Presentation is loading. Please wait.

Presentation is loading. Please wait.

MusicOWL The Music Score Ontology

Similar presentations


Presentation on theme: "MusicOWL The Music Score Ontology"— Presentation transcript:

1 MusicOWL The Music Score Ontology
Jim Jones Institut für Geoinformatik / Universitätsbibliothek (ULB), Universität Münster, Germany Diego de Siqueira Braga Institut für Wirtschaftsinformatik, Universität Münster, Germany Kleber Tertuliano Musikhochschule, Universität Münster, Germany Tomi Kauppinen Aalto University School of Science, Finland WI ‘17, August 2017, Leipzig

2 Agenda Problem Statement (music score discovery)
Proposed Approach (semantic web technologies) MusicOWL Musci2RDF API Music Score Portal Future Work Jones et al. | MusicOWL - The Music Score Ontology

3 Unknown composers? Managing and analyzing music scores content is a known nightmare for digital libraries (information centers). Many of these libraries have a huge old music score archive, but they treat these music scores with a rather historical than musical documents. Known composers (Telemann - Georg Philipp Telemann ( ): Brockes-Passion, Detail der Viola da Gamba-Stimme., Elgar) Easy to find using most of text based searches Rather historical than musical importance Unknow composer (Charles Doisy) - How to find such music scores inside large historical archives if the piece is unknown? - Jones et al. | MusicOWL - The Music Score Ontology

4 Text-based Search Engines
Cello Concerto (E minor) Op. 85 1919 Sir. Edward Elgar Adagio, Moderato Lento, Allegro Molto Adagio Allegro, Moderato, Allegro ma non-troppo … E Minor (tonality) in Portuguese: Mi menor Adagio: Beats per minute Allegro: BPM Jones et al. | MusicOWL - The Music Score Ontology

5 „Raw Data Now!“ phrases rests clefs time signatures notes dynamics
Conventional meta data on the top - but what about „the rest“? dynamics connections Jones et al. | MusicOWL - The Music Score Ontology

6 Going beyond the conventional metadata …
How many voices / instruments play this piece? How many movements does it have? In which tonality(ies) was it written? How many times a particular melody is played along the piece? What is its level of complexity? 1 - What can be achieved by going beyond the metadata? Let‘s take a look what one could ask if this information is taken into consideration. 2 - How can we structure a dataset in such a way, that these questions are addressed? Semantic Web Technologies! Jones et al. | MusicOWL - The Music Score Ontology

7 Semantic Web Technologies – Linked Open Data
Genesis 3:14 Xxxxxx xxxxxx xxxxxxxx xxx xxx xxx xxx. Xxxx xxxx x xx xxx. - God “linked open data describes a method of publishing structured data so that it can be interlinked and become more useful.” Jones et al. | MusicOWL - The Music Score Ontology

8 Semantic Web Technologies – Linked Data (Benefits)
Attaches meaning to data (allowing thematic searches) My repository! Private data and public data can be mixed, enabling companies and individuals to make better decisions and generate innovations. Links different kinds of data sets making the WORLD a single big repository. How to formalize these concepts? Jones et al. | MusicOWL - The Music Score Ontology

9 Semantic Web Technologies – Ontologies
„A set of concepts and categories in a subject area or domain that shows their properties and the relations between them.” - Oxford Dictionary Human ≡ Man ⊔ Woman Man ⊓ Woman ⊑ ⊥ Parent ≡ ∀hasChild.⊤ Mother ≡ Woman ⊓ Parent TBox John Mary Orphan ≡ Human ⊓ ¬∃hasParent.⊤ ChildlessHuman ≡ Human ⊓ ¬Parent MarriedWoman ⊑ Woman ⊓ hasSpouse.⊤ Poygamist ⊑ Human ⊓ 2≥hasSpouse.⊤ HappyMan ⊑ hasSpouse.(Human ⊓ Orphan) hasSpouse hasDaughter hasFather hasMother hasDaughter Ontology in Computer Science: A formal datamodel to represent knowledge, based on concepts and their relationships. Tbox axioms : Teminological Knowledge hasSpouse -> symetric property hasChild -> asymentric property hasMother ⊔ hasFather ⊑ hasParent hasSon ⊔ hasDaughter ⊑ hasChild hasWife ⊔ hasHusband ⊑ hasSpouse RBox Peter Jones et al. | MusicOWL - The Music Score Ontology

10 OWL - Web Ontology Language
Axioms TBox: subclass relationships C ⊑ D RBox: subproperty relationships R ⊑ S, inverse properties R-, transitivity ⊑+ ABox: facts for classes C(a), properties R(a,b), equality a=b, difference a≠b Class Constructors Conjunction C ⊓ D, disjunction C ⊔ D, negation ¬C of classes Property restrictions: universal ∀R.C and existential ∃R.C Number restrictions: ≥n R and ≤n R Closed classes (nominals): {a} Source: OpenHPI (Hasso Plattner Institut) Jones et al. | MusicOWL - The Music Score Ontology

11 MusicOWL Modelling Music Scores
Issue regarding Music scores + Semantic Web techonologies Applying OWL to musical data Jones et al. | MusicOWL - The Music Score Ontology

12 MusicOWL Structure mo:Score mo:movement mo:movement mso:hasScorePart
mso:ScorePart Jones et al. | MusicOWL - The Music Score Ontology

13 MusicOWL Structure mo:Score mo:movement mo:movement mso:hasScorePart
mso:ScorePart mso:hasStaff mso:Staff Jones et al. | MusicOWL - The Music Score Ontology

14 MusicOWL Structure mo:Score mo:movement mo:movement mso:hasScorePart
mso:ScorePart mso:hasStaff mso:hasVoice mso:Staff mso:Voice Jones et al. | MusicOWL - The Music Score Ontology

15 MusicOWL Structure Chord ⊑ NoteSet ⊓ ≥3hasNote.Note
Tetrad ⊑ NoteSet ⊓ (=4)hasNote.Note mo:Score chord:Note mo:movement mo:movement chord:Note chord:Note mso:hasScorePart mso:ScorePart mso:NoteSet mso:hasDuration mso:hasStaff mso:hasVoice mso:hasNoteSet mso:Staff mso:Voice mso:Duration Jones et al. | MusicOWL - The Music Score Ontology

16 MusicOWL Structure mso:Measure mo:Score mso:hasNoteSet mo:movement
chord:Note mo:movement chord:Note chord:Note mso:hasScorePart mso:ScorePart mso:NoteSet mso:hasDuration mso:hasStaff mso:hasVoice mso:hasNoteSet mso:Staff mso:Voice mso:Duration Jones et al. | MusicOWL - The Music Score Ontology

17 MusicOWL Structure mso:Time Signature mso:hasTimeSignature mso:Measure
mo:Score mo:movement chord:Note mso:hasNoteSet mo:movement chord:Note chord:Note mso:hasScorePart mso:ScorePart mso:NoteSet mso:hasDuration mso:hasStaff mso:hasVoice mso:hasNoteSet mso:Staff mso:Voice mso:Duration Jones et al. | MusicOWL - The Music Score Ontology

18 MusicOWL Structure eflatMinor ⊑ Key ⊓ mode.Minor ⊓ tonic.Eb mso:Time
Signature mso:hasTimeSignature mso:Measure mso:hasKey ton:mode mo:Score ton:Mode ton:Key mo:movement chord:Note chord:Note mso:hasNoteSet mo:movement ton:tonic chord:Note chord:Note mso:hasScorePart mso:ScorePart mso:NoteSet mso:hasDuration mso:hasStaff mso:hasVoice mso:hasNoteSet mso:Staff mso:Voice mso:Duration Jones et al. | MusicOWL - The Music Score Ontology

19 MusicOWL Structure mso:Time Signature mso:hasTimeSignature mso:Measure
mso:hasKey ton:mode mo:Score ton:Mode ton:Key mo:movement chord:Note chord:Note mso:hasNoteSet mo:movement ton:tonic chord:Note chord:Note mso:hasScorePart mso:Clef mso:ScorePart mso:NoteSet mso:hasClef mso:hasStaff mso:hasVoice mso:hasNoteSet mso:hasDuration mso:Staff mso:Voice mso:Duration Jones et al. | MusicOWL - The Music Score Ontology

20 MusicOWL Structure mso:Time Signature mso:hasTimeSignature mso:Measure
mso:hasKey ton:mode mo:Score ton:Mode ton:Key mo:movement chord:Note chord:Note mso:hasNoteSet mo:movement ton:tonic chord:Note chord:Note mso:Dynamic mso:hasScorePart mso:Clef mso:hasDynamic mso:ScorePart mso:NoteSet mso:hasClef mso:hasStaff mso:hasVoice mso:hasNoteSet mso:hasDuration mso:Staff mso:Voice mso:Duration Jones et al. | MusicOWL - The Music Score Ontology

21 MusicOWL Structure mso:Time Signature mso:hasTimeSignature mso:Measure
mso:hasKey ton:mode mo:Score ton:Mode ton:Key mo:movement chord:Note chord:Note mso:hasNoteSet mo:movement ton:tonic chord:Note chord:Note mso:Dynamic mso:hasScorePart mso:Clef mso:hasDynamic mso:ScorePart mso:NoteSet mso:hasClef mso:hasArticulation mso:hasStaff mso:hasVoice mso:hasNoteSet mso:hasDuration mso:Articulation mso:Staff mso:Voice mso:Duration Jones et al. | MusicOWL - The Music Score Ontology

22 Proof of Concept (Historical Music Scores)
Data structure very complex… how to create data? Jones et al. | MusicOWL - The Music Score Ontology

23 Dataset Generation PDF, JPEG MusicXML OMR (Sibelius) Music2RDF API
* ULB Münster: 1k music scores from the last 5 centuries - Many scores written by unknown local composers - Scores had a rather historical than musical importance. * Sibelius (by AVID) – choice based on existing license * MusicXML - Lacks semantics, e.g. tonalities / circle of fifths Not easily interoperable (query via XPATH possible) MusicXML OMR (Sibelius) Music2RDF API Jones et al. | MusicOWL - The Music Score Ontology

24 Dataset Generation - Challenges
OMR often confuses lyrics with notes or articulations * PDF perfect * Old German writting (Fraktur?) OMR inaccurate with old music scores (expert check needed) Jones et al. | MusicOWL - The Music Score Ontology

25 WWU Music Score Portal Live demo:
GraphDB Triple Store (Free version) Search criteria: C C C (whole) Expected result: PAGE 29 – Notendruck, Movement 23, measure 1, staff 2, voice 4 Jones et al. | MusicOWL - The Music Score Ontology

26 SPARQL Example 1 – Notes Search
mso: < chord: < note: < SELECT ?score ?measure WHERE { ?score mso:hasScorePart ?part . ?part mso:hasMeasure ?measure . ?measure mso:hasNoteSet ?noteset1 . ?noteset1 mso:hasNote ?note1 . ?note1 chord:natural note:A . ?noteset1 mso:nextNoteSet ?noteset2 . ?noteset2 mso:hasNote ?note2. ?note2 chord:natural note:D .} Jones et al. | MusicOWL - The Music Score Ontology

27 SPARQL Example 2 – Durantion Search
mso: < chord: < note: < SELECT ?score ?measure WHERE { ?score mso:hasScorePart ?part . ?part mso:hasMeasure ?measure . ?measure mso:hasNoteSet ?noteset1 . ?noteset1 mso:hasDuration ?duration1. ?duration1 a mso:Half . ?noteset1 mso:nextNoteSet ?noteset2 . ?noteset2 mso:hasDuration ?duration2 . ?duration2 a mso:Quarter .} Jones et al. | MusicOWL - The Music Score Ontology

28 SPARQL Example 3 – Phrase Search
mso: < chord: < note: < SELECT ?score ?measure WHERE { ?score mso:hasScorePart ?part . ?part mso:hasMeasure ?measure . ?measure mso:hasNoteSet ?noteset1 . ?noteset1 mso:hasNote ?note1 . ?note1 chord:natural note:B . ?noteset1 mso:hasDuration ?duration1 . ?duration1 a mso:Half . ?noteset1 mso:nextNoteSet ?noteset2 . ?noteset2 mso:hasNote ?note2. ?note2 chord:natural note:E . ?noteset2 mso:hasDuration ?duration2 . ?duration2 a mso:Half.} Jones et al. | MusicOWL - The Music Score Ontology

29 SPARQL Example 4 – Chords with 4 Notes
mso: < chord: < note: < SELECT DISTINCT ?score ?measure (COUNT(?note) AS ?size) WHERE { ?score mso:hasScorePart ?part . ?part mso:hasMeasure ?measure . ?voice a mso:Voice . ?voice mso:hasNoteSet ?noteset . ?measure mso:hasNoteSet ?noteset . ?noteset mso:hasNote ?note . ?note chord:natural ?natural .} GROUP BY ?score ?measure ?voice ?noteset HAVING (?size = 4) Jones et al. | MusicOWL - The Music Score Ontology

30 Wrapping up Usage of Semantic Web technologies on Music Score discovery. Semi-automatic approach for Music Score data set creation. WWU Music Score Portal Jones et al. | MusicOWL - The Music Score Ontology

31 Future Work Improve data set creation (support other input formats)
Evaluate the possibility of implementing RDF export in existing Music Notation Software. Your suggestions? Jones et al. | MusicOWL - The Music Score Ontology

32 Vielen Dank! Portal: http://linkeddata.uni-muenster.de/musicportal/
MusicOWL: Music2RDF: Dataset: Images used in this presentation: Jones et al. | MusicOWL - The Music Score Ontology


Download ppt "MusicOWL The Music Score Ontology"

Similar presentations


Ads by Google