Evaluating text-based similarity measures for musical content

Different music information retrieval (MIR) techniques have been developed over time. This, in part is due to the fact that the variety of formats used to represent musical content creates the necessity of using specialized format-dependent algorithms. Nonetheless, techniques that are used in text information retrieval can be used as a starting point for the development of general systems that deal with musical content. In this paper three different information retrieval techniques are evaluated for the task of music information retrieval. Two of them are commonly used for text documents, while the other was designed specifically for musical content. As a first step, a general overview of a MIR system is given, and the concept of a music representation language is introduced. After this, the different techniques evaluated are discussed and explained. Finally, the test methodology and results are presented along with a discussion of the importance that text IR techniques might have on the field.