HoloClean: Weakly Supervised Data Repairing

Very related: Open Refine

> is a powerful tool for working with messy data: cleaning it; transforming it from one format into another; and extending it with web services and external data.

Thanks for sharing! This looks much more practical than HoloClean.

There is also ExceLint [1] for fixing bad data in (Excel) spreadsheets. I never tried it.

[1] https://github.com/ExceLint/ExceLint

The code is on GitHub[1] for anyone interested.

[1] https://github.com/HoloClean/HoloClean

This looks useful.

I wonder if the concept could be used to make predictions about data instead of just preparing it. ie, to interpolate likely missing values like in collaborative filtering?

There are probably several datasets that could be explained with this kind of scalable weak semi supervision

This looks really cool, but I must admit - when I saw the name I hoped it was a HoloLens AR gamification of tidying up my house that gives me points for each dish washed or item of clothing folded. But hey, data is cool too.

Holoclean was presented at this year's SIGMOD conference.