Tools

Union College Academic Register Search (v0.1) (On Campus Only)

A search tool, powered by the Natural Language Toolkit, which matches statements of interest to descriptions and course titles from the Union College Academic Register.


CuDAC: Cue-Based Dialogue Act Classifier

A Java implementation of the cue-based dialogue act classifier. Works from the command line. If true dialogue act is included, package can also calculate accuracy scores on the fly. For details of the classifier, see:

Webb, N., M. Hepple and Y. Wilks

Dialogue Act Classification using Intra-Utterance Features, in Proceedings of the AAAI Workshop on Spoken Language Understanding, Pittsburgh, 2005.


AT-AT: Albany Trial-Annotation Tool

The AT-AT is a Java tool that enables the user to add a range of XML annotations over appropriately coded and segmented dialogue data. AT-AT currently has two distinct functions: (1) Dialogue Act Annotation, and (2) Appropriateness Annotation.

(1) Dialogue Act Annotation Tool. There are two modes of operation of the DAT. The first is a manual annotation tool, that enables users to select one distinct dialogue act per utterance in a dialogue. The dialogue act set is configurable by the user (in XML format), where the interface of the tool (including the set of annotations) is dynamically created at start-up. There is also a (beta) semi-automatic annotation mode, where a subset of annotations is performed automatically using CuDAC – a Cue-based Dialogue Act Classifier.

(2) Appropriateness Annotation Tool. This is a tool that allows users to select appropriateness tags for each utterance, as per the annotation scheme described in the guidelines, above. Once annotation is complete, the tool saves the annotations in XML format, scores the complete dialogue, and outputs a distribution analysis of tags for that dialogue.