Agent Communication Languages (ACLs), such as the Knowledge Query and Manipulation Language (KQML), provide a formal structure for interaction between software agents. Unlike traditional method invocation in programming languages like Python or Java, which assumes shared memory or tightly coupled systems, ACLs are designed for distributed, loosely coupled environments (Labrou, et al., 1999) .
One key advantage of ACLs is their emphasis on semantics. KQML, for example, supports a range of performatives (e.g., “ask”, “achieve”) that indicate communicative intent, enabling agents to reason about and respond appropriately to messages. This supports autonomy and flexibility in multi-agent systems (Mayfield, et al., 1996). ACLs also facilitate interoperability across heterogeneous systems, which is crucial for open environments such as the web or Internet of Things (IoT) systems.
However, ACLs have their disadvantages. The semantic richness they offer can be difficult to interpret consistently across platforms, leading to interoperability issues in practice. Additionally, there is no universally accepted standard for agent communication semantics, which affects scalability and reliability (Singh, 2003). In contrast, method invocation is simpler and better suited to tightly integrated systems where communication overhead and interpretation complexity must be minimised.
Overall, ACLs like KQML are better suited for applications requiring agent autonomy, negotiation, or coordination in distributed environments. However, for straightforward function calls in centralised systems, conventional method invocation remains more efficient and practical.
References
Labrou, Y., Finin, T. and Peng, Y. (1999) ‘Agent communication languages: the current landscape’, IEEE Intelligent Systems and their Applications, 14(2), pp. 45–52.
Mayfield, J., Labrou, Y. and Finin, T. (1996) ‘Evaluation of KQML as an agent communication language’, in Intelligent Agents II Agent Theories, Architectures, and Languages. Berlin, Heidelberg: Springer Berlin Heidelberg (Lecture notes in computer science), pp. 347–360.
Singh, M.P. (2003) ‘Agent communication languages: Rethinking the principles’, in Communication in Multiagent Systems. Berlin, Heidelberg: Springer Berlin Heidelberg (Lecture notes in computer science), pp. 37–50.