Conditional Statements Grounded in Past, Present and Future Grzegorz Skorupa and Radosław Katarzyniak Wroclaw University of Technology, Wybrzeże Wyspiańskiego 27 {grzegorz.skorupa,radoslaw.katarzyniak}@pwr.wroc.pl Abstract. In previous works we defined a model for grounding con- ditional statements extended with modal operators of possibility and belief. It was assumed that statements were told by an agent to describe objects at current moment. Within this work we extend the model by adding time variable. This extension allows agent to describe future and past events at known time moments. 1 Introduction One of the important aspects of agent based systems is communication. Agents usually are supposed to communicate between themselves, but it is also often required that they have abilities to effectively communicate with humans. One of the most popular trends in agent communication is tightly connected with speech act theory [6]. It specifies how agents can ask questions, make commitments, negotiate and i.e. inform each other. Speech act theory mainly specifies protocols and types of communicates but pays little attention to internal message meaning. Within this work we concentrate on the usage of conditional statements. Such statements can be embodied within inform predicate of speech act theory. We assume agent lives in an environment, where she observes objects and their properties. Objects can change properties within time. Agent’s task is to describe the viewed environment. Description is constructed from a class of nat- ural language statements. Those statements describe observed objects and cover: simple statements, complex statements having conjunctions like ‘and’ and ‘or’ and in particular conditional statements of the form ‘If ..., then ...’. The agent has only partial knowledge about the environment and is unable to view or predict all its properties. Based on her knowledge agent can evaluate how likely different properties are to occur. To effectively communicate its predictions and evaluations agent can enrich its utterance with modal operators of belief and possibility. If agent is not sure of something she can say that she believes in it or finds it possible. Our primary aim is to imitate human speech abilities when choosing proper statement to describe a situation. It is of high importance that agent chooses sentences that are not only logically true but also consistent with its mental state and knowledge. Agent should pick the statements in a similar way human does. None of the statements should mislead a human listener. For example if J.-S. Pan, S.-M. Chen, and N.T. Nguyen (Eds.): ICCCI 2010, Part III, LNAI 6423, pp. 112–121, 2010. c  Springer-Verlag Berlin Heidelberg 2010