Utterance Retrieval based on Recurrent Surface Text Patterns - Archive ouverte HAL Access content directly
Conference Papers Year : 2017

Utterance Retrieval based on Recurrent Surface Text Patterns


This paper investigates the use of recurrent surface text patterns to represent and index open-domain dialogue utterances for a retrieval system that can be embedded in a conversational agent. This approach involves both the building of a database of such patterns by mining a corpus of written dialogic interactions, and the exploitation of this database in a generalised vector space model for utterance retrieval. It is a corpus-based, unsupervised, parameterless and language-independent process. Our study indicates that the proposed model performs objectively well comparatively to other retrieval models on a task of selection of dialogue examples derived from a large corpus of written dialogues.
Fichier principal
Vignette du fichier
ECIR.pdf (320.72 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-01436052 , version 1 (16-01-2017)


  • HAL Id : hal-01436052 , version 1


Guillaume Dubuisson Duplessis, Franck Charras, Vincent Letard, Anne-Laure Ligozat, Sophie Rosset. Utterance Retrieval based on Recurrent Surface Text Patterns. 39th European Conference on Information Retrieval, Apr 2017, Aberdeen, United Kingdom. ⟨hal-01436052⟩
227 View
501 Download


Gmail Facebook Twitter LinkedIn More