A couple of new datasets from 2012 ("Voter autrement" in voting station and "Vote au Pluriel" online) #23
Conversation
|
PS: for some reason the Pol.is dataset has a publication date in the future and appears as the most recent one on Preflib. |
|
Would it not be better to merge these with the previous datasets? At least 73 and 71 seems to me merge-able. We did it like this for AAMAS bidding for instance: different files per year (see https://preflib.github.io/PrefLib-Jekyll/dataset/00037). What do you think? |
|
Yes that could make sense, I made separate files because the set of authors on each one is not exactly the same, and they each require their citation... but I could put all citations in "required citations"? Or try to find a paper on which every one is? But true, it might be better to find a way to merge them, because more are coming (see https://theo.delemazure.fr/datasets/). Also I had a question: is it possible to have fractionnal "counts" in the file like "2.31: 1, 2, {3,4}, 5" ? (Asking because in many of these datasets we have weights for votes,to reduce biases) |
|
I think you can put all the papers in the required citations (that no one really use unfortunately...). About weights, it's not yet part of the format specification but strictly speaking why not. I'm just not sure that they should be included in the data for this specific case: the weight are values that you chose yourselves to balance different factors that you found important right? If so, then it's a bit of an arbitrary choice what weight goes where so I would not put them in. Does that make sense? |
|
Ok I'll do this then! Yes these are values chosen to match the distribution of votes in the actual election (otherwise in some election Mélenchon wins with 60% of the votes). For some use case (like running time) it is not very important but if one want to interpret the result it might be important. I can also just mention that the file with weight is somewhere else for people interested. |
|
I did not do the merging of French datasets yet but added another dataset (scotus data). Will take care of the french datasets next. |
|
I wonder if the sanity check failing is because of the line breaks in the Selected Studies. Try to put they all on a single line maybe? |
|
It should be the missing spaces after the comma in the header of the file description section... (yeah sorry, weird choice I've made there) |
|
Should I then wait for you to merge the French datasets before merging? |
|
Yes, I will (hopefully) do that before end of week. |
|
Thanks for finding the bug! |
|
I think I will split the datasets like this, if that is okay with you :
|
|
That looks good to me. It's a good idea to organise per experiment. It's probably what makes the most sense 😊 |
|
Agreed!
…On Wed, Feb 4, 2026 at 1:26 PM Simon-Rey ***@***.***> wrote:
*Simon-Rey* left a comment (PrefLib/PrefLib-Data#23)
<#23 (comment)>
That looks good to me. It's a good idea to organise per experiment. It's
probably what makes the most sense 😊
—
Reply to this email directly, view it on GitHub
<#23 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAJGSMWSFCWU5JO4KF7N3CL4KJBRFAVCNFSM6AAAAACJFFSRDSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTQNBZGA4TMNZQG4>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
--
*Nicholas Mattei*
Associate Professor, Tulane University
***@***.*** | www.nickmattei.net
Stanley Thomas Hall | 305B
+1 504 247 1416
Department of Computer Science
Tulane University
6823 St Charles Ave
New Orleans, LA 70118
|
|
Alleluia :D |
|
Nice, so you're all done? :) |
|
Yep ! |
3 participants
No description provided.