A Recognition Approach For Adversarial Planning Based On Plans

In recent years, plan recognition technology as an important branch of artificial intelligence, has made considerable progress. Plan recognition technology is based on the observed agents fragments, and frivolous actions to infer agent's objectives and its operating trends, which describe the sequence of its development path. Solved a plan recognizer can find many of our results are not aware that there are real results, can also provide results──agent may be found in the track. Identified by the use of many aspects of planning, so research is of great significance.In this paper, based on plans rival plan recognition algorithms, in the town plan, based on two kinds of relationships with the decomposition and abstraction to express planning problems, analysis of plans algorithm weaknesses, the incident of overall and partial and abstract and the specific relationship between the new provisions and use of "test" function of the event (planning) appears the possibility of detection, given the probability method of calculating the reconciliation plan emerged graph description of the algorithm and associated algorithm design.This presentation given by the algorithm is relatively simple in structure, using the method with or nodes, said event (Planning) between the abstract and concrete, the relationship between whole and part, so that the solution map to be better able to explain the observed phenomena. Identified during the planning process, fully rival plan recognition database to speculate agent moves appear in the calculation of the probability of the value of planning simultaneous solution of the expansion plan, generate solution plans, reducing the generation of redundant nodes. In this paper, the algorithm can deal with opponents of planning and partially observable environment, planning and other issues, which in network security and intrusion detection and many other fields have a broad application prospects.In this paper, the algorithm is given based on the use of C + + language algorithm to achieve the design of the plan recognition system, and system validation. Experiment proved that the system can achieve the desired results the theory can be quickly and efficiently identify the agent's goals, objectives of the likelihood and possible action to take.