To SIGIR'99 Home To SIGIR'99 Workshop Home

"The Exploratory Workshop on Music Information Retrieval"

An ACM SIGIR 1999 Workshop
19 August 1999
University of California, Berkeley
Call for Participation

Page Updated: 15 May 1999
Created by: J. Stephen Downie

QuickLinks:

Introduction
Objectives
Background
Approach
Call for Participation
Workshop Organizers
Contact Information

Introduction:

This workshop is intended to bring together Music Information Retrieval (MIR) researchers; Information Retrieval (IR) researchers; computer scientists; musicologists; music bibiliographers; music, digital, and general librarians; and, music providers and music industry members; in the first-ever forum exclusively devoted to issues pertaining to MIR. We are conducting this workshop in order achieve the following objectives.

Objectives:

1. Foster the development of methods and technologies for content-based Music Information Retrieval (MIR) systems by:

a. having those on the user side (i.e, musicologists,librarians,industry members, etc.) inform those on the research side (i.e., IR researchers, computer scientists, etc.) about real-world problems, needs, and opportunities

b. having those on the research side inform those on the user side about the strengths and weaknesses of the various approaches to content-based MIR (both extant and under research & development)
c. having those with a specific research interest in MIR demonstrate to other IR researchers the interesting, unique, and non-trivial research issues involved in the development of MIR systems
2. Foster a framework for future fruitful research into MIR by:
a. having participants explore consensus opinion on the establishment of research priorities, inter-disciplinary collaborations, evaluation standards, test collections, resource sharing, funding opportunities, communication channels, etc.

b. having participants begin work on the establishment of a formal organization of those interested in MIR. Such an organization will have as its mandate those items mentioned above.
3. Contextualize and integrate MIR within the theme of SIGIR '99 by striving to have the participants leave the tutorial with answers (?) to the questions posed in the SIGIR '99 Call for Papers' section on Multimedia Information Retrieval:
http://www.sims.berkeley.edu/conferences/sigir99/mmir.html

Background:

Automating access to music information through the use of digital computers has intrigued musicologists, computer scientists, librarians and music lovers alike. Each has his own purpose in mind and thus there seems to be as many approaches to developing Music Information Retrieval(MIR) systems as there are users and researchers. A half-hour's perusal of the back issues of "Computing in Musicology" (Hewlett & Selfridge-Fields, eds.) will bring this fact to the fore. Some have designed complex suites of computer tools (e.g., David Huron's "Humdrum" [http://dactyl.som.ohio-state.edu/Humdrum/] to analyze all the varied facets of music. Others have tried to automate the thematic catalogue by including incipit or thematic extracts part of a bibliographic record (e.g., RISM [http://www.rism.harvard.edu/rism/] ; also, Huron and Kornstaedt's "Themefinder" [http://musedata.stanford.edu/databases/themefinder/]). Still others have explored the idea of using sophisticated approximate string matching techniques (e.g., McNab et al.'s "Meldex" [http://www.nzdl.org/cgi-bin/gw?c=meldex&a=page&p=coltitle]; Prechelt & Typke's "Tuneserver" [http://wwwipd.ira.uka.de/tuneserver/]). One thing that unites all of these approaches it that they have some kind of shortcoming. The more powerful analytic systems can be very difficult to use, incipit and thematic indexes leave out large amounts of music that might be of interest, and approximate string matches can be computationally expensive without necessarily giving better results.

The recently published "Melodic Comparison: Concepts, Procedures, and Applications" [http://musedata.stanford.edu/publications/cm/idx11.html] is indicative of the growing interest in MIR issues. However, in our recently drafted literature review for our PhD dissertation on MIR we note that despite the growing interest in MIR issues the formal literature remains sparse, particularly where the evaluation of MIR systems is concerned. (Downie [http://alexia.lis.uiuc.edu/~jdownie/mir_papers/chaps1to4e.pdf]). It appears that one reason for the sparseness of the MIR literature is that the various MIR research teams are operating autonomously. We hope that this exploratory workshop will contribute to the formation of a more cohesive MIR research and evaluation programme by affording all interested parties the chance to share their insights and achievements. Perhaps, one day, MIR researchers and stakeholders will look back upon this workshop as the birthplace of a new ACM SIG, one replete with peer-reviewed publications, annual meetings, and TREC-like trials?

Approach:

1. Integrate participants: As stated in the objectives, we intend to bring together rather disparate groups that collectively have an interest in MIR. Our first goal will be to establish and reinforce the commonalities of the participants. We plan to review the issues pertaining to music for the benefit of our IR/CS participants (i.e., the facets of music information, the 1000-year history of MIR, etc.). For our music specialists we will review the fundamental's of the various IR techniques and evaluation metrics. Throughout the course of the workshop we will pay particular attention to the effective communication between the groups. In short,we are striving for a rigourous and accurate discussion of the issues that is at the same time free from jargon and specialist shorthand.

2. Lecture/Demonstration/Discussion Format: After the background material has been covered the primary mode of communication for the workshop will be a series of Lecture/Demostration/Discussion presentations. In the Lecture presenters will explicate the motivations and strengths of a particular approach. Demonstrations (where applicable) of the approach will be made to make the Lecture more concrete. Discussion will be encouraged after each Lecture/Demonstration.

3. Panelist Response: In all likelihood, panelists and presenters will be one and the same (though not necessarily). However, the presenters will be asked to reconstituted themselves as members of the panel. The panelists will provide commentary in response to the various presentations. It is hoped that this phase of the workshop will lead to the identification of some common intellectual threads or points of debate.

4. Constitutional Forum: The last period of the workshop (say between the last coffee-break and the close of the workshop) will be the open-floor discussion (open to participants and presenters alike) concerning how to move forward. We will propose, based upon the discussion, the formation of some kind of more formal organization for those interested.

Call for Participation:

We invite the participation of all interested parties. You may participate at one or more level(s):

General Participant

Presenter

Panelist

Presentations and/or demonstrations are solicited that cover such topics as:

Interface methods (e.g., singing, notation, keyboard,etc.)
Representational issues (e.g., Music Representation Languages, sound files, graphic scores, etc.)
User-needs assessment (e.g., consumer use, researcher use, etc.)
User/system interaction (e.g., query failures, feedback models, etc.)
Informetics (e.g., database properties, query analyses, etc.)
Evaluation (e.g., MIR specific metrics, application of traditional metrics, etc.)
System integration (e.g., development of large-scale systems, identification of real-world MIR situations, etc.)
Retrieval approaches (e.g., boolean, vector space, approximate string matching, etc.)
..and so on...

Formal papers are not required. However, we do want to have an extended abstract (approx. 1k words) and appropriate presentation materials (copies of slides, charts, etc.) for inclusion in the workshop information package.

Submission requirements: Whether you want to participate as a General Participant, Presenter and/or Panelist, we would like you to contact the Workshop Chair ASAP. We would like to have your name, background information, contact information, and the level at which wish to participate. For those wishing to present we would also appreciate brief abstract of your proposed lecture/demonstration along with any special equipment requirements. (NOTE: this does not constitute formal registration, it is merely a means by which we can better plan the workshop).

Submission Information: Send all expressions of interest, proposals, and questions, to the Workshop Chair, J. Stephen Downie [jdownie@uiuc.edu].

Deadlines: 1 July 1999 for submitting proposals for lectures and/or demonstrations; 15 July 1999 for submission of package materials (details of format TBA).

Workshop Organizers:

J. Stephen Downie, Information Retrieval Group, Graduate School of Library and Information Science, University of Illinois at Urbana-Champaign (Workshop Chair), see below.

Craig Nevill-Manning, Computer Science, Rutgers University
Dr. Nevill-Manning holds a Ph.D. in Computer Science from the University of Waikato, New Zealand, and is a founding member of the New Zealand Digital Library project (www.nzdl.org). His research interests are unified by the goal of detecting structure in sequences; sequences of amino acids for protein identification, sequences of words for digital libraries, and sequences of abstract symbols for machine learning and data compression
David Huron, Faculty of Music, Ohio State University
Dr. Huron teaches and conducts research in the field of music cognition at the Ohio State University. Professor Huron heads the Cognitive and Systematic Musicology Laboratory in the School of Music, and is affiliated with the Center for Cognitive Science. He is also an Associate Editor for the journal Music Perception published by the University of California Press.

Background of Workshop Chair:

J. Stephen Downie is a faculty member at the Graduate School of Library and Information Science, University of Illinois at Urbana-Champaign, where he teaches and conducts research in the areas of Music Information Retrieval (MIR), traditional IR, informetrics, Internet service management, indexing and abstracting, digital libraries, and information organization and access. Prof. Downie is a founding member of the University of Illinois' Information Retrieval Group. He is also completing his PhD thesis, "Evaluating a simple approach to Music Information Retrieval," at theUniversity of Western Ontario, under the supervision of Drs. Jean Tague-Sutcliffe (deceased) and Michael Nelson. Stephen holds a B.A. (music theory and composition) and a Master's in Library and Information Science, both earned at the University of Western Ontario.

Since 1992, Prof. Downie's research into MIR has garnered four prestigious awards: 2 Best Doctoral Poster; 1 Best Doctoral Paper; and the Cameron Award for interdisciplinary scholarship. He has published 5 conference papers and given many invited talks and colloquia on the issues involved in MIR. A brief CV can be found at: http://alexia.lis.uiuc.edu/~jdownie/mir_papers/cv_short.pdf

Contact information for Workshop Chair:

J. Stephen Downie, Lecturer,
Graduate School of Library and Information Science,
University of Illinois at Urbana-Champaign,
501 East Daniel St., Champaign, IL,
61820-6212
office: (217) 265-5018 fax: (217) 244-3302 home: (217) 352-7422<<preferred
email: jdownie@uiuc.edu