ABSTRACT
Conversation management system is finally to be provided with a model of human mental-image based understanding according to mental image directed semantic theory. In conversation management system, natural language and mental image description language (Lmd) are translated into each other via a dependency tree employed in conventional natural language processing for grammatical description. The bidirectional translation between a dependency tree and Lmd is operated by mapping rules assigned to functional words, such as verbs and prepositions, indicating how their context in natural language should be mapped into or generated from the counterpart in Lmd. Conversation management system shows the potential for people to control robotic awareness systematically by natural language. Conversation management system sends the final solution to the Animation Generator as well, to be translated into a procedure for animation. Conversation management system can understand User’s assertions and answer the questions where the locus formulas were given in Polish notation.
