General information

Important dates

Submission information

Conference program

     Detailed program

     Keynote speakers

     Special Metadata Session

Accepted papers

Accepted posters/short papers



     Professional visit

     Social events

     Conference closing remarks


Conference venue

Conference committee

Photos from the conference

Conference proceedings


Hosted by


Sponsors and partners

Mairie de Paris

IU seal, red on white, small


ISMIR 2002
3rd International Conference on
Music Information Retrieval

IRCAM – Centre Pompidou
Paris, France
October 13-17, 2002


There is a famous dialogue in Molière’s play “Le Bourgeois Gentilhomme”, between the proverbial Mr Jourdain and his philosophy master, from whom he requires help in writing a little note to a lady of great quality with whom he is in love:

MONSIEUR JOURDAIN: No, I don't want either prose or verse.

PHILOSOPHY MASTER: It must be one or the other.


PHILOSOPHY MASTER: Because, sir, there is no other way to express oneself than with prose or verse.

MONSIEUR JOURDAIN: There is nothing but prose or verse?

PHILOSOPHY MASTER: No, sir, everything that is not prose is verse, and everything that is not verse is prose.

MONSIEUR JOURDAIN: And when one speaks, what is that then?


MONSIEUR JOURDAIN: What! When I say, "Nicole, bring me my slippers, and give me my nightcap," that's prose?


MONSIEUR JOURDAIN: By my faith! For more than forty years I have been speaking prose without knowing anything about it, and I am much obliged to you for having taught me that.

We have been “speaking” metadata for more than forty years, much before the word was coined. The first metadata records must have been written on sumerian and akkadian clay tablets - there are two primitive catalog tablets dating from around 2000 BCE at the temple of Enlil at Nippur.

As the “One-Two-Three-Infinity” title of George Gamow’s famous book illustrates, we may reach intractability when dealing directly with contents of large collections: the induced complexity relates not only to accessing the items (the search and retrieval phase) but also to managing their immediate use and their life-cycle, to handling the rights of their owners and their associated costs (and hence their market value) in compliance with a multiplicity of national and international laws and regulations.

This is why the need to list, describe and index the objects themselves arises:

·         by selecting some characteristic traits of the constituents – a title, an incipit, a shape, a color…;

·         by attempting at providing its “perceived essence” as if it were, e.g., by summarizing it.

whence the emergence of catalogues, which have been the domain of expertise of librarians for millenia.

Now that collections of metadata become themselves computer-accessible and can constitute vast collections, the need to organize them in other ways than purely sequentially in card-catalog boxes arises and raises a host of intellectual, technical and financial issues: their structure (just a title and name, or a deep structure description?), their production (manual or automatic) and maintenance through time (the question of evolving standards and preservation), their identification and organization, their protection, their value... in other words, how to deal with them as data, in addition to dealing with them as “mediators” to data.

This is (in my view) more complex for music than for written text[1]:

·         it is easier to describe words with words (and so for instance it may be easier to produce text summaries that make “textual” sense than music summaries that make musical sense);

·         the authorship/ownership issue (from the composer to the publisher);

·         the multiple possible “incarnations” of music (scores and interpretations).

When you add to this the digital capacity to derive other works (by summarization, by linking) from original works, this becomes quite mind boggling.

The relatively recent work on metadata as emerging from the digital world (the internet, really) has had an important impact on the “librarian” view of metadata. But conversely, the knowledge accumulated by librarians through the ages has also fed the research and development of metadata for digital objects.

Most of the music which is “traded” on the internet can be viewed as “consumer music” which probably has no need for deep metadata structure, authority records [listes d’autorités], and so on. But immediate money and universal usability should hopefully not be the only reasons for intellectual (and artistic) pursuit – there is other food than McDonalds and other music than Brittney Spears.

During this afternoon, we shall have a glimpse of the variety of perspectives on metadata, as they relate to music. Before the break, we will hear Leonardo Chiariglione and Chris Barlas speak on MPEG as an intermediary between the author and the consumer; then Harriette Hemmasi will show the evolution of the “library” view of music metadata as reflected in a real, musical digital library, system, followed by David Datta who will describe another real musical metadata system, in the entertainment world. This session will end with a talk by Eric Scheirer on the issue of the business value of music metadata.

Michael Fingerhut
October 16, 2002

The ISMIR 2002 Web pages will be regularly updated
to include program content and schedule


[1] As Chris Barlas rightly pointed to me, performed text has many of the complexities listed below.