Fuzzy Petri Nets for Human Behavior Verification and Validation

TL;DR

This paper introduces a method using fuzzy Petri nets to verify and validate human behavior models, specifically those represented as fuzzy rule-bases, by constructing and analyzing the models through a case study of professor-student interactions.

Contribution

It presents a novel approach to V&V of fuzzy human behavior models by constructing rule-bases from questionnaires and analyzing them with fuzzy Petri nets.

Findings

Effective detection of structural errors in fuzzy rule-bases

Semantic validation through Petri net reasoning

Application to professor-student interaction case study

Abstract

Regarding the rapid growth of the size and complexity of simulation applications, designing applicable and affordable verification and validation (V&V) structures is an important problem. On the other hand, nowadays human behavior models are principles to make decision in many simulations and in order to have valid decisions based on a reliable human decision model, first the model must pass the validation and verification criteria. Usually human behavior models are represented as fuzzy rule bases. In all the recent works, V&V process is applied on a ready given rule-base. In this work, we are first supposed to construct a fuzzy rule-base and then apply the V&V process on it. Considering the professor-student interaction as the case-study, in order to construct the rule base, a questionnaire is designed in a special way to be transformed to a hierarchical fuzzy rule-base. The constructed fuzzy rule base is then mapped to a fuzzy Petri net and then within the verification (generating and searching the reachability graph) and validation (reasoning the Petri net) process is searched for probable structural and semantic errors.