Two ways of feature extraction: 1. split multi-value cells; 2. split into multiple columns

[antonin_d]

Will OpenRefine consider supporting multiple tables manipulation in the future?

There is interest in thinking about what that would look like at least. @steve has been working on this. He is working as a researcher on this topic:

Table Scraps: An Actionable Framework for Multi-Table Data Wrangling From An...

Personally I would find it really great to make progress in this area, but it feels like there is still much design work to do before we can plan on implementing anything. Your thoughts on the subject are very welcome.

For more thoughts about the limitations of the records mode, see also this thread:

Representing hierarchical data: beyond the records mode?

I am opening this thread to give a summary of an important open problem in OpenRefine’s design. The goal is to encourage people to come up with alternative design proposals or pointers to solutions from other tools. GitHub issues relevant to this are normally classified under the “records” tag. Here are a few places where such a discussion has happened before: #2825 The future of the records mode #2298, which proposes to investigate the Dremel encoding: “a novel columnar storage format for …