The Music Ontology

Yves Raimond, Samer Abdallah, Mark Sandler, Frederick Giasson

Centre for Digital Music, Queen Mary, University of London

Overview

  • Introduction
  • Towards a web of data
  • The Music Ontology
  • A music-related web of data
  • Conclusion and Future Work

Introduction

Music-related datasets on the web



Linked data

  • Resources on the Web can be far more than web pages!
  • Linked data

    • Resources have associated representations, accessed through a dereferencing process, which can be:
      • Human-friendly (HTML, Flash, plain text, audio, video, etc.)
      • Machine-friendly (RDF, Microformats, RSS, etc.)
    • Representations may hold links, allowing an agent (a program or a person) to discover more things

    Linked data

    Linked data

    Vocabularies / Ontologies

    Such data is also linked to a particular model of its domain: an ontology

    • <http://dbtune.org/jamendo/artist/5> rdf:type foaf:Person specifies that this resource is a person, as defined in the FOAF ontology
    • A performance involves some performer, a place, a time
    • Ontologies are linked together: mo:Performance rdfs:subClassOf event:Event...
    • ... And therefore part of the data web

    The Data Web

    Turning the Web into a huge democratic, decentralized, database that can be directly consumed by applications

    The Music Ontology

    A framework for dealing with music-related information on the Semantic Web

    • Complex editorial information (worflow-based)
    • Temporal annotations
    • Cultural information (folksonomies, social networks, etc.)
    • Modular, adaptable, and designed to fit the needs of the community
    • Mesh nicely with Creative Commons RDF license information

    The Timeline Ontology

    Expressing temporal information, eg.

    • This performance happened the 9th of March, 1984
    • This beat is occurring around sample 32480
    • The second verse is just before the second chorus

    This ontology defines:

    • Interval (origin: OWL-Time)
    • Instant (origin: OWL-Time)
    • TimeLine — A backbone for addressing temporal information
    • TimeLineMap — Relationship between two timelines

    The Timeline Ontology

    The Event Ontology


    Event — An arbitrary classification of a space/time region
    • This performance involved Glenn Gould playing the piano
    • This signal was recorded using a XXX microphone located at that particular place
    • This beat is occurring around sample 32480

    The Event Ontology

    FRBR and FOAF

    Functional Requirements for Bibliographic Records:

    • Work — eg. Franz Schubert's Trout Quintet
    • Manifestation — eg. the "Nevermind" album
    • Item — eg. my "Nevermind" copy

    Friend of a Friend:

    • Person
    • Group
    • Organization
    • Social networking information (ask Oscar :-) )

    Music Production concepts

    • On top of FRBR:
    • MusicalWork, MusicalManifestation (Record, Track, Playlist, etc.), MusicalItem (Stream, AudioFile, Vynil, etc.)
    • On top of FOAF:
    • MusicArtist, MusicGroup, Arranger, Engineer, Performer, Composer, etc. — all these are defined classes: every person involved in a performance is a a performer...
    • On top of the Event Ontology:
    • Composition, Arrangement, Performance, Recording
    • Others :
    • Signal, Score, Genre, Instrument, ReleaseStatus, Lyrics, Libretto, etc.

    The Music production workflow

    Current and planned extensions

    Available extensions:


    Possible extensions:

    • Recording devices under Recording
    • Mixing events dealing with Signal objects
    • Symbolic notation under Score and AbstractTimeLine
    • Taxonomy of music processing predicates

    Linking Open Data

    • A W3C Semantic Web Education and Outreach community project.
    • Lots of open data available: Wikipedia, Geonames, Eurostat, Musicbrainz, Magnatune, etc.
    • Let's interlink them using Semantic Web technologies — data mashups
    • A subset of this project is the dbtune project, aiming at interlinking lots of music-related datasets using Musicbrainz as a data hub

    Current Map

    Conclusion and Future work

    Conclusion

    • RDF and HTTP provides a way to create a Web of data
    • Vocabulary (Music Ontology) and data hub (Linking Open Data) for further interlinking of music-related datasets
    • The MOPY Python library (Chris Sutton) allows you to manipulate Music Ontology documents without having to write one line of RDF!
    • The GNAT software allows you to find dereferencable identifiers for items in your audio collection

    Conclusion and Future work

    Further Work

    • Well, more interlinking!
    • Semantic-Web-enabled music collection handler:
      Give me all musical works composed in a city with more than 500 000 inhabitants
      Is there someone nearby really liking this band and the same beer as me, so that we can have a drink tomorrow?
      Place my collection on a timeline and make me listen something composed in the UK in 1560, followed by a rock song recorded in the 60s
      Are there any other performances of this work? Give me one with a small part at 120 bpm
    • Publishing features using Semantic Web technologies! (+interpretation rules)

    Epilogue

    And well, what would be a Semantic Web talk without a google map?

    Questions ?