A talk in three parts - metadata from a market, data and technical perspective.
Speakers : Decibel Music Systems
As part of the British Library's Keeping Tracks symposium on music and archives in the digital age.
21.03.14
http://britishlibrary.typepad.co.uk/music/2014/04/keeping-tracks.html
1. The Secret Life of a DataPhile
Evan Stein
Mark Wainwright
Georgiana Bogdan
Decibel Music Systems
2. PART 1
A Life in Search
Evan Stein
Decibel Music Systems
3. The Talk
Evan Stein: Introduction – Reasons for Decibel as a product
Mark Wainwright: Technical issues with music metadata
Georgiana Bogdan: Metadata collection and processing
4. What is Decibel?
Fact-based metadata system
Social / buying-based recommendations (e.g., Amazon)
Sound / mood-based recommendations (e.g., EchoNest)
Fact-based navigation (e.g., MusicBrainz, Gracenote)
Data and search provided through an API
White-label services for customers’ products
Navigate collections through linked information
Furnish information, sleeve-note equivalent and file tagging
Repertoire, artist and recording normalisation
Insane level of detail
6. A bit of history
Musician
Library of Congress
Studies in Musicology
Switch to computers, thanks to Fernando Pessoa
Manhattan DA
Standard & Poor’s
Decibel
7. Library work
Classification
Retrieval (by classification)
LC / Dewey Decimal, metadata
Knowledge of the domain is a key to good work
There are human databases walking about
8. Musicology
Catalogues and classifications
Works
Instruments
Eras, genres, styles
Biography
Ways of thinking about music
Repertoire, theory, performance practice
Sociology, anthropology, psychology, linguistics
Correlation with other art forms
Performers
9. Law enforcement
Data for hypothesis-formation
Unknown start and end
Non-linear search
Multiple languages, phonetics, semantics
Linkage
11. Why the British Library?
Fact based systems are good for research
You don’t know what the user wants to know until they want to know it
Data-based thinking allows you to follow your train of thought
Good for navigating collections
Improvements in bandwidth and storage
Personal collections are getting larger
Stores and services are also collections
Library collections are being digitised, and physically smaller
12.
13. Digitisation
Information extracted from artefact (record, book, video, etc.)
Cons
Possible lack of context and background
Ignores the artefact
Pros
“Good enough” for most uses
Can be consumed anywhere
16. Graph Database Features
Polymorphism More detail without affecting performance
Recursive Relationships Results are more complete
Associative Structure More interesting questions
23. Graph Database
Polymorphism More detail without affecting performance
Recursive Relationships Results are more complete
Associative Structure More interesting questions
24. PART 3
Metadata Collection and Processing:
A Data Detective’s Investigation
Georgiana Bogdan
Decibel Music Systems
25.
26. Why do we collect metadata?
KEEP
CALM
AND
JOIN THE
DIGITAL
MUSIC
REVOLUTION!
Source: IFPI Digital Music Report 2014
27. Why do we collect metadata?
Because few things matter more.
It is crucial for:
Artists
Music Listeners
Music Providers
Copyright Holders
Music Libraries & Archives
28. Who do we collect metadata for?
Music streaming services
Copyright Collection Society
App Developers
Online radio services
Record Labels
Other music
industry
players
Digital Music Stores
Music
distributors
29. What metadata do we collect?
Comprehensive data model
Graph database for representing and storing data; API for delivering it
Rich data fields; mix of internet sources, research and editorial content
31. How do we collect metadata?
Online legal sources (and the magic of computer programming!)
Research Team + Editorial Team. The Right People!
Data Partnerships
32. How do we keep our metadata evolving?
Keeping an eye on emerging music markets – data & content in local languages
Being aware of the music ecosystem; connecting with the industry players
Directly engaging with music industry professionals. Being social and sociable!