Mark Buckley

A Classification of Dialogue Actions in Tutorial Dialogue

In this paper we present a taxonomy of dialogue moves which describe the actions that students and tutors perform in tutorial dialogue. We are motivated by the need for a categorisation of such actions in order to develop computational models for tutorial dialogue. As such, we build both on existing work on dialogue move categorisation for tutorial dialogue as well as dialogue taxonomies for general dialogue. Our taxonomy has been prepared by analysing a corpus of tutorial dialogues on mathematical theorem proving. We also detail an annotation experiment in which we apply the taxonomy and discuss idiosyncrasies in the data which influence the decisions in the dialogue move classification.

Linguistic Processing in a Mathematics Tutoring System: Cooperative Input Interpretation and Dialogue Modelling

Formal domains, such as mathematics, require exact language to communicate the intended content. Special symbolic notations are used to express the semantics precisely, compactly, and unambiguously. Mathematical textbooks offer plenty of examples of concise, accurate presentations. This effective communication is enabled by interleaved use of formulas and natural language. Since natural language interaction has been shown to be an important factor in the efficiency of human tutoring [29], it would be desirable to enhance interaction with Intelligent Tutoring Systems for mathematics by allowing elements of mixed language combining the exactness of formal expressions with natural language flexibility.

An agent-based architecture for dialogue systems

Research in dialogue systems has been moving towards reusable and adaptable architectures for managing dialogue execution and integrating heterogeneous subsystems. In this paper we present a formalisation of ADMP, an agent-based architecture which supports the development of dialogue applications. It features a central data structure shared between software agents, it allows the integration of external systems, and it includes a meta-level in which heuristic control can be embedded.

Verification of human-level proof steps in mathematics education

Automated mathematics tutorial systems need support from a reasoning module which can verify the correctness of students’ contributions. However, current systems typically do not reason at a level similar to the student’s reasoning level, and do not fully account for underspecified or ambiguous inputs. We present a domain-independent method for automatically verifying correct proof steps and detecting standard reasoning errors. We use a depth limited BFS proof search to determine and maintain multiple possible interpretations consistent with the given proof step, we are able to resolve or otherwise propagate underspecification and ambiguity which occurs due to unrestricted user input. Our approach has been implemented in Ωmega.