Δημιουργία Οντολογίας για τη συσχέτιση μουσικής και συναισθήματος

Abstract

Music plays an important role in our lives and its impact on our emotions has been a focus of research for many years. Composers tend to use different musical features and combinations in order to influence how listeners feel. This study explores the connection between various musical features and the emotional responses they create. It introduces an ontology, which is a structured framework to represent knowledge using concepts, relationships and axioms. The ontology classifies the most common musical features into distinct categories and offers properties which link them with valence(positivity) and arousal(intensity) values. Then, a Python script is presented, which is used to popu late the ontology with instances of musical pieces derived from MusicXML files. The script extracts the musical features from such files and creates instances of them. It then uses these instances in order to calculate valence and arousal values for every mea sure of the musical piece analysed. By populating the ontology with many instances of musical pieces by four popular composers, a knowledge graph is constructed. Using the knowledge graph, several experiments are presented that make use of the SPARQL query language, demonstrating how the structure can be applied to analyze music and compare composers. Lastly, the study discusses how the ontology is evaluated, answers competency questions, and suggests possible future improvements and applications.