#metabrainz

we can create a request object

and a task object

request object can have attributes related to the actual request (host, port, etc. reducing the number of arguments of get, post functions) and task object can have priority and all associated with it (again reducing arguments)

each task can have a response parser associated with it

and we can have a subclass of a task dispatched which hooks a particular type of response parser to each request object

does this sound good?

s/dispatched/dispatcher

this way we can support multiple different response parsers at the same time while maintaining a single queue and WS

zas?

yes, looks ok to me

So I Just downloaded and fired up the musicbrainz vbox vm

After running the docker comand recommended in the install docs, I was able to hit port 5000 but the server on port 8080 didn't seem to start

and I'm not sure what to do / how to get it to start

I tried the reindex command, no joy

the psql command on port 5432 doesn't seem to work either

cjnolet: hi. what's your issue with psql? where are you running the psql command?

So last night I went through nightmares trying to get the musicbrainz server installed from source on a Fedora 24 machine. I think it came down to the version of perl being different so there were differences in the packages that were being installed

so today, I broke down and downloaded the VM. I have it running in VBox on a mac pro

Upon initially starting the VM, I had to run the "reset-containers" script to get the server on port 5000 to work

I have it reindexing right now, it's taking awhile to do.

I'm trying to build a metadata repository to do some machine learning research for a Ph.D program. I just need this to work, hehe

yeah, reindexing will take a while

remember, reindex is only needed if you want to do search

for example, I run a copy of the database, but with no index

ah

Yeah one of the things I'd like to be able to do is search

ok 👍

I'm trying to mash up IDs from various different sites.

cjnolet: Neat! For what, if I may ask?

A lot of my research is geared towards crowd dynamics- trying to read and influence crowd behavior.

(Cross-referencing by external IDs is one of the things I like to promote with MB, but I don't know too many people/projects that are actively doing it. :))

cjnolet: Anything public?

I've found a surprising amount of companies that have been doing it- but they are all proprietary solutions and they are all charging a ton of money!

lol

cjnolet: With MusicBrainz data specifically?

And (primarily) based on actual IDs (rather than matching by similar (looking) strings etc.)?

Fresno, at the moment, I don't have anything publicly to offer, no. I'm hoping that will change! There are some interesting things I'd like to do with soundcloud data but I'm going to need a streaming index like shezam to do it

YES!

Oh man. I saw this the other day! Kudos!

I'm a committer on a couple apache projects. I absolutely value open source. I'd love to try and see how my research / prototyping could help.

(Turned out it was actual Spotify IDs in their URLs, and I ended up making https://community.metabrainz.org/t/spotify2musi... to hopefully get more Spotify IDs into the db.)

cjnolet: FWIW, we're currently very interested in data visualisations of our data.

And you should also check loujine's musicbrainz-dataviz project.

I've been infatuated with some of these mood-based models I've been seeing lately

It would be really neat to be able to visualize people's taste profiles based on the common musicbrainz tags. I don't know if there's enough overlap with them

but that would be realy neat. Like cluster portions of their tastes that seem to be describable with a subset of tags

Well, I would say that what I tag = what I like/my taste.

except more people listen to music than tagging

it's almost like you'd need an archive of people's listening habits

cjnolet: Are you familiar with AcousticBrainz and ListenBrainz?

alastairp: If only there was an open one of those... :)

cjnolet: did you solve your access problems to psql?

Freso: I've seen those listed in my research but Ihavn't dived into them yet. Are they different analytics bundles/companies?

alastairp: Still waiting on the reindexing, lol

29%

cjnolet: They're both MetaBrainz projects too. :)

Ah! So I had a girl Elizabeth reach out to me. I'm actually trying to get a startup off the ground as well and we're interested in commercial licensing for the metabrainz data too

cjnolet: ListenBrainz is the open "archive of people's listening habits".

Elizabeth Bigger is her name

cjnolet: She goes by Quesito in here.

ah gotcha! Dude.... this is what I love about irc, lol

I've been an efnet junkie since 1997

I'm actively ingesting the million song dataset

And AcousticBrainz is the, well, open acoustic information database. :) alastairp is the lead of AcousticBrainz, so he probably has a much better "elevator quip" to introduce it. :p

we have more songs in acousticbrainz ;)

AcousticBrainz is basically an alternation to 1M dataset.

what do initial commercial licensing costs usually look like for a startup with very limited financial backing?

^ evelator pitch: "more than a million songs"

Only with about 2.5M unique recordings.

cjnolet: we are very friendly

and understanding that startups don't have money

cjnolet: No clue, but AFAIK it's reasonable. Quesito will know.

We certainly want to pay our fair share, if our product succeeds.

Ok

Alternation? wat? alternative

Compared to MB, AB and LB are both pretty young. :)

(I mean, LB isn't even in beta yet.)

"Normally start-ups spend vast amounts of money to create these data sets before building recommendation systems. We aim to short-cut this process and put the power to create recommendation engines into the hands of everyone. We plan to release the data under the CC0 License."

You guys are great

We try. :)

Has anyone attempted anything with the ISNI dataset as of yet?

(Reading your meeting notes from over a year ago)

I've also been looking at linking ISRCs from Spotify to musicbrainz to see how many are there/missing

cjnolet: No.

(For whether anyone's done anything with the ISNI dump.)

Huh. You actually do use "hobby horse" in English too. #TIL.

weird. I just read about "hobby horse" two days ago .. :o

also elizabeth isn't "a girl". she's our little cheese " Supporter Catalyst"

:O

:D

lol

The only correspondence I've done with her so far is getting an email after I created an account. Hopefully next week I can schedule a phone call.

cjnolet: I'm glad you found our IRC channel either way. :) It bodes well for the relationship!

Is this dataset still available by chance: https://multimediaeval.github.io/2017-AcousticB...

alastairp: ^ anything you know about?

cjnolet: Note that alastairp and a lot of others MeB'ers are in Europe (a handful of us in Barcelona), and it's around midnight her (midnight CEST, so 11 PM in UK, IE, and Portugal, and 1 AM in Eastern Europe) - and it's weekend.

I'm not sure alastairp checks IRC in his off hours, so any AB inquiries may not get answered until he's back in office on Monday.

Ah, good point. I'm in Maryland, USA and it's only 6pm here, lol

Oh, I guess he was around 1½ hour ago, so maybe he does sometime check IRC in off-hours. :)

Thos genre annotations are wonderful. I'm trying to figure out how to secure a useful standard set of genres that are pretty well represented everywhere

Good luck. :p

alastairp and reosarevok, may have some thoughts on that.

reosarevok is currently travelling though.

Is anyone around? Need help with CB docker tests

Oh wow. So acousticbrainz has a lot of the audio features (and more) that the echo nest / spotify provide. This is so wonderful