本論文參考JAM及Jason的推理週期，探討兩者的優點及特性，簡化後提出一套點單的agent推理週期。使用Java agent development framework (Jade)作為底層平台，在該平台上實作belief-desire-intention (BDI) model來賦予代理人思考能力。
此model使用Java撰寫，消除了使用logic programming language的平台不易撰寫程式的缺點。此外，具有模組化的model，讓使用者只需撰寫部分的code，便可生產出可執行的代理人程式，使代理人程式比以前容易開發。 ;The intelligent agent system has become an important technique in the distributed system, and it is widely applied in the fields of computer science. Because the concept of intelligent agent system is similar to human thinking, the system is thus able to simulate human thinking by automatically executing program tasks. However, we noticed the difficulty of using logic programming language and the lack of modularized framework have made the development of agent programs very slow. We figure that the software industry needs a model and a platform that are practical and simple for rapid development of intelligent agent systems.
This paper reviews JAM and Jason’s reasoning cycles, discusses their advantages and characteristics, and then proposes a simple agent reasoning cycle. This reasoning cycle uses Java agent development framework (Jade) as the underlying platform, on which the belief-desire-intention (BDI) model is implemented to endow agent with thinking ability.
This reasoning cycle is written in Java to eliminate the difficulty of writing such programs using logic programming language on certain platforms. In addition, a modularized framework is used to enable users to generate agent programs by writing only a small part of code, making the development of agent programs easier than before.
