iris2hu / l2c-rater Goto Github PK

• evn: Linux Debian9 4.9.267 / Python 3.6.3
• unzip http://model.scir.yunfutech.com/model/ltp_data_v3.4.0.zip to ltp_models
• python3.7 main.py -m b -t wp -dp data/
• Result is data/essay_linguistic_indices.csv
  
  Then ?

Hi, I am very interested in your work. And I think your work is very nice and try to reproduce the results which are in the paper. And I found the features selected after step-wise regression with default hyper-params are different from those in paper, I got 29 but the paper reports 33. And in the code, I didn't see anything about writing-error features which are also mentioned in your work.

• For Feature Selection: I think maybe the hyper-parameters are different from the ones you used in your experiment, so if it is convenient, maybe you provide the parameters used in the experiment to me.
• For Writing-Error Features: I am not sure whether I missed it in your code or I should use some methods like statistics to create them.
  ：）

Hi, I am interesting in your work and have a question about the dataset used in your paper. I went to the website (http://hsk.blcu.edu.cn/) and found that it can be downloaded 500 essays randomly. So I want to ask that how to access to the same dataset used in your paper. Thanks in advance!

the model dir is empty, what should I to download, thank you