Agent Behaviour Recognition Research

AI research focuses more and more on explainable technology that takes into account the results of programs and products. One of the main important aspects of this direction is the ability to recognize the behavioral patterns of our application for making intelligent and informed decisions. In this work, we analyze the planning agent's behavior in solving a simple logistics problem. Also in this work, we will show a new method of recognizing behavior that is based on machine learning methods. The idea is to view the plan as plain text, not as a structure of action names and objects. Thus, it is possible to use text analysis methods that can be used to extract the structure that is at the heart of the plan. To summarize these methods, you can try to recognize the behavior that the agent follows. We conducted several experiments with several methods of trait extraction and several classification methods for two well-known planning areas. This approach shows promising results, it can have a good development. This paper consists of 43 pages, 4 tables, 9 figures and 31 references in bibliography.