Re:Cleaning algorithm finds 20% of errors in major image rec...
by polm23 on Thursday April 16, 2020 @01:24PM
(#22891557)
Multiple reasons, but to name a few:
- Don't want to deal with vandalism
- Hosting static data is dramatically easier than making a public editing interface
- You want reference versions of the dataset for papers to refer to so that results are comparable. Sometimes this is used as a justification for not fixing completely broken data, like with Fasttext.
- Building on the previous point, large datasets like this don't play nice with Git. There are lots of "git for data" things but none of them are very mature, and most people don't spend time trying to figure something out.
- Don't want to deal with vandalism
- Hosting static data is dramatically easier than making a public editing interface
- You want reference versions of the dataset for papers to refer to so that results are comparable. Sometimes this is used as a justification for not fixing completely broken data, like with Fasttext.
https://github.com/facebookresearch/fastText/issues/710
- Building on the previous point, large datasets like this don't play nice with Git. There are lots of "git for data" things but none of them are very mature, and most people don't spend time trying to figure something out.