Abstract
Online and distance learning environments have changed dramatically over the last 20 years and are now sophisticated interactive learning environments. However, much more improvement is possible, and some of that improvement might come from mining some of the technologies developed as part of intelligent tutoring systems. Intelligent tutoring systems combine the best of human tutoring by capturing one on one tutoring interactions between a teacher and student on all topics for a learning module and converting them to a computerized version. The computerized version is designed to gauge the understanding of the student and adapt the instruction, modeling, hints, interactions, and activities to particular students. The systems are usually designed to assess the student’s learning continuously and scaffold the learning of the student. Ideally, these interactions will mimic human tutoring that has been shown to significantly improve learning beyond large group instruction.
TopBackground
The first intelligent tutoring systems were developed in the 1990s, building on research combining the disciplines of user modeling and effective tutoring strategies (Corbett, Koedinger, Anderson, 1997). These early intelligent tutoring environments “adopted the human tutor as their educational model and sought to apply artificial intelligence techniques to realize this model in “intelligent” computer- based instruction.” (Corbett et al., 1997). Human tutors have been shown to be most effective and efficient in improving learning (Cohen, Kulik, and Kulik, 1982) and that fact remains true even today (Corbett et al., 1997, Meyer and Wijekumar, 2007). Thus mimicking human tutors using computers is a worthwhile approach to extend to distance learning environments (Meyer, Middlemiss, Theodorou, Brezinski, McDougall & Bartlett 2002). Currently, intelligent tutoring systems include the research on the learning cycle of interactivity, assessment, and feedback to help student achieve their learning goals using computer tools (Meyer and Wijekumar, 2007).
The definition of intelligent tutoring technologies combines the processes of tutoring and intelligence. First, tutoring requires creating computer systems that can imitate what human tutors accomplish with learners. Specifically, the human tutors model how to solve the problem or apply their learning. The human tutors provide activities for the learner to use the skills, observe the student’s performance on the practice tasks, scaffold and guide the learner using hints and prompts, assess student learning, and provide feedback. Therefore, the tutoring component of intelligent tutoring technologies should attempt to mimic the ideal human tutors. Second, the intelligence in the computer systems are designed to “learn” from their interactions. Just as human tutors learn how to adapt to different types of learners over a period of time, it is important for the computer to save the tutor’s interactions with learners and use that information to create new interaction pathways for computer-learner interactions.
Key Terms in this Chapter
Direct and Indirect Speech: are communications terms that differentiate a statement that describes exactly what is intended by the speaker vs. providing related statements with the understanding that other humans can infer the intentions of the speaker. For example, if a person walks into a room and states “it is hot in here”, in a real life situation this could mean many different possibilities. First, it could mean that the actual temperature in the room is hot. Second, it could mean that the speaker was talking about the mood of the people in the room. Finally, it could also mean that the speaker would like to change the setting on the thermostat.
Intelligent Tutoring Technologies: describe a wide variety of computer based tutoring systems that use many different pathways through tutoring interactions between the computer and students. Some of the tools use programmed rules for interacting with the students and others learn from each interaction by saving each activity and new responses provided by the students. The tutoring systems follow a script that initiates interactions with learners, models expert behaviors to the students, requests interaction from the learner, and scaffolds their learning. The scaffolding of learning includes customized feedback, hints, and prompts.
Cognitive Task Analysis: is the process of conducting problem solving scenarios using experts and novices to identify the ideal problem solving path and steps as well as the places in the process where novices have difficulties. Frequently, think-aloud problem solving is conducted by the researcher to document what types of thinking and cognitive processes are involved in the process. These step by step processes designed to document what should be learned is used as a tutoring script in an intelligent tutoring environment.
Intelligent Tutoring System for the Structure Strategy (ITSS©): was designed to improve reading comprehension with any person by helping them learn and apply the Structure Strategy. The Structure Strategy was designed by Dr. Bonnie J.F. Meyer in 1975 and has been researched extensively in all age groups showing significant improvement in reading comprehension. ITSS created user models by capturing tutoring interactions between human tutors and students and converted them to computerized interactions. ITSS was funded by the US Department of Education. Further information about the project can be found at http://itss.br.psu.edu/
User Modeling: is the process of conducting human tutoring sessions between a tutor and learner. During these sessions, each concept is presented to the learner by the tutor and they engage in a tutoring interaction where the tutor gauges the learner’s understanding, learning style, and needs and adapts their tutoring to the needs of the learner. These interactions are video taped, transcribed, and coded. Once these interactions are reviewed, the information is used to program a computerized tutor to behave in a similar manner.
“Intelligent” Computer-Based Instruction: is designed to create computerized learning environments that behave similar to human instructors. The processes of human intelligence that are paralleled in these environments include modeling of expert problem solving behavior, providing interactive learning activities, assessing student responses in real-time, and providing appropriate feedback and scaffolding to the learner.
SQLServer™: is a commercially available database used in enterprise data management.
Cognitive Tutor Development: is a process described in detail by Anderson et al., 1995 where the team documented excellent tutoring interactions in detail to enable the creation of a computerized tutor.
Latent Semantic Indexing: is a text-to-text matching technique that provides a matrix version of large corpora text that can be compared to other texts for similarities. The approach is used in assessing essays and in natural language processing of student writing. There are commercial tools available for the product including LSI by Telecordia (2008)