Exhaustive simulation of consecutive mental states of human agents

摘要

We develop a generic software component for computing consecutive plausible mental states of human agents. The simulation approach to reasoning about mental world is introduced that is based on exhaustive search through the space of available behaviors. This approach to reasoning is implemented as a logic program in a natural language multiagent mental simulator NL_MAMS, which yields the totality of possible mental states few steps in advance, given an arbitrary initial mental state of participating agents. Due to an extensive vocabulary of formally represented mental attitudes, communicative actions and accumulated library of behaviors, NL_MAMS is capable of yielding much richer set of sequences of mental state than a conventional system of reasoning about beliefs, desires and intentions would deliver. Also, NL_MAMS functions in domain-independent manner, outperforming machine learning-based systems for predicting behaviors of human agents in broad domains where training sets are limited.We evaluate the correctness, coverage and maximum complexity of the NL_MAMS and discuss its integration with other reasoning components and its application domains. The proposed component is intended to be integrated into eBay human behavior simulation system, predicting behavior of buyers and sellers in normal and conflict situations. Also, NL_MAMS can be a part of any software system where modeling of human users is necessary, such as a personalized assistant, a tutoring or decision support system, advisor, recommender and conflict resolver.