by Holger H Hoos
Musical databases are growing in number, size, and complexity, and they are becoming increasingly relevant for a broad range of academic as well as commercial applications. The features and performance of musical database systems critically depend on two factors: The nature and representation of the information stored in the database, and the search and retrieval mechanisms available to the user. In this paper, we present an experimental database and retrieval system for score-level musical information based on GUIDO Music Notation as the underlying music representation. We motivate and describe the database design as well as the flexible and efficient query and retrieval mechanism, a query-by-example technique based on probabilistic matching over a clustered dataset. This approach has numerous advantages, and based on experience with a first, experimental implementation, we believe it provides a solid foundation for powerful, efficient, and usable database and retrieval systems for structured musical information.