Jackal

A Brief History of the Jackal Project


Jim Mayfield and Daughter Emily
In the spring/summer of 1996, Jim Mayfield crafted the first UMBC (among the first anywhere) Java implementation of KQML. Later referred to as Jackal 1.0, it is this package, Luwak, on which subsequent versions of Jackal were initially based. Luwak was put into the service of the then young CIIMPLEX effort, where it's integration and extension was engineered by R. Scott Cost.
R. Scott Cost

Not long after, Johns Hopkins Applied Physics Lab triumphed over UMBC in a bid for Mayfield's continued employment. Geographic distance made continued coordination of Luwak/Jackal's development difficult, and UMBC decided to commit itself to a splinter implementation, officially Jackal 2.0. (Development of Luwak did not continue beyond this point.) This version was largely an extended version of Jackal 1.0, and was fairly cumbersome, but met the immediate needs of the project. The principal advance was the integration of KQML conversation protocols (Finin and Labrou) as an explicit layer of the communication protocol.

As the demands of the CIIMPLEX project grew, the need for a vastly improved KQML implementation was felt. Considering the lessons learned in Jackal's first two versions, Cost undertook a what ended up being a complete rewrite of the package. This included a substantial change in the underlying architecture and the addition of a great deal of complex functionality. Consequently, the rewrite was slow, and Jackal 3.0 was not fully available until the last quarter of 1997, although alpha versions were in use considerably earlier. Significant changes from version 2.0 include the blackboard message interface, declarative, loadable conversations, and modular, plug-and-play transport modules. After 3.0, Jackal split into two different development branches; Olympus, which was closely tied to the IBM project, and Everest, which had more of a research thrust.

Jackal owes a tremendous debt of gratitude to those that have supported and guided its continuing development:
  • Principal Investigators at UMBC
    • Tim Finin
    • Yun Peng
    • Yannis Labrou
  • Original CIIMPLEX Agent Development Team at UMBC
    • Xiaocheng Luan
    • Wenlan Feng
  • Our colleagues at:
    • University of North Carolina, Charlotte (Bill Chu et al.)
    • University of Florida, Gainesville (Stanley Su et al.)
    • IBM Corporation

Milestones:

  • Spring 1996, Labrou finishes dissertation on conversation policies for KQML
  • Summer 1996, Mayfield completes Luwak (later referred to as Jackal 1.0)
  • Winter 1996/97, Jackal 2.0 released (internal)
  • Winter 1997/98, Jackal 3.0 (3.0.6 later 'Olympus') released (internal) [12/17/97]
  • Summer 1998, Jackal presented at AAAI
  • Winter 1998, Jackal 3.1 (later 'Everest') released (internal) [12/31/98]
  • Summer 1999, Cost completes dissertation on Jackal and conversation management
  • Autumn 1999, Jackal and conversation mechanisms presented at IJCAI
  • Autumn 1999, First public binary release of Jackal, through IBM AlphaWorks
  • Winter 2000/01, EECOMS Project closes
  • Winter 2001, Look for Jackal 4.0 release
 
Last updated: 12/29/2000