#metabrainz

/

      • CatQuest
        ss?
      • alastairp
        for source separation we'd have to do it at an earlier stage - I don't think there's enough data in the acousticbrainz data to do anything with it
      • ruaok
        but I guess we would need to re-run the low level data.
      • shivam-kapila
        Screenshot
      • pristine___
        > Sync already feedbacked recs
      • ruaok
        alastairp: yerp.
      • CatQuest
        yea it was borked so I used ircloud too
      • pristine___
        Will be a separate PR shivam-kapila
      • CatQuest
        see KassOtsimine link
      • alastairp
        it's a pretty active area, especially with new deep learning techniques
      • both Deezer and Sony presented some pretty amazing systems at ismir last year
      • shivam-kapila
        pristine___: sorry but that doesnt make sense. Thats an incomplete feature then
      • pristine___
        I makes senses
      • alastairp
        I don't know if the tutorial just covered current techniques, or if they were proposing some new systems
      • pristine___
        Because I have made a separate branch
      • CatQuest
        ok I'm out now
      • pristine___
        The PR will be hugeeee if I do everything in one PR.
      • shivam-kapila
        Then branch of the child pr from this and merge it into this
      • Or make feedback pr totally separate
      • And only refactor in this
      • ruaok
        alastairp: I think we should keep an eye on this stuff and choose the most promising tools to include in the next low level scan for AB... when ever that will be.
      • alastairp
        yeah
      • pristine___
        shivam-kapila: did you look at the PR. I made a separate less file.
      • I am not able to render the emojis in line with the text
      • :(
      • shivam-kapila
        I did
      • And I am confused what happened
      • ruaok
      • shivam-kapila
        pristine___: can we do refactor first so we can merge it today
      • And then move to feedback
      • Bkz its still some work imo
      • pristine___
        What refactor
      • shivam-kapila
        We added a new component
      • pristine___
        Yeah
      • shivam-kapila
        But we didnt clean up the redundant code
      • We have in Recent listens
      • CatQuest
        eh. is the upperline/menu/thing on jira-tickets borked for anyone else?
      • shivam-kapila
        Listen card
      • And so on
      • Lets do the cleanup and move to feedback in separate pr
      • CatQuest
        also seems every page has horizontal scrolling too
      • I'm fairly sick of these "maintenance" jira updates every 3-4 weeks that changes colours, makes it harder to read and borks things up tbh
      • shivam-kapila
        Also col-sm-9 etc are bootstrap classes
      • pristine___
        The component isn't ready in my opinion. I mean I plan to put the non functional emojis properly, merge it, remove redundant code, merge it, syn feedback, merge, to have small PRs.
      • CatQuest
        by all means ypdat security, but could you not f-upp the layout and such, tnx.
      • pristine___
        Col-xs-9?
      • Or sm?
      • shivam-kapila
        Xs
      • pristine___
        Hmm.
      • shivam-kapila
        So rec-col-xs-9 is not gonna work properly
      • pristine___
        I guess that's causing the issue.
      • Right
      • shivam-kapila
        pristine___: I am unable to align myself with your planned schedule
      • Emojis are no good there
      • pristine___
        shivam-kapila: but for #rexommendations id, I should define col-xs-9 again right?
      • I mean I cannot use it from #listens id?
      • shivam-kapila
        You dont need to ideally
      • pristine___
        Oh
      • shivam-kapila
        #recommendations, #listens in less
      • Comma separated selectors share styling
      • pristine___
        What?
      • shivam-kapila
        Ah selectors in css is like how we select ids, class, element etc
      • Did you get chance to look into css mdn docs i linked
      • pristine___
        > #recommenders, #listens in less
      • No shivam-kapila I guess I missed the link :(
      • shivam-kapila
      • I have a class now. I will get back to you after that.
      • Mr_Monkey is starting to look at PR #1133
      • reosarevok
      • Not quite a dress, but :p
      • ruaok
        wow, nice.
      • I want an updated wikidata shirt. do they sell those?
      • or do I have to go beg lydia for one?
      • reosarevok
        We asked for them nicely and got them for free :p
      • Begging Lydia sounds like a plan :p
      • ruaok
        will do then. :)
      • I'd offer a summit shirt of ours, but she wouldn't wear it.
      • reosarevok
        There's also socks, although those are more standard
      • Mr_Monkey
        shivam-kapila, pristine___ : For what it's worth, I think alastairp's idea for expanding +1/++1/-1/--1 is not very clear (~= not used anywhere that i'm aware of).
      • I think sticking to more established feedback methods would be a better idea. Emoticons should work quite well, and can be supplemented with text on hover to clarify meaning.
      • Also, I do like the idea of showing all the emojis only once the user hovers over a feedback component.
      • alastairp
        ok, cool
      • if it's not used then there's no reason to use it
      • shivam-kapila
        Mr_Monkey: yeah I align with your thoughts
      • Mr_Monkey
        And finally, if the "this is not a good recommendation" feedback is separate (I agree on that), perhaps we could just have some text button, clearly separate from the emojis? I think it would suggest that we're not so much talking about emotions (the emojis are there for that).
      • reosarevok
        So is "I really dislike this" 💩 ?
      • Mr_Monkey
        🤮
      • pristine___
        Mr_Monkey: I am okay with what community thinks because I am not very clear about using 5 emojis. I personally would have only used thumbs up and thumbs down.
      • Mr_Monkey
        Ah, that's another story :) We do need to figure out how many levels of feedback we want.
      • alastairp
        one thing that discussed in person is that the meaning of feedback is quite personal
      • e.g. ruaok has said before that if he says "no" to a song, he never wants to hear it again
      • whereas sometimes people might want to say "no, not here (or not now), but in other places OK"
      • Mr_Monkey
        Right. I do think emoticons can help with that, compared to the functionally-equivalent stars rating for example.
      • shivam-kapila
        alastairp: then we can have a hide rec for that
      • Mr_Monkey
        I think for this idea of situational feedback "not here (or not now), but in other places OK" we'll just have to rely on if the user skipped the song.
      • If a user really doesn't like a song at all, 😠 face is pretty clear for that, compared to thumbs down for example
      • ruaok
        alastairp: I wonder if we should have a survey of how people may be interesting in giving feedback. and then distilling them into a coherent set of feedback values -- I personally dont see why "never hear this again" and "not, her but maybe ok in a diff context" should not both be available options.
      • assink has quit
      • assink joined the channel
      • alastairp
        ruaok: yeah, I think that might be useful. at least it'd give us a wider understanding of how people rate stuff
      • ruaok
        I do worry about one aspect of that -- most people are unfamiliar with giving such feedback, so we would need to educate before ask for feedback
      • alastairp
        I would flat out ask in a survey 1) do you use a music streaming platform, 2) do you ever use the +1/-1 buttons
      • and some questions about if they think that the system "listens" to them
      • let me start a document with some suggestions for questions
      • ruaok
        perfect
      • reosarevok
        yvanzo: see also https://github.com/metabrainz/musicbrainz-serve... - I answered your comment, but not sure if that's enough or you'd want to change something :)
      • alastairp
        > Very few studies have investigated which personal and situational characteristics exactly motivate and inhibit users to provide preference feedback to the system (see Pommeranz et al. 2012, for a notable exception). This is an important issue, as many recommender systems rely on explicit feedback (e.g. users' ratings) to give good recommendations.
      • > Privacy concerns may reduce users' tendency to disclose personal information (TeltzrowandKobsa 2004; Chellappa and Sin 2005; Berendt and Teltzrow 2005; Ackerman et al. 1999). On the other hand, if it positively influences their user experience (i.e. in terms of better recommendations), users may be more willing to provide feedback (Spiekermann et al. 2001; Brodie et al. 2004; Kobsa and Teltzrow 2005)
      • v6lur joined the channel
      • ruaok
        it will remain critical or us to maintain user trust in everything we do, to uphold the latter sentiment.
      • BrainzGit
        [musicbrainz-server] reosarevok merged pull request #1633 (master…MBS-8725): MBS-8725: Allow mediums to have an unknown tracklist https://github.com/metabrainz/musicbrainz-serve...
      • BrainzBot
        MBS-8725: Allow mediums to have an unknown tracklist https://tickets.metabrainz.org/browse/MBS-8725
      • reosarevok
        Please let's try to properly test this in beta ^ :)
      • (I *think* it works fine this time, but I would really like to avoid further messes)
      • yvanzo: also https://github.com/metabrainz/musicbrainz-serve... was fixed according to your request :) (I think!)
      • reosarevok stops sending PRs
      • alastairp
        ruaok: yeah, absolutely. I think that we're in a good position to do this because we are explicitly open
      • open data repository, open code
      • travis-ci joined the channel
      • travis-ci
        Project bookbrainz-site build #3494: passed in 4 min 56 sec: https://travis-ci.org/bookbrainz/bookbrainz-sit...
      • travis-ci has left the channel
      • pristine___
        Mr_Monkey: #1133 is ready for review.
      • I will post a ss here
      • Mr_Monkey
        Great :) I'll look at it today
      • pristine___
      • assink has quit
      • Mr_Monkey: please ignore the emoji eolors, i will do that in a separate PR, the CSS thing.
      • assink joined the channel
      • Last_updated is on 8th October
      • So everything works perfectly.
      • Thanks!
      • I guess I will add the window here as well, so users will know the exact dates, for eg your listens were last updates on 8 October. `Listens submitted between 1 October and 7 October have been used to generate recs`
      • Something like this.
      • MajorLurker has quit
      • _lucifer
        alastairp: did you see the sentry weekly report ? It seems that after friday release events dropped from 4.4k to ~200.
      • yvanzo
        I enabled SSO for Tickets using either MusicBrainz or GitHub.
      • alastairp
        _lucifer: no, I didn't. that's good news! we must have had a bunch of bugs that we weren't aware of
      • _lucifer
        yeah, right
      • yvanzo
        There is a small display bug that may hide these options on the starting page, I reported it already.
      • reosarevok
        yvanzo: oh, neat
      • shivam-kapila
        reosarevok: nice tee
      • pristine___: I looked in detail now and I see a lot of code that we wont need for rec component
      • pristine___
        Comment on the PR please, I will remove. I wasn't clear when I was writing the first draft :)
      • But things make a lil sense now.
      • shivam-kapila
        Yes I can relate
      • pristine___
        :p
      • BestSteve has quit
      • BestSteve joined the channel