In conjunction with the ACM+IEEE Joint Conference on Digital Libraries (JCDL 2001) a meeting of the satellite site [SATSITE] participants in the Indiana University Digital Music Library project was held 27 June 2001 in Roanoke, Virginia.
Indiana University: Kristine Brancolini, Mary Wallace Davidson, Andrew Dillon, Jon Dunn, Eric Isaacson, Jim Halliday, Harriette Hemmasi, Matt Jadud, Doug Pearson and Rob Pendleton
Institut de Recherche et Coordination Acoustique/Musique (IRCAM): Michael Fingerhut
Loughborough University: Cliff McKnight
Northwestern University: Claire Dougherty, and Bill Parod
University of Oxford: Matthew Dovey
University of Illinois at Urbana-Champaign: J. Stephen Downie
A collection of selected project documents was provided to the participants and discussed. The documents included:
In the research and development phase of the project we don't want to restrict content. Content will be placed in the system according to what people need to conduct their DML work - evaluation, testing and pilot-phase instruction. The DML Copyright Research Area will carry on activity in parallel with development and pilot phases of the other Research Areas. Final guidelines regarding content in the system will be implemented upon completion of the Copyright Research Area work. The distinction of development versus production phase content guidelines will apply to Indiana and to satellite site use of the system.
User interface to DML will be through bibliographic search and retrieval, the Multimedia Music Theory Teaching [MMTT] application and through the Indiana University OnCourse [ONCOURSE] integrated, web-based course management system and portal. The interface via OnCourse is relevant solely to Indiana DML implementations.
A staged delivery plan has been developed for the DML system and applications. First phase deliverables are scheduled for release at the end of year, 2001. Subsequent releases of the DML will occur at six month intervals.
Indiana is not currently working on a content-based search and retrieval mechanism, but we do want to be able to plug in content-based tools in the future. Currently, all search and retrieval is metadata-based. The modular DML system architecture is designed to facilitate incorporation of new features and capabilities in the future.
Over the past six months, work on the DML has concentrated on requirements definition, metadata specification, user interface for searching and result sets manipulation, and interfacing to the media. Very soon, the user interface design for overall system functionality will be available for review. In accordance with the system development milestones [MILESTONES], final system requirements definition will be completed soon.
DML application software will be made available to the satellite site community in distinct version releases, plus patches. The system will be Java-based and we will expose the class libraries. Some form of Java, "web start" control of versioning will likely be employed.
The DML repository and server will be based on a relational database, Oracle is being considered. Media streaming will utilize MP3 in Apple QuickTime, using the Darwin server. OS will be Linux or other Unix variant. Applications will be in Java. Authentication technology is an open question.
Discussion took place regarding the selection of Oracle for database management system (DBMS). It's recognized that the use of a licensed DBMS such as Oracle can pose problems for distributed deployment of the DML server. Oracle does provide certain features, such as full text searching and thesaurus support which would be difficult and time consuming to duplicate in an open sourced DBMS.
A question was raised regarding DML integration or alignment with the MPEG-7 standard for multimedia information search, filtering, management and processing. Currently, work in the DML metadata research area is focused on identification of the specific metadata elements required. Initial implementation is guided by a MARC mindset. The DML approach is to practical and currently feasible implementations. Developing standards, e.g. MPEG-7, are being tracked for their potential, and the DML is being architected to readily accomodate new standards and technologies. The DML Metadata team welcomes input.
Matthew Dovey; University of Oxford
Oxford music faculty interest in satellite site participation is geared to use of DML in teaching practices. Faculty will expect to utilize the DML for listening lists and online courseware for continuing education. Online courseware is an area of Oxford expertise. Oxford also posesses expertise in metadata which may provide value to the DML. Interests include preservation and metadata for preservation. Oxford has a collaboration agreement with John Hopkins for development of optical recognition of scores to MIDI notation, and can provide linkages of the project to DML. Also, researchers at Oxford have done some systems work that may be of use in the DML, particularly a Java implementation of Z39.50.
Claire Dougherty and Bill Parod; Northwestern University
Northwestern is in the beginning phases of developing music streaming capability in the music library. A question was raised regarding location of a satellite site DML library - would Northwestern implement a repository and server locally or contribute to a shared system. Initially, the satellite sites will access a central repository and server at Indiana. Distributed systems can be implemented later, according to individual site requirements. People at Northwestern involved in the project include Peter Webster (Music faculty), Don Roberts (head of Music Library), Claire Dougherty (Digital Media Services in the Main Library) and Bill Parod (IT division, Academic Technical Support).
Cliff McKnight; Loughborough University
Cliff is with the Information Sciences department, and has interest in usability, copyright and preservation. Andrew Dillion (IU) expressed great interest to have the satellite sites evaluate usability of the applications.
J. Stephen Downie; University of Illinois at Urbana-Champaign
Stephen is intriqued by the possibilty of integration of distributed satellite sites into a network. An application at a single site could go out to multiple servers in a network of sites to pull various results sets and media types from distributed libraries, e.g. MIDI from one library, score from another and music from a third.
Michael Fingerhut; IRCAM
IRCAM is not a satellite site, but Michael Fingerhut attended the meeting at invitation. IRCAM researchers have experience in the development of an audio/video delivery technology and system that may provide lessons-learned and technique value to the DML project. Michael is particularly interested in networking issues, including network quality of service; encoding methods; and has performed research into delivery methods appropriate to the users' location and use of music. He believes that a metadata strategy based on MARC is a good idea. Michael brought our attention to the WEDELMUSIC project [WEDEL], which provides a method to distribute interactive music via the Internet while respecting publisher rights and protecting from copyright violation.
The 2nd annual International Symposium on Music Information Retrieval 2001 [ISMIR] will be held 15-17 October 2001 at Indiana University.
Listservs focusing interests for specific Research Areas will be created.