container-id	patterns-knowledge-graph
name	Patterns Knowledge Graph
description	Knowledge graph containing data about patterns extracted using the [FONN tools](https://github.com/polifonia-project/folk_ngram_analysis), and software for creating that knowledge graph
type	KnowledgeGraph
release-date	07/06/2023
release-number	v0.1.0
work-package	WP3
project	polifonia-project
bibliography	main-publication Forthcoming paper in SWODCH 2023 publication Polifonia Deliverable 3.4
related-components	folk_ngram_analysis
links	https://github.com/polifonia-project/patterns-knowledge-graph https://zenodo.org/record/ (TODO)
credits	This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement N. 101004746.
funder	name url grant-agreement Horizon 2020 Framework Programme https://cordis.europa.eu/programme/id/H2020-EC https://cordis.europa.eu/project/id/101004746
has-part	P2KG-Pipeline patterns-knowledge-graph-datasets

Abstract and highlights

The Patterns Knowledge Graph contains data about patterns in folk / traditional music. The patterns are those extracted using the FONN tools.

• Allows exploration of patterns as linked open data
• Multiple corpora, multiple definitions of patterns (e.g. $n$=(4, 5, 6), accent-level versus note-level)
• Running public SPARQL endpoint and MELODY data stories as demonstrations.

Data Sources

The PatternKG currently contains pattern data extracted from two important repositories:

• MTC-ANN, Meertens Tune Collections - Annotated. This set of 360 tunes from the Dutch folk tradition contains tune family annotations. See https://www.liederenbank.nl/mtc/
• The Session (annotated subset). This set of 315 tunes in 10 tune families is from the Irish folk tradition. The Session itself is 40k tunes, crowd-sourced. This subset has been chosen and annotated with tune family information based on the musicological literature. See https://thesession.org/ for the original (large) collection.

The tune family annotation of The Session (315 tunes) is published here for the first time. It was carried out by Danny Diamond based on the musicological literature with extensive manual checking.

Requirements and installation

1. Use pip install -r requirements.txt to install the necessary libraries. They are:

PyYAML
rdflib
jams

2. You will also need to install SPARQL-Anything. We are currently using version 0.6.0 (see #1). Download the 0.6.0 jar file from: https://github.com/SPARQL-Anything/sparql.anything/releases, and copy it to the sparql-anything directory here in the patterns-knowledge-graph repo.

Running the software

Then, you can run the software by running:

cd P2KG-Pipeline
python pattern2kg_pipeline.py

It will read the data (metadata in csv format, and pattern occurrences in pickle format) from inputs/. It will write out the knowledge graph to the RDF directory in .ttl files.

Testing locally

You can run Blazegraph locally to test your Knowledge Graph. Download it from: https://blazegraph.com/ and put it (e.g.) into a /test directory. From there run, e.g.:

shell $ java -server -Xmx4g -jar blazegraph.jar

And it will start a UI in the browser. From there you can upload the .ttl files and issue test queries.

Deployment

To instead deploy, we copy our .ttl to the Polifonia server https://polifonia.disi.unibo.it/fonn/sparql, which provides the data via a public SPARQL endpoint.

We also provide some data stories via MELODY:

• https://projects.dharc.unibo.it/melody/fonn/statistics_on_the_session_annotated_subset_and_meertens_tune_collections_mtcann_pattern_kg
• https://projects.dharc.unibo.it/melody/fonn/tune_families_in_the_session_and_mtcann

Anyone can create new MELODY data stories, using the same SPARQL endpoint. See https://projects.dharc.unibo.it/melody/ for more information.

Bringing your own data

If you have pattern information and metadata representing some new corpus, you can copy it to a subdirectory in inputs/ before starting, and it will be processed. The metadata should be in csv format, with at least an identifiers field (string), a title field (string; can be empty), and a tune family field (string).

Citing this repository

If you would like to use this software and knowledge graph, please cite this repository as follows:

@software{shahid_patternkg_2023,
	address = {Galway, Ireland},
	title = {Pattern Knowledge Graph},
	shorttitle = {{PatternKG}},
	url = {https://github.com/polifonia-project/patterns-knowledge-graph},
	publisher = {University of Galway},
	author = {Shahid, Abdul and Diamond, Danny and McDermott, James},
	year = {2023},
}

This Pattern KG and associated software in this repository form part of Polifonia Deliverable D3.5, part of WP3.