search

iai-group / www2018-table

23 stars 9 forks source link

How to create "features.csv" #3

Open g1644222 opened 2 years ago

g1644222 commented 2 years ago

Your research is very interesting.

Especially the feature file of table was an interesting attempt. I wanted to make "features.csv" myself.

If you have a program that created "features.csv", I would like you to publish it. It may be difficult because the program is not open to the public, but please treat me well.

imsure318 commented 2 years ago

the methods have a lot of data dependencies that's why we do not release the whole pipeline in the original repository. But I can share more with you: https://drive.google.com/file/d/1aeSXxnXMcz4Yc2E2f__dmKSzmYexSTyc/view In the file of metadata.py, you can find what are all the necessary files (paths to WDC corpus, the entity linking data from the WWW paper, the files for word2vec and rdf2vec). You can directly check STR.py and pool_ranker.py to see when the dependencies are used.

g1644222 commented 2 years ago

Thank you for the quick response.

Thank you for making it public. I'll try to make "features.csv" with the program as a reference.

g1644222 commented 2 years ago

I was creating features.csv based on your file, but I could not find a certain class.

I am calling the "WDCTables" class from data_loader.py in the file you shared with me. However, I cannot find "WDCTables" in data_loader.py . Is it intentionally left out, or is it an older version?

imsure318 commented 2 years ago

Hi, this script is used to reproduce the same method in another dataset - WDCTables. You need to change the code a bit.