monkey: ansh: I think I’ve finished with the LB landing page mockups. I added a few details to make it more engaging, and that we can expand on/add to later if we want: https://www.figma.com/file/YRbCOtFHBez8XmMdCKbG...
relaxoMob has quit
relaxoMob joined the channel
Lotheric has quit
relaxoMob has quit
lucifer
aerozol: can you ask the person to link to their page where it is not working?
relaxoMob joined the channel
relaxoMob has quit
lusciouslover has quit
aerozol
lucifer: done. I suspect it's something like they want to link it to a different *release*, but because it shares the recording it won't change. But let's see...
Seems like manually relinking that should work? And because the recordings are so similar our auto matching has picked the wrong one. But see what you think
lucifer: Might be interesting for you? This runs pretty seamlessly with a bunch of platforms (browser based and no install required, supports LB, pretty visualisations :P ): https://ampcast.app/
just about every user is more than 50% -- this doesn't seem right either.
this kinda smacks of issues with the scaling.
it seems that the range of values is between 50% and 70%. which is very odd to say the least.
if I take out the + .5 the results are nicer, but the process makes no sense.
lets just take the result as it is returned by spark and not mess with it. perhaps we missed that these steps were taken internally.
lucifer: please generate the data one more time on test-similarity-fix . lets see if that makes more sense.
Kavach
Hello, everyone I wanted to introduce myself to the communtity, I am currently pursuing my PhD in Recommendation Systems, with integration of biophysical sensing, cognative thinking and emotions. I new to open source and would love to get involved Listen Brains music recommendation as it aligns with my PhD topic
lucifer
mayhem: on it
DarkMode joined the channel
mayhem
👍
Kavach: hi!
are you an LB user and have you listened to our recommendations?
(just curious)
we're not experts in recommendation -- we certainly have some issues to hammer out still.
but we're certainly open to someone helping to improve things.
Kavach
No, I haven't used it yet, looking forward to try it.
Can you guide me how to get involved and start solving things
mayhem
sure.
what do you consider recommendations in your world. we have collaborative filtering recommendations and programmatic playlist generation (radio)
take a look at all three and tell me what interests you.
Kavach
In my world, recommendation systems can be built from various techniques, Colloborative filtering is one the technique.
Are you using Deep learning models for the Collaborative filtering
mayhem
we're not -- we're not sure how/what of our data we have we could plug into a deep learning model.
I sure would love to explore this front, but we don't have enough resources to get to it. so, you'd be most welcome if you're into DL.
Kavach
I am into my first year of the PhD, so I am also new in this.
But Using Deep learning in Recommendation System is gathering a lot of attention because of its results
lucifer
at the moment, we are using Matrix Factorization using Alternating Least Squares for Collaborative filtering.
mayhem
if we can identify the needed training data to build something, I'd be very interested.
Kavach
Okaaay, I understand.
Matrix Factrorization in the industry has been there since alot of years for the recommendations systems.
Neural Colloborative Filtering along with Matrix Factorization also yields amazing results
mayhem
well, lucifer and I are the key people who can help you with this.
lucifer
sounds like it would need the same dataset we use at the moment but a better algorithm to process it
Kavach
It's sounds amazing,
Excited to get involved and found like minded people.
Is this the place everyone uses to commiunicate or there are some other place as well
monkey
lucifer, mayhem, ansh : Just wanted a quick :+1: or :-1: from the whole LB team: please have a look at the latest designs for the LB homepage before we move forward with implementation.
aerozol: cover art download for 2023 started, should be able to calculate a new gradient next week.
monkey: are the slogals/taglines part of the deal or is that for a later discussion?
monkey
AFAIK the whole page as-is is what is proposed to be implemented
Kavach
Yes, lucifer
that sounds good.
mayhem
ok, then I'll take up the text portions -- the graphics all look good to me.
monkey: "Connect with ListenBrainz" I don't quite get that. "Connect with your Music"?
monkey
Comments on the ticket pleas e:)
mayhem
oki.
monkey
mayhem: you're looking at the worng mockups I think
(or I'm blind). Look at the very last comment
mayhem grumbles about jira not being a good place to review art
No, Figma is, but y'all aren't using it :p
mayhem
if someone specifically asks me to look at a figma, I will. if someone tosses a "here look at this figma" without engaging me, I'm bound to miss it. this has very little to do with figme. both aerozol and you have this idea that I loathe figma and refuse to use it. can we please drop that?
monkey
Sure
mayhem
thx
lucifer
Kavach: yes this is the place for for all the communication.