Galaxy Zoo Talk

'Favourites' dataset?

  • dindon by dindon

    Is there a dataset available of favouriting/collecting patterns across objects? This could simply be a mapping from object identifier to the number of users who favourited it (ideally also number of users who classified it).

    Unfortunately I didn't see anything like this at data.galaxyzoo.org. I think this would be a really interesting axis to explore though.

    Posted

  • Capella05 by Capella05 moderator

    No, as far as I am aware there is not. Accessing a volunteers favourites / individual collection is a somewhat tricky subject. The line between what a volunteer considers personnel as opposed to what they consider open source differs a lot from individual to individual. You give permission for your actual classifications to be used in any capacity, but your contributions here on Talk are a different matter.

    I will leave it up to one of the project scientists to give you a more comprehensive reply.

    Posted

  • KWillett by KWillett scientist, admin, translator

    Yeah - that's a good question. We do keep track of which subjects that individuals have favorited, so that you can see and access your own. As Capella05 said, though, releasing that information publicly likely isn't in the cards - even as an anonymized set, we want to respect our volunteers' privacy as much as possible.

    We have shared the top ten images for a few projects as selected by the number of favorites before, though:

    Posted

  • zutopian by zutopian in response to KWillett's comment.

    We do keep track of which subjects that individuals have favorited, so that you can see and access your own.

    For new users' information: He means, that one can see his/her Favourites on the classification page.: Click "Profile"

    As Capella05 said, though, releasing that information publicly likely isn't in the cards - even as an anonymized set, we want to respect our volunteers' privacy as much as possible.

    Actually, Capella's and your statements are false, as far as I know! Those images, which a GZ volunteer marked as favourite, are automatically displayed in a collection "Favorites" in GZ Talk! E.g. AGZ000bhf8 : There are currently displayed five "Favorites" collections (with usernames), which contain this image! This means, that 5 users marked this image as favorite while/after classifying! Even if a user doesn't use GZ Talk, his/her favorite images are displayed in a GZ Talk collection, as far as I know!
    As far as I remember, the former Zooniverse team member Arfon had realized this! I am not sure, if he had informed about this in a blog post or GZ Talk. I am going to check!
    EDIT: UPDATE: I found the related blog post by Arfon.: Here it is.:

    New Talk Feature: Automatic Favourites Collection July 25, 2013
    http://blog.zooniverse.org/2013/07/25/new-talk-feature-automatic-favourites-collection/

    I guess, that you and Capella either had missed reading the above Zooniverse blog post two years ago or maybe you didn't remember!

    PS: Here is a related discussion concerning "private vs public" collections.: http://talk.galaxyzoo.org/#/boards/BGZ0000006/discussions/DGZ00011ii

    Posted

  • Capella05 by Capella05 moderator in response to zutopian's comment.

    Sorry Zutopian, I think you might of misread the original question 😃

    I am referring to the datasets that GZ provides (Free and on-line!) of the data collected from the various surveys that all of us volunteers have classified.

    Dataset:

    A data set (or dataset) is a collection of data. Most commonly a data set corresponds to the contents of a single database table, or a single statistical data matrix, where every column of the table represents a particular variable, and each row corresponds to a given member of the data set in question.

    The Zooniverse does not provide a dataset of users collections / favourites. There might of been several social posts that utilised the data (see 2nd post above), but it is not actively collated.

    You are perhaps referring to Talk collections? Totally different, and even then, they are not available as a dataset. They might be displayed on the Talk interface, but I will point out that they are neither accurate or a good representation of a collection.

    ~ Capella05

    Posted

  • zutopian by zutopian in response to Capella05's comment.

    Sorry Zutopian, I think you might of misread the original question 😃
    I am referring to the datasets that GZ provides (Free and on-line!) of the data collected from the various surveys that all of us volunteers have classified.
    (...)
    You are perhaps referring to Talk collections? Totally different, and even then, they are not available as a dataset. They might be displayed on the Talk interface, but I will point out that they are neither accurate or a good representation of a collection.

    In my opinion, it is strange, that you and KWillett have concerns concerning sharing "favourites data" in the "classification data sets", though in GZ Talk the favourites of users are automatically shared! Favourites per user is actually more private than data in a dataset.: The favourites are displayed together with a username in Talk, but in a data set the data is anonymous. As I mentioned before, even the favourites of users, who actually don't use Talk, are displayed!
    In conclusion, I think, that the favourites per user should no longer be displayed automatically in Talk!
    BTW, there had been a discussion about "private vs public Talk collections" before, as you know, but not about the automatic display of Favorites in Talk!

    Posted

  • Capella05 by Capella05 moderator

    In my opinion, it is strange, that you and KWillett have concerns concerning sharing "favourites data" in the "classification data sets", though in GZ Talk the favourites of users are automatically shared!

    Will not be an issue going forward. Automatic collections are not supported on Talk V3 and you can set your personnel collections to 'private'. Galaxy Zoo will be moving over to the 'new' Talk at some point, so there is no point making the changes here.

    Posted

  • dindon by dindon

    Thanks all for the responses. As Zutopian points out, it seems like favourites are already a matter of public record (I had considered just writing a crawler to gather the data, but it looks like the pages are very JS-heavy).

    I appreciate your commitment to preserving your users' privacy, but I have a hard time understanding how a user would see it as a violation of their privacy for data about their favourites to be released (in aggregate, no less). What nefarious purposes could this information be put to?

    Posted

  • Capella05 by Capella05 moderator in response to dindon's comment.

    To answer that question, you would have to know the reason / motivation for a particular user to add a particular image to their 'favourites' collection.

    That we do not know.

    Some volunteers use it as repository for subjects that they want to follow up on for their own research, and I respect their privacy. Most of the time, it is new volunteers that are still finding their feet, and clumping together objects that they have questions about.

    There could also be a thousand different reason between those two!

    No way to tell the difference 😃

    VOLCROWE have done quite a lot of research in this area. I would suggest contacting them.

    For what reason are you interested in this data?

    Posted

  • zutopian by zutopian in response to Capella05's comment.

    VOLCROWE have done quite a lot of research in this area. I would suggest contacting them.

    I would like to mention the below blog post about VOLCROWE, especially for information of those volunteers, who don't know, what it is!

    Introducing VOLCROWE – Volunteer and Crowdsourcing Economics, February 26, 2014
    http://blog.zooniverse.org/2014/02/26/introducing-volcrowe-volunteer-and-crowdsourcing-economics/

    Posted

  • dindon by dindon in response to Capella05's comment.

    For what reason are you interested in this data?

    I just think it'd be fun to try to infer what properties of an object make it "interesting" to someone. Especially if you could build a model based on the raw image alone - you could use it to predict interesting galaxies from an unannotated sample!

    Posted

  • zutopian by zutopian in response to KWillett's comment.

    (...)
    We have shared the top ten images for a few projects as selected by the number of favorites before, though:
    â—¦ http://daily.zooniverse.org/2014/12/22/zooniverse-advent-day-22-milky-way-project-top-10/
    â—¦ http://daily.zooniverse.org/2014/12/20/zooniverse-advent-day-20-galaxy-zoo-top-10/
    â—¦ http://daily.zooniverse.org/2014/12/04/zooniverse-advent-day-4-your-favourite-snapshots/

    I found following earlier (2009) blog posts, where most favourited GZ images (temporary favourites) had been presented.:

    Your top ten galaxies… 24 February 2009
    http://blog.galaxyzoo.org/2009/02/24/your-top-ten-galaxies/

    Top Ten Galaxies 03 March 2009
    http://blog.galaxyzoo.org/2009/03/03/top-ten-galaxies/

    Your new favourite galaxy 25 March 2009
    http://blog.galaxyzoo.org/2009/03/25/the-new-top-ten-galaxies/

    Posted