Of possible interest: Informetrics and Music

J. Stephen Downie jdownie at UIUC.EDU
Thu Jan 27 05:00:35 EST 2000

Hi Connie et al.:

I couldn't help but pick up that Connie Wilson is working on a review article
for ARIST on informetrics.  As part of my recently completed PhD research
(defended 31 August 1999) I did a wide variety of informetric analyses on the
*musical* content of a folksong database (9354 songs). A fundemental
thesis of the dissertation was the idea that by n-grammming an
interval-only representation of the melodies, one could treat the
resultant n-grams as a kind of artificial text. As text (or what I
called "musical words") the music information retrieval problem could
be converted to a text retrieval problem. To this end, it followed that
I should examine the "musical words" (i.e., n-grams) in the same manner
as I would analyze "real text." Thus, I performed a whole whack (that is
the technical term :) ) of traditional text-based informetric
analyses upon my "musical word" n-grams. These analyses include:

Term discrimination analyses/Space density (a la Salton, Crouch, Dubin)
Model fitting:
        Mandelbrot Zipf
        Generalized Waring
        Zero-truncated, Generalized Inverse Gaussian (a la Dietmar Wolfram's
        PhD work)
Entropy analyses (a la Shannon, Losee, Yavuz)
        Term frequencies
        Document frequencies
A stub of the thesis can be found at:

This stub includes the front matter (most importantly the abstract, and TOC),
and the bibliograpy. I have also published two smallish papers at the Canadian
Association for Information Science conferences (1998; 1997) that contain
preliminary informetric results. The official citations can be found in the
bibliography of the thesis stub.

I would be glad to field any questions, if anyone is interested.

J. *Stephen* Downie

   "Research funding makes the world a better place"
J. Stephen Downie, PhD
Assistant Professor,
Graduate School of Library and Information Science,
University of Illinois at Urbana-Champaign
(217) 351-5037

More information about the SIGMETRICS mailing list