軟體業界極需實用而簡單的model 及平台 來迅速開發智慧代理人程式 ,而現有開發agent的平台Jason 及Jadex都頗繁雜不易使用,Jason 使用Agent-Speak Language,過於抽象不易理解及撰寫;而Jadex 使用XML 來記錄goal 與plan head 過於複雜且不易使用。 本研究用流行的Java agent development framework (Jade) 作為平台發展一套model 稱作Jadexx ,希望它可以讓agent-oriented programming paradigm 繼承object-oriented programming paradigm而流行。它不像Jadex 是個要安裝的平台。它一方面提供上層代理人思考能力,另一方面藉著既有而穩定的Jade 平台提供下層代理人的溝通,以簡單便利的開發智慧代理人。 Jadexx使用Java撰寫,並且goals只以OWL檔紀錄,改進Jason與Jadex不易撰寫的缺點,此外實作meta-plan改進Jadex的meta-level reasoning的不足。
The software industry extremely needs a practical and simple platform to develop agents quickly. However, the existing software platforms such as Jason and Jadex are all difficult to use. Because Jason uses Agent-speak language, it’s so abstract that it is not easy to understand. Jadex uses XML to store the goals and plan head, this is too complicate. Based on the popular Java Agent Development framework (Jade) platform, we develop a model called Jadexx. Unlike Jadex, Jadexx needs no installation. We hope that, through Jadexx, agent-oriented programming can follow the popular object-oriented programming paradigm to be widely accepted. On one hand, Jadexx provides high-level agent deliberation. On the other hand, the underlying Jade provides low-level communication among agents. Jadexx uses Java to implement, and using only OWL files to store goals. It thus improves the usability, compared with Jason and Jadex. In addition, Jadexx implements meta-plan to remedy the weakness of Jadex meta-level reasoning.