The Bucket Matching System: Design and Initial Implementation

                                B. Danette Allen
                             b.d.allen@larc.nasa.gov        
                   CS 845: Introduction to Digital Libraries
                                   2/19/98


Buckets are digital library (DL) construct to provide both aggregation and
intelligence to archival objects [1].  Buckets provide aggregative qualities 
by allowing like data objects to be grouped in a user definable manner.
Buckets act as intelligent agents in that they can communicate with other
buckets, with people, or with arbitary network services.  The mechanism for 
bucket communication is known as the Bucket Matching System (BMS).

Some of the proposed behaivors for the BMS have been proposed in [1], but 
design will document the core methods for bucket <-> BMS communication.
The initial implementation will demonstrate the BMS by allowing buckets
to "register" their metadata with the BMS, and a separate process in the BMS
will find likely "matches" (based on similarity of metadata) of the registered
buckets.   We will study the Linda [2] and Common Object Request Broker
Architecture (CORBA) [3] communications models for applicability to the BMS.




1.  M. L. Nelson, K. Maly, S. N. T. Shen and M. Zubair, "Buckets: Aggregative,
    Intelligent Agents for Publishing," Old Dominion University CS TR-97-41,
    December 1997.

2.  N. Carriero and D. Gelernter, "Linda in Context," Communications of
    the ACM, 32(4), 1989, pp. 444-458.

3.  S. Vinoski, "CORBA: Integrating Diverse Applications Within Distributed
    Heterogeneous Environments," IEEE Communications, 14(2), 1997.