Towards the Automatic Generation of a Semantic Web Ontology for Musical Instruments

  • Sefki Kolozali
  • Mathieu Barthet
  • György Fazekas
  • Mark Sandler
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6725)


In this study we present a novel hybrid system by developing a formal method of automatic ontology generation for web-based audio signal processing applications. An ontology is seen as a knowledge management structure that represents domain knowledge in a machine interpretable format. It describes concepts and relationships within a particular domain, in our case, the domain of musical instruments. The different tasks of ontology engineering including manual annotation, hierarchical structuring and organisation of data can be laborious and challenging. For these reasons, we investigate how the process of creating ontologies can be made less dependent on human supervision by exploring concept analysis techniques in a Semantic Web environment. Only a few methods have been proposed for automatic ontology generation. These are mostly based on statistical methods (e.g., frequency of semantic tags) that generate the taxonomy structure of ontologies as in the studies from Bodner and Songs [1]. The algorithms that have been used for automatic ontology generation are Hierarchical Agglomerative Clustering (HAC), Bi-Section K-Means [2], and Formal Concept Analysis (FCM). Formal Concept Analysis is a well established technique for identifying groups of elements with common sets of properties. Formal Concept Analysis has been used in many software engineering topics such as the identication of ob jects in legacy code, or the identication and restructuring of schema in ob ject-oriented databases [5]. These works are important since ontologies provide the basis for information and database systems [6].

In this study, we present a novel hybrid ontology generation system for musical instruments. The music ontology is a Semantic Web ontology that describes music-related information (e.g., release, artist, performance), but does not provide models of musical instruments. Hence, there is a need to develop a separate instrument ontology to deepen how music knowledge is represented on the Se- mantic Web. Such complementary knowledge on musical instruments can be useful to develop music recognition and recommendation systems based on semantic reasoning. This work is a preliminary step which focuses on automatic instrument taxonomy generation in Ontology Web Language (OWL). The taxonomy of musical instruments given by Hornbostel and Sachs [3] was considered as the basis for our instrument terms and initial hierarchical structure. The hybrid system consists of three main units: i) musical instrument analysis, ii) Formal Concept Analysis, iii) lattice pruning and hierarchical form generation.

In the musical instrument analysis unit, the system analyses the relationships between 12 predened classes of musical instruments (chordophones, aerophones, bowed, struck, reed pipe instruments, edge instruments, brass instruments, double reeds, single reeds, with valves, without valves, true utes) and 10 musical instrument individuals (bassoon, cello, clarinet, flute, oboe, piano, saxophone, trombone, tuba, violin) using a Multi-Layer Perceptron (MLP) classier. The underlying audio features used in the recognition and classication system are the Line Spectral Frequencies (LSF) which are derived from linear predictive analysis of the signal. The LSF characterise well the timbral differences between instruments, since they are related to the formant structure of the sounds spectral envelope which is an important aspect of the timbre. In this study, we used a database of isolated tones of various pitches and dynamics to train and test the system.

In the formal concept analysis unit, the binary relations obtained from the MLP outputs are used to identify the common attributes/individuals shared by different objects, and generate a concept lattice form. In the lattice pruning unit, the concepts are ordered in the hierarchical concept order and a well structured hierarchical form using reduced label ling technique is created. Finally, the reduced concept lattice labels and the class hierarchy of the instrument ontology is coded to the Ontology Web Language by using the OWL API java library [4]. The results have given a conclusive evidence in favor of the hierarchical similarity to the taxonomy of musical instruments used as a reference [3]. To the authors knowledge, this is the rst study to investigate automatic ontology generation system in the context of audio and music analysis. In further stages, the outcomes of this work may be used in an application to establish an automated mechanism to identify and classify music sources and annotate them with RDF meta-data. We will also incorporate a wider set of musical instruments and use more OWL language features (e.g., ob ject/data properties) which is important for Resource Description Framework (RDF) triples (subject S, predicate P, object O) and semantic reasoning. An instrument ontology is currently in development to be used as a ground truth for the further experiments.


Resource Description Framework Musical Instrument Formal Concept Analysis Semantic Reasoning Line Spectral Frequency 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    Bodner, R., Song, F.: Knowledge-based approaches to query expansion in information retrieval. In: McCalla, G.I. (ed.) Canadian AI 1996. LNCS, vol. 1081. Springer, Heidelberg (1996)Google Scholar
  2. 2.
    Cimiano, P., Hotho, A., Staab, S.: Comparing conceptual, divise and agglomerative clustering for learning taxonomies from text. In: Proceedings of the European Conference on Artificial Intelligence (2004)Google Scholar
  3. 3.
  4. 4.
    Horridge, M., Bechhofer, S.: University of Manchester and Clark & Parsia LLC and University of Ulm (2010),
  5. 5.
    Snelting, G.: Concept lattices in software analysis. In: International Conference on Formal Concept Analysis, pp. 272–287 (2003)Google Scholar
  6. 6.
    Yahia, A., Lakhal, L., Bordat, J.P., Cicchetti, R.: An algorithmic method for building inheritance graphs in object database design. In: Thalheim, B. (ed.) ER 1996. LNCS, vol. 1157, pp. 422–437. Springer, Heidelberg (1996)CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Sefki Kolozali
    • 1
  • Mathieu Barthet
    • 1
  • György Fazekas
    • 1
  • Mark Sandler
    • 1
  1. 1.Centre for Digital MusicQueen Mary University of LondonLondonUK

Personalised recommendations