LILLA and SILLA algorithms

explosion / spaCy

💫 Industrial-strength Natural Language Processing (NLP) in Python

https://spacy.io

MIT License

30.02k stars 4.39k forks source link

LILLA and SILLA algorithms #270

Closed knyghty closed 8 years ago

knyghty commented 8 years ago

I'm back with another alignment-related feature request.

I'm looking to implement LILLA and SILLA as described in papers: https://pdfs.semanticscholar.org/c883/8797233934980176e4a3d9269b711fe99b36.pdf http://pss.sagepub.com/content/23/8/931.full.pdf https://pdfs.semanticscholar.org/eff2/1c747cd965aa3430598dc536196e4048e8c0.pdf

I guess at least the latter may be dependent on #59

honnibal commented 8 years ago

These look like good suggestions --- they should be easy to implement. A dependency-based version of their SILLA measure looks like it should work, so I don't think constituency parsing will be necessary.

knyghty commented 8 years ago

Another day, another new paper: http://www.thoughtfill.com/docs/doyle_yurovsky_frank_www16.pdf

Looks like it could replace those two - I don't see why we couldn't use dependency trees and run it on that for a syntactic measure also.

I have some small doubts though, like they only used it on Twitter data, makes me wonder if they tried it on longform text and maybe didn't get the results they wanted...

lock[bot] commented 6 years ago

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.