldn't find any uses of class Conv_attention however.

I guess the implementation for attentive pooling is inside def forward(self, x) defined in class Attention

code of def forward():
u = self.linear_project(x.contiguous().view(-1, input_size)).contiguous().view(batch_size, seq_len, -1)
atten_weights = self.softmax(batch_matmul(u, self.representation))
s = attention_mul(x, atten_weights)
return s

In equation tanh is being applied after linear transformation followed by attention calculation which involves multiplication with another vector (u_w).
while in method batch_matmul(), it seems that tanh is being called after multiplication with u_w.

Is it right? I am not much familiar with torch.
If so, how both would yield the same results?

Can you please provide equivalent tensorflow code for the same?

Dont understand which program should run first please upload files that are needed for this project please replay

i coundnt run the program since there are lot of files is missing please upload all the files that are needed

where to start running the code

I have quick question: which file should I first run?

I guess I need to first train the model, so I ran train.py, but it gave me "FileNotFoundError: [Errno 2] No such file or directory: 'config_hcnn_attent_7.json'".

Could you please let me know where I can get the file config_hcnn_attent_7.json, or should I start from the other file?

I also tried make_data.py, it gave "FileNotFoundError: [Errno 2] No such file or directory: './data/data/text_dict'", the make_vocab.py gave "FileNotFoundError: [Errno 2] No such file or directory: './data/data/abstract_train'"

Could you please upload all the missing files, or let me know where to download these files?

Code for attention in CNN module

Sanity Check

I believe that a lot of people are curious about it and I am just the first one who is not too shy to ask. The question is:
Could you show the acceptance probability of this paper with your current methods?

The dataset link is broken. Could you please provide other possible links?

Clean_latex.py not working

Dear authors,

I think that 'make_single_tex' function in clean_latex.py is missing.

Did you intend it?