Personal Music Knowledge Base

Semantic Federation of Music and Music-Related Information for Establishing a Personal Music Knowledge Base

Outline

#Motivation #Concept #Implementation #Evaluation #Conclusion

Problem Description

#Music is perceived and described very subjectively #People often get lost in their steadily growing, multi-placed, digital music collections #Personal music collection use cases require a huge amount of fuzzy and exact information #Several music information sevices are available to assist users by organising their music collections #Users cannot seamlessly interact with all services directly from their music player applications #Personalisation aspects of music collection management tasks are often neglected

Solution Proposal

#Available music and music-related knowledge has to be >gathered up >semantically federated >integrated into a uniform knowledge base #A personalised semantic aggregation of music metadata #Data is personalised represented in an appropriated visualisation #A continuum of Yves Raimond's dissertation project [1]

Background

#REST architectural style #Semantic Web technologies #Linked Data publishing guideline

Requirements

#Knowledge representation: >Semantic relations >Uniform naming >Expressive power >Reasoning #Knowledge management: >Information resources have to be creatable, accessible, modifiable, removable >All information processing capabilities are secured by an access control mechanism >PMKB KMS is guided by the principles of the REST architecture style

The Personal Music Knowledge Base

The PMKB - Workflow

#User account initialisation #Individual information extraction #Information service choice #Proactive update notification #Information exploration #Personal association and context

A Personal Music Knowledge Base

#Knowledge representation: >Ontological KRM powered by Semantic Web KRLs >All new created or extensively modified ontologies are published on purl.org/smiy/ >All occurred KR issues are mainly resolved as a outcome of this thesis #Knowledge management system: >Only sketched-out due to the massive KR modelling work

Ontologies and Vocabularies Overview

A Music Playlist Example

Cognitive Characteristics Ontology

Property Reification

#Can be described by the Property Reification Vocabulary

A Cognitive Characteristics Example

@prefix foaf: <http://xmlns.com/foaf/0.1/> .
@prefix cco: <http://purl.org/ontology/cco/core#> .
@prefix wo: <http://purl.org/ontology/wo/core#> .
@prefix ex: <http://example.org/> .

ex:APerson
	a foaf:Person ;
	foaf:name "John Wayne" ;
	cco:skill <http://dbpedia.org/resource/Football> ;
	cco:expertise <http://dbpedia.org/resource/Football> ;
	cco:interest <http://dbpedia.org/resource/Football> ;
	cco:habit [
		a cco:CognitiveCharacteristic ;
		cco:topic <http://dbpedia.org/resource/Football> ;
		cco:characteristic cco:skill ;
		wo:weight [
			a wo:Weight ;
			wo:weight_value 6.0 ;
			wo:scale ex:AScale
			] ;
		cco:activity <http://sw.opencyc.org/concept/Mx4rwJRiEpwpEbGdrcN5Y29ycA>
		] ;
	
	...

Knowledge Management System

#Knowledge base category: triple store >With inbuilt reasoning capacities and SPARQL implementation #KRM is published by following the Linked Data principles #HTTP protocol for supporting some REST constraints #WebID and WAC as authentication and authorisation mechanism #RDFizer for mapping non-RDF KRs into Semantic Web KRs #N3Logic and SPIN as preferred rule languages #Provenance data at least by utilising the Info Service Ontology #Change propagation and synchronisation via Atom + AtomPub, PSHB, SUP, DSNotify Eventset Vocabulary #OWL, SKOS and SPIN for stereotype profile and categorisation modelling

AudioScrobbler RDF Service

PMKB ID3 Tag Extractor

public static void createPlayCounter(Model model, int playCount,
	String playCounterTitle, Resource mediaObject)
{
	Resource playCounter = ModelUtil.generateRandomResource(model);
	model.addStatement(playCounter, RDF.type, PBO.PlayBackCounter);
	addIntegerLiteral(model, playCounter, CO.count, playCount);
	addStringLiteral(model, playCounter, DC.title, playCounterTitle);
	model.addStatement(playCounter, PBO.media_object, mediaObject);
}

Evaluation (1)

#A social evaluation of the outlined ontologies #Reutilisation: >Ordered List Ontology in Stories Ontology and Sport Ontology (interest) #Application: >Recommendation Ontology in OpenRecommender project and dbrec [2] / seevl (interest) >Counter Ontology, Play Back Ontology and Recommendation Ontology in the Libre.fm (interest) #Reviews and Mentionings

Evaluation (2)

#Indexing: >Manually: Schemapedia, Ontology Design Patterns repository, Sindice >By maintainers: Linked Open Vocabularies information service, Linked Open Data cloud cache >SEO via embedded RDFa (with the help of SpecGen) + sitemaps (with the help of Remote Sitemap Generator)

Conclusion

#PMKB KMS: >With uniform communication endpoints and ontological KRM >Personalisation due to the whole information flow life cycle >Strong aspect on change propagation #PMKB KRM: >Utilises as much as possible and appropriated existing Semantic Web ontologies >Missing gaps are mainly closed due to own contributions >Proposed vocabularies and ontologies are successful applied and evaluated #Realising the PMKB vision is a bigger project ;)

Future Work

#Implementation of the sketched-out PMKB KMS #Further refinements of the stereotype, context and user modelling #Implementation of a reference information service quality ontology and a full information service rating system

Sources

[1]Yves Raimond. A Distributed Music Information System. PhD thesis, Queen Mary, University of London, November 2008 [2]Alexandre Passant. dbrec - Music Recommendations Using DBpedia. In Proceedings of the 9th International Semantic Web Conference (ISWC 2010), pages 209–224. Springer, 2010

Abbreviations (1)

RESTRepresentational State Transfer PMKBPersonal Music Knowledge Base KMSKnowledge Management System KRMKnowledge Representation Model KRLKnowledge Representation Language KRKnowledge Representation SPARQLSPARQL Protocol and RDF Query Language HTTPHyperText Transfer Protocol WACWeb Access Control RDFResource Description Framework N3Notation 3 SPINSPARQL Inferencing Notation

Abbreviations (2)

PSHBPubSubHubbub SUPSimple Update Protocol OWLWeb Ontology Language SKOSSimple Knowledge Organisation System SEOSearch Engine Optimisation

Musical Feature Categorisation

Music and Music-Related Information

REST Architectural Style

#Component interface constraints: > (Uniform) Identification of resources >Universal action semantics >Manipulations of data through representation >Self-descriptive messages >Hypermedia as the engine of application state

Semantic Web Technology Stack

Linked Data Publishing Guideline

#Name reference mechanism #Information resource delivering mechanism #Resource description mechanism #Resource connection mechanism #Information resource modification mechanism

Resource Description Levels

Property Reification Definition

@prefix prv: <http://purl.org/ontology/prv/core#> .
@prefix cco: <http://purl.org/ontology/cco/core#> .
@prefix ex:  <http://example.org/> .
 
ex:SkillReification a prv:PropertyReification ;
   prv:shortcut cco:skill ;
   prv:shortcut_property cco:characteristic ;
   prv:reification_class cco:CognitiveCharacteristic ;
   prv:subject_property cco:agent ;
   prv:object_property cco:topic .
   
   ...