MATE Deliverable D1.1

Supported Coding Schemes

(Linköping University)

Coding book:
Authors: Nils Dahlbäck and Arne Jönsson
Title: A coding manual for the Linköping dialogue model

Further information can be found in:

Number of annotators:
None at the moment. Previously, four people used the scheme.

The dialogues has been analyzed by linguists, a psychologist and computer scientsits. However, the main work on tagging the dialogues was done by two students, one cognitve science student and one computer science student.

Number of annotated corpora:
The corpus used when developing Linlin consists of 30 dialogues with 1749 utterances. The dialogue model was also applied to 100 SUNDIAL dialogues with around 700 utterances and to 10 Waxholm dialogues with around 400 utterances. (all Swedish)

Evaluations of scheme:
No k -statistic is available  for the LinLin scheme, but in a pairwise agreement for LinLin1 97% was achieved.

Underlying task:
LinLin was designed to a written human-to-(simulated)computer information retrieval dialogue. It is now also applied to the AIRPLANE corpus with a human-to-human instructional dialogue.

List of phenomena annotated:


S: [Welcome to Cardata] DO
U: [show mercedes] Q
S: [Wait...] DC 

[Cardata can answer questions about a number of car models, concerning manufacturer, model, year, country of manufacturing, disposition to rust and size class as well as question about price, security, space and technical data.] A

[Any particular wishes?] Q

U: [cost and space] A

Mark-up language:
Nb's mark-up language (pseudo-SGML  - not fully compliant)

Existence of annotation tools:
Nb (NotaBene) for coding, perlscript for analysis.

Used in the S-DIME (Swedish dialogue move engine) project.

Contact person:
Arne Jönsson (arnjo@ida.liu.se)
Dept. of Computer and Information Science
Linköping University

Last Modification: 27.8.1998 by Marion Klein