Multi-Agent Systems as Parallel Logic Programs

We present two simple approaches to implementing MAS using currently available technologies for parallel logic programming (Prolog). The first approach employs the Linda tuple space concept introduced in the early eighties: agents are independent Prolog processes that communicate via the tuple space. The second approach uses multi-threading: agents share the same knowledge base but each of them is executed by its own independent thread within one same Prolog session. We have implemented platforms incorporating each approach and we shall briefly explain them, listing their advantages and drawbacks.