Web Services for Music Information Retrieval

In the emerging world of networked and distributed digital libraries, the Web services framework will be a key to facilitating simple inter-application communication between them. Yet, despite the popularity of Web services in the business sector and their seemingly obvious applicability to the digital library domain, and to MIR in particular, the adoption of these new protocols has not been widespread. To demonstrate the tremendous potential of Web services for MIR, this paper presents an application using the Google and Amazon.com databases to generate clusters of related musical artists based on cultural metadata. The use of cultural metadata to determine artist relatedness is valuable and interesting because it captures emergent popular opinion about music. Starting from an initial seed artist, Amazon Listmania! lists are traversed to find potentially related artists. Google is used to determine which of these candidates are in fact related by assessing the co-occurrence of the two artists’ names on Internet web pages. A list of artists related to the seed is returned once a given number of artists is found. The positive results generated by the system illustrate the use of Web services for exploiting the vast amount of untapped data that are available today and highlight their importance for the future, when even more musical data will become available.