Rule-Drove Social Force Behavior Modeling

The study on behavior modeling of human crowd has recently been considered as very important. Social force model which formalize individuals' mental and emotional elements as forces has remarkably established the base of study on human crowds' dynamics. Nevertheless, social force model doesn't give details about how critical parameters in formulas of social force are fixed or changed according to agent's decision-making process. We propose, therefore, modeling methods of rule-drove social force behavior which is in charge of perception, reasoning and motor control. Being different from other currently wide used models such as Soar and ACT, we tried to represent perceptive elements (short term knowledge) in semantic net by using an OWL/XML version and query them by SPAQL, an access control model for data searching. There are three kinds of rules (long term knowledge) are adopted to support decision-making process, and we also studied in this paper the mechanism of rules matching. At last a Scenario was given and simulated.