OLM information

The following information is provided as background information about Open Learner Models. It is reproduced from the paper ďBringing Chatbots into Education: Towards Natural Language Negotiation of Open Learner ModelsĒ (Kerly et al. 2006).

You can also see a selection of screenshots from Open Learner Models here.

    Learner Modelling and Open Learner Modelling

    Intelligent Tutoring Systems employ a learner model to infer the learnerís knowledge and to provide an adaptive interaction. While many ITSs do not reveal the contents of the learner model to the learner, it has been argued that opening the learner model to the ITS users can in fact provide opportunities for learner reflection and deep learning that enhances the learning experience (e.g. [20], [21], [22], [23] and [24]).

    Open learner models are therefore accessible to the user. They are inferred from the learnerís interaction with the system (as in any ITS), and may also include contributions obtained directly (explicitly) from the student. As a pedagogical goal, learner reflection is endorsed by many theories, including Dewey [25], Schn [26], and Kolb [27].

    Bull & Pain [28] and Dimitrova [21] proposed that both learner reflection and model accuracy could be increased through a process of negotiation of the learner model contents and implemented the Mr. Collins and STyLE-OLM systems respectively. In this method the learner model is collaboratively constructed and maintained by both the system and the learner. In both the above systems, the learner was required to discuss their beliefs with the system, arguing against the systemís assessment if they disagreed with it, and providing supporting evidence or argument for their own beliefs when they differed from the system. This interaction supported the increased learner reflection intended to benefit learning, and produced a more accurate learner model on which to base system adaptivity.

    In order to support the negotiation functionality, the learner model must store distinct records of the learnerís and the systemís beliefs about the learnerís knowledge. Two separate belief measures were maintained in the Mr. Collins [28] system, each of which was taken into account by the system in providing adaptive interactions. Bakerís notion of interaction symmetry [29] was applied to the system, ensuring that all dialogue moves necessary for negotiation were available to both the student and the system. Laboratory studies of the Mr. Collins system [28] found that students were interested in being able to see the contents of their learner model. They were keen to use negotiation to improve the accuracy of the learner model and most students also wanted the system to challenge them if it disagreed with theirattempts to change their confidence in their performance.

    Previous open learner model systems have employed menu-selection or conceptual graphs to achieve negotiation of the learner model contents. While laboratory trials ([28], [21]) of these systems suggested the potential for engaging learner reflection and enhancing the accuracy of the learner model, the negotiation methods used may be restrictive or unnatural. We propose that natural language negotiation through a Chatbot may offer users the flexibility to express their views in a naturalistic and intuitive way.