Animated pedagogical agents are life-like characters used as cognitive tools to support student learning. Pedagogical agents are better substitutes for the expert/tutor module of a traditional intelligent tutoring system as they provide sophisticated, real-time problem-solving advice with strong visual appeal that make learning more engaging and effective. In addition to this, animated pedagogical agents provide another important benefit: motivation. In a multi-agent environment, different agents are used for different tasks. But the main goal of a pedagogical agent is to perform a given tutoring function for the student‘s benefit. This paper presents a brief introduction of different types of agents that are used for learning and gives a survey on current developments in this field. This paper also describes the architecture and roles of an animated agent called "Pundit" which has been developed for Web-based Intelligent Learning Environment for Digital Systems (WILEDS).
Keywords: Pedagogical Agent, Intelligent Tutoring Systems.
1 Introduction: Intelligent Tutoring Systems (ITS) are complex systems that involve several different types of expertise: knowledge on the subject matter, knowledge on the learner’s knowledge, pedagogical expertise, etc [Frasson, Mengelle & A?meur 1997]. All these knowledge need to be improved in order to be adapted to the learner and the different learning situation. To make this improvement, intelligent agents are introduced which represent a new generation of human computer interface (HCI) design and are reactive, instructable, adaptable and cognitive. Animated pedagogical agents [André, Rist & Müller 1997; Johnson, Shaw & Ganeshan1998; Lester & Stone 1997; Piesk & Trogemann 1998; Sch?ch, Specht & Weber 1998] are life-like autonomous agents that facilitate human learning by interacting with learners and make computer-based learning more engaging and effective [Johnson 1998]. Besides serving as pedagogical expert and cognitive tools for learning, pedagogical agents can play different roles in a learning system. They can collaborate with learners and other agents thus create a cooperative learning environment and also are able to play a powerful motivational role.
We discuss about the general definition and taxonomy for agents in section 2. In this section, we also discuss about the pedagogical agents and their categories. Section 3 briefly discuses about the desired attributes for any pedagogical agent. Following section presents a survey on the current developments on this area. This section also shows the three architectural categories of pedagogical agents. Section 5 introduces internal architecture of Web-based Intelligent Learning Environment for Digital System (WILEDS). We present Pundit in section 6, focusing on its system architecture, followed by conclusions.
2 Pedagogical Agents: Russel et al. [Russel & Norvig 1996] defined agents as a combination of architecture and program. According to them, architecture means hardware components. So, an agent is able to perceive the environment through hardware sensor and act upon the environment through effectors. But from the software viewpoint, an agent is mainly a computer program which has a specific plan of action defined in a limited domain and a behaviour space. This allows an agent to change its interaction with the outside world at the right moment depending on the stimuli from the environment. This technology combines artificial intelligence (reasoning, planning, natural lanuage processing, etc.) and system development techniques (object-oriented programming, scripting languages, human-computer interface, distributed processing, etc.). Franklin et al [Franklin & Graesser 1996] provides taxonomy of agents, which is shown in Figure 1.
Figure 1: Natural Taxonomy of Agents
Pedagogical agents are a type of autonomous agents who incorporate multiple characteristics of Task-Specific and Entertainment Agents. Pedagogical agents can be divided into two categories: A. Goal-Driven (tutor, mentor, assistance), B. Utility-Driven. The Table 1 shows the characteristics of goal-driven pedagogical agents.
Tutor Mentor Assistant
Knowledge about environment Strong Strong Medium
Domain Expert Strong Strong Strong
Student Model Strong Medium Weak/Nil
Pedagogical aspects Strong Medium Weak/Nil
Table 1: Characteristics of Goal-Driven Pedagogical Agents
The utility-driven agents are used for the pedagogical purposes like help students to find reference learning materials, schedule group meetings, remind the deadline for submitting assignments etc.
3 Desirable Attributes for Pedagogical Agents: Pedagogical agents appear to the student commonly as animated 2D or 3D cartoon style characters. Research shows that an animated pedagogical agent should have the following properties (Table 2):
Autonomous Has to be able to perform the majority of its problem-solving task without the direct intervention of human or other agent. It has to have a degree of control over its own actions and internal state.
Reactivity Has to be reactive since they are able to sense changes in the environment and respond to it over a certain period of time.
Goal-directed Must have proactiveness. They are to exhibit goal-directed behaviour by taking the initiative [Wooldridge & Jennings 1995].
Communicative Must respond to learners with a combination of verbal communication and non-verbal gesture such as gaze, pointing, body stance, and head nods. They can convey emotions such as surprise, approval, or disappointed.
Persona effect They should give the learner an impression of being lifelike and believable. They are able to interact with the learners on an ongoing basis.
Subservient Act on behalf of someone else, most of the times tutor.
Adaptive Act according to the learner’s cognitive state.
Social ability Capable of communicating with humans or other agents through some kind of agent communication languages in order to satisfy its design objectives.
Table 2: Properties of a Pedagogical Agent
Besides the above-mentioned properties, pedagogical agents can be mobile, capable of moving from one physical place to another. They can learn from the environment by observation.
4 Survey of Pedagogical Agents: In the recent years, animated pedagogical agents in the interface of the learning systems have become increasingly popular. In this section, we illustrate a brief description of several implemented animated pedagogical agents and their attributes.
A. Herman-the-Bug: Herman-the-Bug [Lester, Stone & Stelling 1999] is a lifelike agent whose visual and verbal actions are controlled by a real-time behaviour-sequencing engine [Stone & Lester 1996] in response to changing problem-solving contexts. It performs a broad range of activities including walking, flying, swimming, shrinking, expanding, fishing, bungee jumping, teleporting and acrobatics. It inhabits Design-A-Plant, a learning environment for the domain of botanical anatomy and physiology.
B. Cosmo: Cosmo [Lester, Voerman, Towns & Callaway 1999] is a 3D character that performs the role of the Internet Advisor, a learning environment for the domain of Internet packet routing. Cosmo is designed to study spatial deixis in pedagogical agents, i.e. the ability of agents to dynamically combine gesture, locomotion and speech to refer to objects in the environment while they deliver problem-solving advice [Johnson, Rickel & Lester 2000].
C. Adele: Adele [Johnson, Shaw & Ganeshan 1998] is designed to operate over the Internet. Adele architecture implements key pedagogical functions: presentation, student monitoring and feedback, probing questions, hints and explanations. Adele-based courses are currently being developed for continuing medical education in family medicine and graduate level geriatric dentistry.
D. Steve: Steve [Rickel & Johnson 1998] is a stereoscopic 3D character that cohabits with learners in networked immersive virtual environments and has been applied to naval training tasks such as operating the engines aboard US Navy surface ships.
E. SmartEgg: SmartEgg [Mitrovic & Suraweera 2000] is developed on the same architecture as that of Adele and it inhabits in SQLT-Web, an intelligent SQL tutor on the Web.
F. PPP Persona: PPP Persona [André, Rist & Müller 1997] is an animated pedagogical agent for interactive WWW presentation. The persona guides the learners through Web-based materials by showing, pointing, explaining and verbally commenting textual and graphical output on a window-based interface.
Johnson identified that the current implementations of pedagogical agents can be categorised into three architectures:
I) Behaviour Sequencing Approach: Behaviours are assembled from a collection of prerecorded primitive animations, sounds and speech elements. These elements are organised in a behaviour space. Herman-the-Bug is designed according to this approach [Stone & Lester 1996].
II) Layered Generative Approach: Animations are generated in real-time. The architecture is divided into cognitive decision making layer and perception motor layer that is responsible for monitoring the environment and generating the animation. The cognitive layer continually evaluates the state of the environment and makes the decision about the agent’s action. The perception motor layer carries out these actions. Steve’s architecture is an instance of layered generative approach [Rickel & Johnson 1998].
III) State Machine Compilation Approach: It addresses the issue of real-time adaptation of the agent’s behaviour, while limiting the real-time animations. This approach is also based on the behaviour space in a similar manner to the behaviour sequencing approach. However, these behaviours are executed by a state machine that can adapt at run time to student’s action. PPP Persona [André, Rist & Müller 1997] has implemented this approach.
5 Web-based Intelligent Learning Environment for Digital System (WILEDS): WILEDS [Kazi, Kassim & Ranganath2000] is a Web-based intelligent tutoring system that helps student to learn Digital System Course. Empirical studies [Bloom 1984] shows that effective individual tutoring is the most powerful mode of teaching. Logistically and financially, however, individual tutoring for all students is not possible, but web-based intelligent learning environment can bring personal tutoring experience to a broader audience. This philosophy drove us to build WILEDS. André et al. [André, Rist & Müller 1998] stated that animated pedagogical agents could cause learners to feel that on-line educational material is less difficult. They can increase learner’s motivation and attention [Lester, Converse, Kahler, Barlow, Stone & Bhogal 1997]. An animated pedagogical agent called Pundit was introduced to WILEDS with the expectation of making the learning process less difficult and more engaging.
We explain the internal architecture of WILEDS in the context of "Combinatorial Logic Circuits" chapter of Digital Systems course. Students are given a logic expression (dynamically generated based on the student model) to find out the simplified and minimized form. Student’s work is assessed by the system online and necessary feedback is given on the basis of the correctness of the work.
WILEDS‘s system model comprises five components (Figure 2) that are as follows: 1. Student Model. 2. Pedagogical Module. 3. Domain Knowledge. 4. Communications Module and 5. Expert Model. This system model is based on the model designed by Joseph Beck et al. [Joseph, Stern & Erik 1996], which is derived from the model developed by Woolf [Woolf 1992]. In WILEDS, Pundit performs the role of the tutor (expert).
One of the main technical goals for this system model is to utilize the solid foundation of intelligent tutoring systems as a platform for an intelligent web-based learning environment.
WILEDS‘s external architecture is implemented by six main software components: the client side software (browser), the WWW server, the Applet, the Servlet, JDBC-ODBC driver and the Student database. The browser (i.e. MSIETM, NetscapeTM) and the WWW server are the basic software for HTTP communications over the Internet. The applet is the engine for Communication module, expert model and the pedagogical module. The servlet in conjunction with the JDBC-ODBC driver makes communication between the student database and the applet.
Figure 2: Relationship among the different components in WILEDS’s internal architecture
6 Pundit – Animated Pedagogical Agent for WILEDS: Pundit is a 2D character of a wise man (Figure 3) implemented as a Java Applet whose main function is to give feedback on student’s work. Pundit inhabits in WILEDS which is a fully functional ITS.
Figure 3: Pundit – Animated Pedagogical Agent for WILEDS
System architecture (Figure 4) of Pundit is much simpler. It has two components:
1. Animated Persona
2. Reasoning Engine
Figure 4: System Architecture of Pundit
Animated Persona is connected with the behaviour space that has a library of gestures and dialogues. The behaviour of Pundit is static, based on predefined gestures and dialogues. Reasoning Engine is connected with the Expert Model and the Student Model of WILEDS. Expert Model computes the correct answer for any given problem and Student Model holds the student’s current knowledge about the subject matter. Feedback is generated by the Reasoning Engine based on the cognitive information of the student and his work, then presented by the Animated Persona. Currently, Pundit has only the Congratulatory behaviours. Pundit congratulates students when a correct answer is submitted and displays disappointment after an incorrect submission.
7 Conclusion: Animated pedagogical agents, although still in its infancy, offer a great deal of promises for interactive learning environments. By broadening the communication bandwidth between teachers and students, animated pedagogical agents are slowly bringing the ITS towards its target.
This paper has given an overview of animated pedagogical agent and its characteristics. A survey of current development on this area is also furnished. Lastly, described the system architecture of Pundit – an animated agent for WLEDS.
References:
André, E., Rist, T. & Müller, J. (1997). WebPersona: A Life-Like Presentation Agent for Educational Applications on the World Wide Web. Proceedings of the workshop “Intelligent Educational Systems on the World wide Web”, 8th World Conference of the AIED Society, Kobe, Japan 1997André, E., Rist, T. & Müller, J. (1998). Integrating Reactive and Scripted Behaviors in Life-like Presentation Agents. In K. P. Sycara and M. Wooldridge (Eds.), Proceedings of the Second International Conference on Autonomous Agents, pp 261- 268, 1998
Bloom, B. S. (1984). The 2-sigma problem: The search for methods of group instruction as effective as one-to-one tutoring. Educational Researcher, Vol.13, pp 3-16
Franklin, S. & Graesser, A. (1996). Is it an Agent, or just a Program?. In Proceedings of the Third International Workshop on Agent Theories, Architectures and Languages. Springer-Verlag. ~
Frasson, C., Mengelle, T. & A?meur, E. (1997). Using Pedagogical Agents In a Multi-Strategic Intelligent Tutoring System. Proceedings of the workshop “Pedagogical Agent in AI-ED’97”, 8th World Conference on AI and Education. Japan, August 1997
Johnson, W. L. (1998). Pedagogical Agents. In Proceedings of ICCE’98, vol. 1, pp 13-22. China Higher Education Press and Springer-Verlag, Beijing. 1998
Johnson, W. L., Rickel, J. W. & Lester, J. C. (2000). Animated Pedagogical Agents: Face-to-Face Interaction in Interactive Learning Environment. International Journal of Artificial Intelligence in Education, Vol 11, pp 47-78
Johnson, W. L., Shaw, E. & Ganeshan, R. (1998). Pedagogical Agents on the Web. Workshop on WWW-based Tutoring, ITS’98, San Antonio, Texas, 1998
Joseph, B., Stern, M. K. & Erik, H. (1996). Applications of AI in Education. ACM Crossroads, Sept. 1996
Kazi, S. A., Kassim, A. A. & Ranganath, S. (2000). WILEDS: Web-based Intelligent Learning Environment for Digital Systems Course. To appear in the Proceedings of International Conference on Education and ICT in the New Millennium, Kualalumpur, Malaysia, October 2000.
Lester, J. & Stone, B. (1997). Increasing Believability in Animated Pedagogical Agents. Proceedings of the First International Conference on Autonomous Agents, pp. 16-21, California, February, 1997
Lester, J. C., Converse, S. A., Kahler, S. E. Barlow, S. T., Stone, B. A. & Bhogal, R. S. (1997). The Persona Effect: Affective Impact on Animated Pedagogical Agents. In Proceedings of CHI’97, pp 359-366. ACM Press, 1997
Lester, J. C., Stone, B. A. & Stelling, G. D. (1999). Lifelike Pedagogical Agents for Mixed-initiative Problem Solving in Constructivist Learning Environments. User Modeling and User-Adapted Interaction 9:1-44
Lester, J. C., Voerman, J. L., Towns, S. G. & Callaway, C. B. (1999). Deictic Believability: Coordinating Gesture, Locomotion, and Speech in Lifelike Pedagogical Agents. Applied Artificial Intelligence. Vol. 13, pp 383-414
Mitrovic, A. & Suraweera, P. (2000). Evaluating an Animated Pedagogical Agent. In the Proceedings of the Fifth International Conference, ITS 2000, Montreal, Canada, June 2000, pp 73-82
Piesk, J. & Trogemann, G. (1998). Presenting Educational Contents in Nonlinear Narrative Structures by Conventional Virtual Actors. Proceedings of ED-MEDIA’98, pp. 1885-1888, 1998
Rickel, J. & Johnson, W. L. (1998). STEVE: A Pedagogical Agent for Virtual Reality. Appears in the Proceedings of the Second International Conference on Autonomous Agents, Minneapolis/St. Paul, ACM Press, May 1998
Russel, S. & Norvig, P. (1996). Artificial Intelligence: A Modern Approach. New Jersey, Prentice Hall, 1996
Sch?ch, V., Specht, M. & Weber, G. (1998). “ADI” – An Empirical Evaluation of a Tutoring Agent. Proceedings of ED-MEDIA’98, pp. 1271-1282, 1998
Stone, B. A. & Lester, J. C. (1996). Dynamically Sequencing an Animated Pedagogical Agent. In Proceedings of the Thirteenth National Conference on Artificial Intelligence. pp 424-431, 1996
Wooldridge, M. & Jennings, N. R. (1995). Intelligent Agents: Theory and Practice. London: QM & Wc, 1995
Woolf, B. (1992). AI in Education. Shapiro, S. (ed.) Encyclopedia of Artificial Intelligence. John Wiley & Sons, Inc., New York, NY, 1992. pp 434 - 444
中国论文网(www.lunwen.net.cn)免费学术期刊论文发表,目录,论文查重入口,本科毕业论文怎么写,职称论文范文,论文摘要,论文文献资料,毕业论文格式,论文检测降重服务。
