aai-institute / pyDVL

pyDVL is a library of stable implementations of algorithms for data valuation and influence function computation
https://pydvl.org
GNU Lesser General Public License v3.0
89 stars 9 forks source link

Doc/glossary #537

Closed mdbenito closed 3 months ago

mdbenito commented 3 months ago

Description

This PR adds a glossary page. I almost exclusively took terms from abbreviations.md but we should add more that are relevant to pydvl. Links to wikipedia / other sources might be in order too.

@schroedk the influence part is again a stub, please check it for consistency and completeness. @AnesBenmerzoug let me know of any terms that you miss and I'll try to add them.

At some point we will want to link to these from other docs. So maybe I'll try adding explicit anchors with { #anchor }, although that has not really worked out so well in the past.

Changes

Checklist

AnesBenmerzoug commented 3 months ago

The new Glossary page makes the Methods page a bit redundant because it contains descriptions of some of the methods as well as links to the implementations. Perhaps we can keep method descriptions in Methods and only put other terms in the Glossary.

mdbenito commented 3 months ago

The new Glossary page makes the Methods page a bit redundant because it contains descriptions of some of the methods as well as links to the implementations. Perhaps we can keep method descriptions in Methods and only put other terms in the Glossary.

Well... you suggested it 😅 But I'm not sure it's redundant. The methods page is a quick list with links for reference for newcomers ("is my algorithm here?"). Whereas the glossary is for quick definitions that can be linked from anywhere. I would imagine that in time, if we really want the glossary to be useful, we'd cross-ref to it from all over the docs, possibly adding a little tooltip with the definition (using e.g. the couple of js+css files suggested here).

AnesBenmerzoug commented 3 months ago

That sounds like a lot of work. I think we can merge it now as it is.