Hacker News Reader

Comment by zozbot234 | Hacker News Reader

What's the licensing status for this data? Not sure if you're aware of this, but Wikipedia is working on pretty-printing encyclopedic English text (or natural language text more generally) starting from a language-independent, semantic ("frame" based) compositional representation, and Universal Dependencies is expected to be an important part of the solution. See e.g. https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2... . It would be especially compelling from this POV to have UD parses of short snippets of public domain text, e.g. famous historical quotations, known idioms, proverbs etc. These parses could then be added directly to Wikidata's lexicographical ("lexeme") data set.

View on HN

CC BY SA 3.0: https://github.com/amir-zeldes/gum/blob/master/LICENSE.txt

I didn't know about that project, that's really cool! I'd be curious to know whether the person who devised this scheme was aware of structured meaning representations (UCCA, AMR, ...), and if so, why they chose to create a new meaning representation. Maybe the goals of the project and/or the constraints of Wikidata necessitated this.

Anyway, GUM (and its sister corpus EWT) does have a lot of parsed permissively-licensed text, so whoever's in charge should definitely consider using them. (Amir, the maintainer, is also super friendly and would respond to an email.)

> UCCA, AMR, ...

Are descriptions for these available anywhere? (The articles on https://en.wikipedia.org/wiki/Universal_Conceptual_Cognitive... and https://en.wikipedia.org/wiki/Abstract_Meaning_Representatio... are not very helpful.) I did find an interesting description of UMR https://github.com/umr4nlp/umr-guidelines/blob/master/guidel... , which looks like it might be intended as a generalization of AMR and quite in line w/ what the project is looking for.

AMR: https://github.com/nschneid/amr-tutorial

UCCA: https://aclanthology.org/P13-1023.pdf

And yeah, you could consider UMR an improved version of AMR. AMR is often lampooned with the false name "American Meaning Representation", referring to its unapologetic focus on English only. UMR tries to undo this sin and has a lot of the same people behind it, as well as some new features such as document-level context.

Would you be willing to add those references to https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Related_a... (creating a new section for UMR) and perhaps https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Related_a... if you're aware of any references that talk about "creating complex, non-trivial text starting from an abstract meaning representation"? That would be the best way of making this discussion useful for what Wikipedia is apparently trying to do.

It's been a while since I worked with any of them so it'd take some review form me to even write a blurb confidently. Sorry! On your latter point though, NLG from AMR is a topic that's been studied since its inception. See the AMR bibliography[1] and CTRL+F "generat".

[1]: https://nert-nlp.github.io/AMR-Bibliography/

Reply on news.ycombinator.com