This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Children's conceptual understanding can be developed using guided reading exercises based on non-fiction texts. Although ‘reading for understanding' is a skill that children are encouraged to acquire at an early age, most researchers agree that students find it difficult to learn using this method and continue to need support from their teachers as they acquire cognitive skills and learn new concepts. There is a need for research into new and better ways for providing support for children's learning during text-based activities.
It is generally acknowledged that successful teachers are skilled at promoting the sort of reasoning processes that develop children's conceptual understanding. Experienced tutors have comprehensive domain knowledge, know their students personally and are able to adapt their explanations to individual students who, as a consequence, gain deeper understanding of new concepts. As a result of receiving this kind of help, children are able to understand new concepts more effectively and become autonomous readers. Without such support, the students are less likely to be able to understand the new concepts presented in their guided reading lessons. Therefore, it is important to find ways of optimising the support available to help students and guide their reasoning processes.
One way forward might be to employ more human tutors. Unfortunately, most educational institutions find it difficult to provide human tutors in sufficient numbers to support students adequately and this is especially true in the case of guided reading because it is an activity that demands a series of one-to-one interactions between the tutor and the student. It often proves too costly and impractical to provide such intensive support.
An alternative solution would be to use computers in the classroom to replace some or all of a human tutor's role and support students as they learn new concepts. This would enable institutions to provide the one-to-one support necessary for effective learning from guided reading at a much lower cost than at present. A pedagogical agent based on recent advances in research into multimedia and artificial intelligence and capable of explaining new concepts to students could simulate some or all of the functions of an experienced human tutor. The domain and student knowledge available to such an agent would be represented in knowledge bases, and artificial intelligence methods used to infer explanation strategies that best matched a student's learning needs. If based on an appropriate learning theory, this new pedagogical agent would generate adaptive explanations similar to those supplied to students by human tutors. Research in the field of Artificial Intelligence in Education (AIED) has shown that existing tutoring systems do not in general support cognitive tasks in a way that leads to conceptual understanding (Luckin & du Boulay, 1999; Aist, 2001). Thus, the goal of our research is to build and test a pedagogical agent that is capable of supporting children's conceptual understanding.
The design of a pedagogical agent would require two lines of enquiry: theory and experiment. A theoretical consideration of the interpretation process would define the tasks to be supported by the pedagogical agent and experimental work would identify the explanation strategies to be employed by the computer tutor. This combination of theoretical and practical work would create a framework of teaching principles and strategies based on an appropriate learning theory.
This thesis makes the case for the use of a pedagogical agent capable of supporting children as they learn new concepts through guided reading activities. A pedagogical agent designed in such a way that it simulates the explanation strategies applied by human tutors should support children's learning effectively. Our hypothesis is that an interaction between a child and a pedagogical agent that supports the child's interpretation process will help the child better understand new concepts, and that educational systems incorporating the pedagogical agent will be more effective at explaining new concepts than educational systems without it.
Following the discussion above, our research questions are:
- How do children understand new concepts in guided reading exercise?
- How do human teachers support children's conceptual understanding in one-to-one interactions?
- How do we design a pedagogical agent capable of supporting children's conceptual understanding?
- Is the pedagogical agent effective and useful at supporting children's conceptual understanding?
1.1 The Methodology of This Research
The objective of this study is to design and test an architecture for a pedagogical agent capable of supporting children's conceptual understanding. Following Self's (1999) ITS research methodology that combines theoretical and empirical investigations, we have conducted this research undertaking the following steps:
- Investigation of a learning theory that explains how humans understand new concepts. We refer to the learning theory called schema theory (Bartlett, 1958) that explains how humans understand a new concept; the teaching principles of this learning theory will be analysed in order to derive design principles for our pedagogical agent.
- Examination of tutoring strategies used by human teachers to identify how the agent will explain new concepts to children. These strategies will define the teaching knowledge of the agent and will be used as a key source for dialogue planning.
- Precise description of the behaviour of the agent including a mechanism for utilizing the domain expertise, reasoning about the student's conceptual understanding, and conducting explanatory dialogue to promote schema-based cognitive tasks.
- Implementation of the pedagogical agent to illustrate the validity of the formal description. It is critical that the agent will be implemented in an educational system to demonstrate its role in learning environments.
- Evaluation of the pedagogical agent integrated into an educational system in a real setting with students to validate the design principles and examine the potential benefits of the approach.
1.2 Possible Contributions of the Work
This work is expected to contribute to the fields of AIED, education and knowledge-based systems.
Artificial Intelligence in Education (AIED). The design of an interactive pedagogical agent capable of explaining new concepts to primary schoolchildren and to support their conceptual understanding will be a new application of artificial intelligence techniques to multimedia educational systems. In particular, a design architecture for a pedagogical agent based on the principles of schema theory will be established and the new agent will build on an examination of dialogue strategies to support schema-based cognitive tasks and interact with the students to support cognitive tasks. Using schema-based reasoning, the agent will be capable of simulating the help offered by human teachers by generating adaptive explanations tailored to the needs of an individual student.
Education. An analysis of schema theory will provide new teaching principles that could be considered by the designers of pedagogical agents. An examination of dialogue strategies based on the principles of schema theory will provide insight into how cognitive tasks intended to develop conceptual understanding might be supported. The agent will adapt its explanation according to the thinking characteristics of the primary schoolchildren, and this will result in guidelines that human tutors can employ when teaching reasoning skills to children in this age group.
Knowledge-based systems. Our new pedagogical agent will make a contribution to the field of knowledge-based systems by utilising schematic knowledge. The reasoning and decision-making processes of the pedagogical agent will be based on schematic domain knowledge and will draw on information contained in a model of the child's schematic knowledge. The pedagogical agent will include a computational model of natural language interaction between the agent and a child in a one-to-one interaction that is novel because it demonstrates an application of the generation of explanations through the use of reasoning based on schema theory.
1.3 The Structure of This Thesis
This thesis is organised into four parts:
Part 1: Motivation and methodology
Our motivation and methodology are presented in chapters 2 and 3. In Chapter 2 we set out our motivation for our research through a review of relevant existing work and thereby establish the theoretical basis of the research and identify the main issues requiring further investigation. In this chapter we consider the difficulties faced by children when learning new concepts and view the problem from cognitive, educational and computational perspectives. In Chapter 2 we also discuss in detail the potential benefits of activating prior knowledge and performing the modification of existing schemas when devising personalised explanations. We summarise the issues that have not yet been addressed in previous research and justify the need for our study.
Chapter 3 describes the experimental basis of the computational framework proposed in this thesis. We discuss the strategies used by experienced human tutors to diagnose students' knowledge, activate relevant prior knowledge and explain new concepts. In this chapter, we also discuss how the pedagogical agent will interact with the student. The findings of the experimental study presented in Chapter 3 identify the agent's speech acts, dialogue episodes and dialogue strategies.
Part 2: Computational Framework
The second part consists of chapters 4 and 5 that form the core of this thesis. In Chapter 4, we propose a novel design architecture: this is the computational framework we will use to design a pedagogical agent based on the results of the theoretical and experimental work discussed in chapters 2 and 3. We describe the functionality and operation of the main components of the agent and present in detail the knowledge representation and the student modeling mechanism.
Chapter 5 describes the support to be provided to the children by the agent and its dialogue mechanism. In this chapter we define how the adaptive explanation of new concepts will be generated and communicated to the student using a template-based natural language. Chapter 5 is an in-depth presentation of our computational framework for the personalised interactive explanation.
Part 3: Prototype and validation of the framework
The prototype phase of our research, which includes implementation and evaluation, is presented in chapters 6 and 7. In Chapter 6, we present the pedagogical agent integrated in an educational system to demonstrate how it works in educational settings. This chapter demonstrates an application of the ITS design approach proposed in this thesis; in it we demonstrate how to develop the pedagogical agent following the computational framework defined in chapters 4 and 5, and how to integrate the agent into existing multimedia educational systems. Chapter 7 presents an evaluation of the pedagogical agent carried out to study the agent's possible effectiveness and its usefulness in supporting children's conceptual understanding. The study provides a validation of the design principles adopted during the development of the agent.
Part 4: Conclusion and future extension
In Chapter 8, we present the conclusion of this thesis. This chapter describes our achievements, identifies the work's limitations and makes suggestions for future work that would lead to improvements to the computational framework presented in this thesis. The thesis provides a design architecture for the interactive explanation of new concepts that is open to further refinement, as pointed out in Chapter 8.