Skip to main content
Kent Academic Repository

Descriptive rule discovery in folk music

Neubarth, Kerstin (2014) Descriptive rule discovery in folk music. Master of Science by Research (MScRes) thesis, University of Kent,. (doi:10.22024/UniKent/01.02.50441) (Access to this publication is currently restricted. You may be able to access a copy if URLs are provided) (KAR id:50441)

PDF
Language: English

Restricted to Repository staff only
[thumbnail of 191MScR-CS Neubarth.pdf]
Official URL:
https://doi.org/10.22024/UniKent/01.02.50441

Abstract

A well-known problem in descriptive rule discovery is the large number of rules that may be discovered, making it difficult to gain an overview of the domain, identify interesting rules and understand relations between rules. This thesis extends association rule discovery to extracting rules of different association categories, in the context of computational folk music analysis. Folk music research has an established tradition of analysing relations between folk music genres, geographical regions and musical characteristics, corresponding to associations of different direction, different strength and both positive and negative associations, and relating associations at different hierarchical levels of genres or regions.

Nine association categories are identified in a qualitative analysis of reference articles on European folk music and are formalised as systematic combinations of rule templates and rule evaluation measures imposing constraints on associations. During rule generation, candidate rules are tested against the association constraints and rules satisfying constraints labelled with the respective category. In multi-level association rule mining, association categories are exploited to determine relations between hierarchically related rules: more specific rules are grouped under more general rules, based on hierarchical relations between genres and regions, and contributing, specialised and deviating sub-rules are distinguished, based on association categories.

The method is applied to the Cancionero Vasco, an iconic collection of Basque folk music. The corpus contains 1902 folk tunes encoded in symbolic format and annotated with genre and region information. The annotation vocabulary is organised in two ontologies, which define hierarchical relations among genres and among regions. Music content is represented by musicologically motivated global features. The corpus is mined for content–genre, content–region and genre–region associations. Discovered rules and rule groups, labelled with association categories, can be translated into statements similar to those in folk music surveys, and individual rules confirm and extend previously stated observations about Basque folk music.

Item Type: Thesis (Master of Science by Research (MScRes))
Thesis advisor: Johnson, Colin
DOI/Identification number: 10.22024/UniKent/01.02.50441
Additional information: The author of this thesis has requested that it be held under closed access. We are sorry but we will not be able to give you access or pass on any requests for access. 10/02/2022
Uncontrolled keywords: data mining association rule discovery computational folk music analysis
Subjects: Q Science > QA Mathematics (inc Computing science)
Q Science > QA Mathematics (inc Computing science) > QA 76 Software, computer programming,
Divisions: Divisions > Division of Computing, Engineering and Mathematical Sciences > School of Mathematics, Statistics and Actuarial Science
Depositing User: Users 1 not found.
Date Deposited: 09 Sep 2015 15:00 UTC
Last Modified: 10 Feb 2022 14:17 UTC
Resource URI: https://kar.kent.ac.uk/id/eprint/50441 (The current URI for this page, for reference purposes)

University of Kent Author Information

Neubarth, Kerstin.

Creator's ORCID:
CReDIT Contributor Roles:
  • Depositors only (login required):

Total unique views for this document in KAR since July 2020. For more details click on the image.