Clouds of Music

The Vision – Cultural Meta Data Extracted From the WEB This application is dedicated to the enrichment of large music archives with cultural metadata extracted from the internet and serves as a neat showcase for using web information retrieval techniques to refine legacy metadata.

The Business View

Creating and maintaining high quality metadata is often a very expensive task to be performed by content administrators or 3rd party providers like Gracenote or AMG. The information retrieval techniques incorporated in the demonstrator ‘Clouds Of Music’ can be used as tools to optimize this task by supporting the content administrator. Furthermore cultural aspects can be incorporated in that generation process.

The data extracted from the web can be used to improve or refine the existing metadata.

The Technical View

Similarity values are calculated based on the number of web pages that mention two given artists (co-occurrence analysis). Combining co-occurrences with legacy information about an artist's genre affiliation, an artist's importance for her genres can be derived. Using both genre affiliation and artist similarity, genre affilation can be computed automatically.

The techniques used in ‘Clouds of Music’ are domain independent!

Website

Online-Demonstrator

Partners

	3united mobile solutions AG (a VeriSign Company)
	Johannes Kepler Universität Linz - Department for Computational Perception

Sponsor

Impulsprogramm Creativwirtschaft

Publications

Making Large Music Collections Accessible Using Enhanced Metadata and Lightweight Visualizations
Florian Kleedorfer, Ulf Harr, Brigitte Krenn
In Proceedings of the 3rd International Conference on Automated Production of Cross Media Content for Multi-Channel Distribution (AXMEDIS '07), Barcelona, Spain, pp.138-144, 2007
Pdf Bibtex