Knowledge Graph Construction From MusicXML: An Empirical Investigation With SPARQL Anything

Ratta, Marco and Daga, Enrico (2022). Knowledge Graph Construction From MusicXML: An Empirical Investigation With SPARQL Anything. In: Proceedings of the Musical Heritage Knowledge Graphs Workshop (Presutti, Valentina; Buffa, Michel; Steels, Luc; Trubert, Jean-François; Daga, Enrico and Meroño Peñuela, Albert eds.), CEUR-WS.

Abstract

Multimodal knowledge graphs are gaining momentum because of their ability to integrate multiple types of representations.
In particular, Musical Heritage Knowledge Graphs combine rich contextual information -- metadata from encyclopedic KGs, with symbolic content -- the scores encoded in a music ontology.
In this paper, we explore the application of SPARQL Anything -- a tool for façade-based knowledge graph construction (KGC) -- for integrating musical content encoded in MusicXML.
Specifically, we investigate the hypothesis that SPARQL is flexible enough to handle relevant tasks for musical knowledge graph construction such as (a) extracting melodic information, (b) extracting N-grams of musical information, (c) supporting the analysis of those N-grams and (d) populate a musical note ontology.
We contribute a collection of reusable queries for extracting musical features from MusicXML files to construct Musical Knowledge Graphs.
Crucially, we discuss friction points in using the façade-based approach (either in querying the façade or transforming the data) and provide recommendations on how to improve the usability of SPARQL for musical KGC tasks.

Viewing alternatives

Download history

Item Actions

Export

About

Recommendations