Current and past projects

Using a chatbot conversation to enhance the learning of social theory

Andrew Teal, School Programmer, Human and Health Sciences

Rudy van Kemenade, Senior Lecturer in Sociology, Behavioural Sciences,

Chris Cameron, Lecturer in Sociology, Behavioural Sciences,

Graham R Gibbs, Head of Department, Behavioural Sciences.

In collaboration with

Dr. Dianne Phillips, SITU Director and Lecturer in Sociology, Manchester Metropolitan University.

Aims

• To develop four knowledge bases for key social and/or political theorists for use with an open source chatbot via the WWW.
• To test them in a teaching context using the chatbot in BlackBoard.
• To disseminate experience in production and teaching
• To make available all materials produced to the UK academic community.

Top of page

Background

Chatbots are computer programs that have the ability to parse natural language questions and, by referring to a knowledge base, generate natural language answers. They are often encountered as interfaces to help systems and web-based search engines (eg. Word help system and Ask Jeeves web search engine). AliceBot is a prize-winning artificial intelligence system that supports conversations over the WWW (http://www.alicebot.org/). AliceBot can be modified to take on the personality and knowledge base of named individuals. This has already been done with e.g. Elvis Presley and John Lennon. As some authors have noted, there is an educational potential here if further personalities such as Napoleon or Ghandi are created. If a sufficiently high-quality knowledge base is constructed, then there is scope in higher education for using such a chatbot as a substitute for the expert academic.

For example, many students of sociology experience problems in learning social theory and philosophy. They .

• . find it very hard to get started with reading.
• . need cartoon level introduction, leading on to more complex material.
• . find it hard to know what is relevant discussion, so seminar discussion often wanders off topic.
• . need the information provided to reflect their stage of learning.
• . want to talk with experts when they need help. This is not usually in seminars/tutorials but when essays are being written, which increases the burden on staff via remedial work.

In this context, an accessible chatbot with a knowledge base reflecting key areas of social theory or philosophy could make an important educational contribution.

Many universities are moving rapidly to the use of Virtual Learning Environments (VLEs) or Managed Learning Environments (MLEs) such as BlackBoard to support student learning. Whilst it is fairly easy in these systems to make course documents and lecture and seminar notes available to students, there is still much work to be done, both developmentally and in staff training, to create more innovative ways of using these systems in teaching. This is especially so in sociology, where it seems that the only additional use of MLEs is for online discussion. Many academics, rightly, are concerned that this will take up even more of their time in supporting students. One important response to this is to encourage students to converse with and learn from each other. This has been tried over the last few years by systems such as coMentor. Whilst this frees up staff time, one disadvantage is that the quality of discussion and hence of the information that others might vicariously experience may be questionable. The use of a chatbot in such a context, whilst it cannot create the same kind of quality of conversation as human participants, can, at least, ensure high accuracy of the content of the conversation.

Top of page

Proposed work

This project proposes to develop a small number of knowledge bases for use in social theory. These will consist of personality and knowledge linked with key social theorists, such as Marx, Weber, Durkheim, Giddens or Foucault. The knowledge bases will be designed to be sensitive to the knowledge levels of potential student users and to be open to "tell me more about." type questions. The knowledge base will indicate where views or ideas are still debated and where views may contradict or conflict with those of other theorists. Either spontaneously or following an enquiry, the knowledge base will give pointers to further reading.

Alongside the creation of high quality knowledge bases, the project will undertake some minor software developments. These will involve changing the chatbot user interface so that users can opt to record the content of the conversation and be directed to additional information.

The system will be implemented using AliceBot (www.alicebot.org). This uses Artificial Intelligence Markup Language (AIML), a derivative of XML (Extended Markup Language) to express the knowledge base upon which the chatbot's parsing of questions and construction of responses are based. There are several advantages in using AliceBot.

1. The team at Huddersfield now have considerable expertise in the creation of XML files through involvement in the Co3 JCIEL-funded project. This is investigating the use of IMS (Instructional Management Systems) standards to facilitate interoperability between three MLEs. IMS is a developing set of international standards for educational content and materials, which are expressed in XML. However, the current project will also require the work of experienced and expert academics to produce high quality knowledge bases. These will then be converted into the appropriate AIML format.

2. AliceBot is open source software, a great advantage in an educational setting since the software can be freely used and, if necessary, modified. There are versions that run on all major servers. It is a requirement of the open source licence that any software modifications should themselves be open source. It is proposed that the AIML knowledge bases created should be open source and hence freely available to all.

3. AliceBot already has a Web interface, so it can easily be incorporated into existing MLEs. This can be easily modified so that additional features can be incorporated.

4. AliceBot is not a rule-based system. There is no need to go through a very complicated elicitation process with experts to produce the knowledge base. The focus can therefore be much more on the quality and accuracy of the information and the way it is presented to student users. Moreover, the knowledge base can be added to and amended by later developers.

Top of page

Evaluation

These developments will be tested out in two ways.

Stage 1. Staff and selected students at Huddersfield and Manchester will test out the knowledge base by using the chatbot. They will check it for accuracy, interpretation and relevance to the kinds and level of questions being asked.

Stage 2. This will evaluate the use of the chatbot in a teaching context at Huddersfield (using BlackBoard) and at Manchester. There are two ways in which the material can be used.

• As a general advisor/expert. Students will have a research question or essay title they need to research and will use the chatbot to generate content that can be included in their assessed work. Conversation with the chatbot will be recordable and can then be cut and pasted into an essay. Students will be required to edit this information into a coherent essay, just as they would with information gleaned from texts.

• As a substitute lecturer. The role here will be remedial or supplementary to information provided by teachers. The knowledge base will be customised to respond to instructions such as "tell me more about."

This should produce a better understanding both of the form and type of content that best matches student needs and also information about the best ways in which the chatbots can be used educationally.

Top of page

Deliverables

• Four key theorist knowledge bases.

• A functioning web-based chatbot modified to provide extra services and linked information website

• A report on the teaching evaluation with recommendations for teaching use.

• A workshop for academics interested in using the materials and the system.

Plan of work

The project will run for 11 months. It will start in August 2001. During the first semester 2001-2 staff will develop and check the knowledge bases. In Dec 2001 the research assistant will be appointed. During the second semester January-May 2002 the material will be evaluated in a teaching context and the web site will be established. In June 2002 final report will be produced and the workshop will be offered.