| From the variety of agent communication languages that have been proposed since the early 1990's, it can be seen that there is no general agreement on what constitutes a necessary core of semantics for agent communication languages. This has led some researchers to observe that agent communication languages have been implemented ad hoc. This dissertation develops and evaluates an approach towards representing the semantics of agent communication languages explicitly in the notation of conceptual graphs. The speech act theories of Austin, Searle and Habermas that are cited as having motivated the semantics of some agent communication languages are also represented in conceptual graph notation. These explicit representations support comparing the semantic features of agent communication languages against those of established speech act theories. The explicit representations also support comparing different agent communication languages with each other. The comparisons provide a means of systematically identifying which semantic features from the speech act theories are implemented in agent communication languages. The comparisons also identify some semantic features from these theories that are not currently implemented, but that may be useful for agent communication languages to incorporate in the future. Because there is no general agreement to date as to what comprises a core semantics for software agent communication acts or how these languages should be specified, a semantic communication framework, such as the Composite Semantic Communication Framework developed in this dissertation, may serve to clarify and identify which semantic features are currently employed in various agent communication languages, and also serve as a basis from which to explicitly specify and communicate agent communication language semantic features. As such, it is intended to support both software developers that implement multi-agent systems and also software agents that may benefit from reasoning about agent communication language semantic features. |
