Improve documentation on generating train/valid/test datasets

Living-with-machines / DeezyMatch

A Flexible Deep Learning Approach to Fuzzy String Matching

https://living-with-machines.github.io/DeezyMatch/

Other

139 stars 34 forks source link

Improve documentation on generating train/valid/test datasets #108

Open kasra-hosseini opened 3 years ago

kasra-hosseini commented 3 years ago

[ ] Document some of the methods that we have tried in our experiments so far to generate negative examples.
[ ] Add experiments/case-studies so far

Related docs:

[ ] https://github.com/LinkedPasts/LaNC-workshop/tree/main/deezymatch
[ ] @mcollardanuy wrote: "We already have a "Tips / Suggestions on DeezyMatch functionalities" section (https://github.com/Living-with-machines/DeezyMatch#tips--suggestions-on-deezymatch-functionalities) that we could then move to that separate document"

kasra-hosseini commented 3 years ago

@mcollardanuy I just created this branch feature/108-negative-examples from develop to address this issue.

kasra-hosseini commented 3 years ago

We do have some documentation/tutorials on this:

https://github.com/LinkedPasts/LaNC-workshop/tree/main/deezymatch

mcollardanuy commented 3 years ago

Ah great, thanks! I will also collect the scripts we have to create the different string pairs datasets, maybe we can unify them somehow. I'll work on this on Friday or next week.

kasra-hosseini commented 3 years ago

Great, thanks! Whenever you have time...I am also happy to help.

I think it would be good to have a separate directory (maybe in DeezyMatch/examples? but, of course, please feel free to create a new directory if needed) where we collect all those scripts and tutorials/docs regarding experiments/use-cases. We can then link that directory from the README.