Stephanie J. Cammarata

Strategies of cooperation in distributed problem solving

Distributed Artificial Intelligence is concerned with problem solving in which groups solve tasks. In this paper we describe strategies of cooperation that groups require to solve shared tasks effectively. We discuss such strategies in the context of a specific group problem solving application: collision avoidance in air traffic control. Experimental findings with four distinct air-traffic control systems, each implementing a different cooperative strategy, are mentioned.

Extending a relational database with deferred referential integrity checking and intelligent joins

Interactive use of relational database management systems (DBMS) requires a user to be knowledgeable about the semantics of the application represented in the database. In many cases, however, users are not trained in the application field and are not DBMS experts. Two categories of functionality are problematic for such users: (1) updating a database without violating integrity constraints imposed by the domain and (2) using join operations to retrieve data from more than one relation. We have been conducting research to help an uninformed or casual user interact with a relational DBMS. This paper describes two capabilities to aid an interactive database user who is neither an application specialist nor a DBMS expert. We have developed deferred Referential Integrity Checking (RIC) and Intelligent Join (IJ) which extend the operations of a relational DBMS. These facilities are made possible by explicit representation of database semantics combined with a relational schema. Deferred RIC is a static validation procedure that checks uniqueness of tuples, non-null keys, uniqueness of keys, and inclusion dependencies. IJ allows a user to identify only the “target” data which is to be retrieved without the need to additionally specify “join clauses”. In this paper we present the motivation for these facilities, describe the features of each, and present examples of their use.

Dependencies and graphical interfaces in object-oriented simulation languages

An object-oriented style of computation is especially well-suited to simulation in domains that may be thought of as consisting of intentionally interacting components. In such domains, the programmer can map the constituent domain components onto objects, and intentional interactions (e.g. communications) onto message transmissions. However, some events or interactions between real world objects cannot be modeled as naturally as we might like. Improper modeling of these interactions inevitably leads to inconsistent simulation states and processing errors. The research reported in this paper identifies two categories of simulation activities that are unnatural and difficult to implement in object-oriented simulations: (1) scheduling events which depend on the continuous aspect of time; and (2) presenting a graphical display of a simulation so that any changes in the simulation state are immediately visible. Following a discussion of these deficiencies, we present a methodology for performing these tasks that is transparent to the simulation programmer. Our approach utilizes extensions to the Ross object-oriented language allowing a programmer to declaratively specify characteristics of the simulation dealing with time dependent attributes and graphics display strategies. The example presented in this paper demonstrates the many advantages of our declarative approach to maintaining consistency. With these capabilities, we expect object-oriented simulation languages to become increasingly attractive for modeling dynamic systems.