Context Based Reasoning

• A. J. Gonzalez, W. J. Gerber, R. F. DeMara, and M. Georgiopoulos, "Context-driven Near-term Intention Recognition," Journal of Defense Modeling and Simulation, Vol. 1, No. 3, August, 2004, pp. 122 * 143.
  
  [ FULL DOCUMENT ] (Full document will be loaded onto a new page.)
  
  
• A. Gallagher, A. J. Gonzalez, and R. F. DeMara, "Modeling Platform Behaviors Under Degraded States Using Context-Based Reasoning," Proceedings of the 2000 Interservice/Industry Training, Simulation and Education Conference (I/ITSEC-2000), Orlando, Florida, U.S.A., November 27 * 30, 2000.
  
  [ FULL DOCUMENT ] (Full document will be loaded onto a new page.)
  
  
  
• Saeki, S. and Gonzalez, A. J., "Soft-Coding the Transitions Between Contexts in CGF's: The Competing Context Concept," Proceedings of Computer Generated Forces and Behavior Representation Conference, 2000.
  
  [

  ABSTRACT

  | FULL DOCUMENT ] (Full document will be loaded onto a new page.)

  
  This article describes a constraint-based approach for implementing the competing context concept in the Context-Based Reasoning (CxBR) Paradigm. CxBR is an automated reasoning paradigm that can simulate human tactical behavior simply and effectively by using an intuitive identifier called a Context. A Context, which addresses all the conditions in the current situation, controls the behavior of the AIP (Autonomous Intelligent Platform) in a tactical simulation. When the situation changes, the currently active Context searches for a possible next Context that addresses the changing situation. Once it finds one that addresses the conditions in the new situation, it deactivates itself and activates this newly found one. An AIP can be intelligently controlled through a continuous transitioning from one Context to the next as the situation demands. However, it can be very difficult to define all of Context shifting by “hard-coding” which Context would follow any other Context. While this can be realistic in some situations, such as actions upon reaching a certain phase line, or action upon orders from a superior, hard-coding the transitions in all situations is unrealistic. This would require either uncanny predictability by the system developer of the situations to be faced by the AIP, or an excessively large and complex set of Contexts, each activated by certain very specific conditions in the simulation. Both are highly unrealistic. Thus, the competing context approach has been developed to address the cases where several Contexts are able to address the situation acceptably well. This competing context concept can determine the “best” Context for the new situation and its immediate goal by using a constraint-based technique and time-warp simulation. An example of this will be provided. The competing context concept has one other significant benefit besides "soft-coding" the tactics: It can pave the way for easy on-line learning.
  
  
• Gonzalez, F. G., Patric, G., and Gonzalez, A. J., "Autonomous Automobile Behaviour Through Context-Based Reasoning," Florida Artificial Intelligence Research Society Conference, 2000.
  
  [

  ABSTRACT

  | FULL DOCUMENT ] (Full document will be loaded onto a new page.)

  
  Today’s driving simulators are used in vehicle research and design as well as in training. However, most simulators are not convincing because the degree of realism is not adequate. To achieve greater realism, a simulator must include autonomous vehicles in the environment.
  
  An essential feedback component to the driver of a car is the vehicular traffic sharing the same road environment. A car simulator must provide information about all elements of an environment that affects the driving task, such as the road layout, traffic control devices, and any autonomous vehicles or objects that exhibit reactive behavior. The simulator must also include logical information, such as information about passing lanes, lane direction, and position of one object relative to another object.
  
  The intent of this research is to develop a system that produces an interactive traffic model that behaves autonomously and intelligently. Furthermore, this model is to be effective, computationally efficient and developed with relative ease.
  
  
• Saeki, S. and Gonzalez, A. J., "Competing Context Concept: Experimental Results," Interservice / Industry Training Systems and Education Conference, 2000.
  
  [

  ABSTRACT

  | FULL DOCUMENT ] (Full document will be loaded onto a new page.)

  
  This article describes an innovative approach for implementing tactical decision-making for Computer Generated Forces (CGF). It is called the competing context concept, and it is associated with the Context-Based Reasoning (CxBR) Paradigm. CxBR is uniquely suited to represent tactical decision-making. It facilitates the simple and effective representation of human tactical behavior by using an intuitive identifier called a Context. In CxBR, there are three kinds of Contexts that are hierarchically defined: (1) Mission Context, (2) Main Context and (3) Sub-Context. The Main Contexts and Sub-Contexts provide intelligent control functions for an Autonomous Intelligent Platform (AIP) in a simulation, and address all conditions in current situation. When the situation changes, this Main Context searches for a possible next Main Context that addresses all conditions in the new situation. Upon finding such a new Main Context, it deactivates itself and activates this newly found one. No matter how the situation changes, an AIP can be controlled intelligently through a sequence of transitions among various Contexts, from the current Context to another appropriate Context. In many cases, it can be easy as well as appropriate to predefine the Context transitions based on one event. However, it can be difficult to predefine (i.e., "hardcode") these transitions in more complex tactical situations such as those typically involving military tactics as they depend on several variables. Therefore, there may be more than one viable context to which the control of the AIP can transition. This can be difficult to predefine without a multitude of rules. In such cases, it is beneficial to define the current situation as set of needs to be addressed by the AIP in order to accomplish its mission and/or survive. Likewise, the Contexts to which the control of the AIP can potentially transition are designed to address some or all of these needs. The contexts then can be said to compete for the right to become the next activated Context to control the AIP. The successful Context would ideally be the one that best addresses the identified needs of the situation currently faced by the AIP. To implement the competing context concept, we have proposed a constraint-based approach. This approach consists of four processes: (1) Situation interpretation metrics generation, (2) Relevant context group selection, (3) Context attribute matching and (4) Time-warp simulation. This article describes our continuing effort to realize the situation interpretation and soft-coding function in order to generalize the competing context concept. Experimental data, which supports the revised prototype’s performance, are described.
  
  
• Gonzalez, A. J. and Ahlers, R. H., "Conntext-Based Representation of Intelligent Behavior in Training Simulations," Naval Air Warfare Center Training Systems Division Conference, 1998.
  
  [

  ABSTRACT

  | FULL DOCUMENT ] (Full document will be loaded onto a new page.)

  
  This article presents, describes and evaluates a novel behavior representation paradigm that can effectively and efficiently be used to model the behavior of intelligent entities in a simulation. Called Context-based Reasoning (CxBR), this paradigm is designed to be applicable whenever simulation of human behavior is required. However, it is especially well suited to representing tactical behavior of opponents and teammates in simulation-based tactical training systems. Representing human behavior in a simulation is a complex and difficult task that generally requires significant investment in human effort as well as in computing resources. Conciseness and simplicity of representation and efficiency of computation, therefore, are important issues when developing models of intelligent opponents. We believe that this paradigm is an improvement over the rule-based approach, currently a common technique used in representing human behavior. We have preliminarily tested CxBR in two different prototype systems. Evaluation of the prototype shows that the context-based paradigm promises to meet the desired levels of simplicity, conciseness and efficiency required for the task.
  
  
• Gonzalez, A. J. and Ahlers, R. H., "Context-Based Representation of Intelligent Behavior in Simulated Opponents," Computer Generated Forces and Behavior Representation Conference, 1996.
  
  [

  ABSTRACT

  | FULL DOCUMENT ] (Full document will be loaded onto a new page.)

  
  This article describes and evaluates a concise, yet rich representation paradigm that could effectively and efficiently be used to model the intelligent behavior of opponents in a simulation-based tactical training system. This feature would be quite useful in the training process for two reasons: 1) the trainee would face a realistic enemy who is knowledgeable about tactics in the domain of interest and, 2) the instructor would not be burdened with playing the part of the enemy in those training systems where this is commonly done.
  
  The representation paradigm proposed is based on the idea that applicable tactical knowledge is highly dependent upon the situation being faced by the decision maker (Le., the context). A combination of script-like structures and pattern-matching rules in an object-oriented environment could serve to hold all knowledge pertinent to the context present at a specific time. This paradigm has been preliminarily tested in a prototype system that incorporates the knowledge of a submarine tactical officer on a patrol mIssion. Evaluation of the prototype shows that the context-based paradigm promises to meet the desired levels of conciseness and effectiveness required for the task.
  
  
• Gonzalez, A. J. and Ahlers, R. H., "A Novel Paradigm for Representing Tactical Knowledge in Intelligent Simulated Opponents," International Conference of Industrial Engineering Applications and A.I. and Expert Systems, 1994.
  
  [

  ABSTRACT

  | FULL DOCUMENT ] (Full document will be loaded onto a new page.)

  
  The investigation described in this paper is an on-going effort aimed at developing a concise, yet rich representation paradigm that could effectively and efficiently be used to model the intelligent behavior of opponents in a simulation-based tactical training system. This feature would be quite useful in the training process for two reasons: 1) the trainee would face a realistic enemy who is knowledgeable about tactics in the domain of interest and, 2) the instructor would not have to be burdened with playing the part of the enemy in those training systems where this is commonly done.
  
  The representation paradigm proposed is based on the idea that applicable tactical knowledge is highly dependent upon the situation being faced by the decision maker (i.e., the context). A combination of script-like structures and pattern-matching rules in an object-oriented environment could serve to hold all knowledge pertinent to the context present at the time. This paradigm was tested in a prototype system that incorporated the knowledge of a submarine tactical officer on a patrol mission. The prototype was implemented in CLIPS 5.1, a rule and object-based language developed by NASA. The results of tests with the prototype show that the context-based representation paradigm promises to meet the desired levels of conciseness and effectiveness described in the paper. Further research to be conducted in this area is also discussed.
  
  
• Gonzalez, A. J. and Ahlers, R. H., "Concise Representation of Autonomous Intelligent Platform in a Simulation Through the Use of Scripts," Florida Artificial Intelligence Research Society Conference, 1993.
  
  [

  ABSTRACT

  | FULL DOCUMENT ] (Full document will be loaded onto a new page.)

  
  The focus of the investigation described in this paper is the development of a concise, yet rich knowledge representation paradigm that could be effectively and efficiently used to model the intelligent behavior of simulated agents in a simulator-based tactical trainer. The behavior of these agents would be similar to that of an adversary who would react to a students action in a manner representative of enemy tactics. The availability of this feature would be of significant utility to the training process for two reasons: 1) the student would face a realistic enemy who is knowledgeable about tactics in the domain of interest and, 2) the instructor would not have to be burdened with playing the past of the enemy in those training systems where this is commonly done.
  
  The hypothesis presented is that a combination of a script-like structure and pattern-matching rules in an object-oriented environment could serve as the desired representation and reasoning paradigm. This hypothesis was tested through the development of a prototype system that implemented the knowledge of a submarine tactical officer on a patrol mission. The prototype was implemented In CLIPS 5,1, a rule and object-based expert system shell developed by NASA. The results of the prototype show that the combination of scripts and rules in an object-oriented environment promises to meet the requirements described above.