Searching and retrieving multi-levels annotated data - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2015

Searching and retrieving multi-levels annotated data

Brigitte Bigi
Jorane Saubesty

Résumé

The annotation of recordings is related to many Linguistics subfields as Phonetics, Prosody, Gestures or Discourse... Corpora are annotated with detailed information at various linguistic levels thanks to annotation software. As large multimodal corpora become prevalent, new annotation and analysis requirements are emerging. This paper addresses the problem of exploring annotations in order to extract multimodal data in the linguistic field ranging from general linguistic to domain specific information. The answer choose to fulfill this purpose is a useroriented approach: the data can be extracted without any specific knowledge or skill. The paper exposes two ways to filter the annotations by a predicative approach: 1/ single filters, i.e. search in one tier depending of the data content, by the extraction of the time values and the duration; 2/ relation filters, i.e. search on annotations of a tier in time-relation with annotations of another one. This system is distributed in SPPAS software under the terms of a public license.
Fichier principal
Vignette du fichier
bigi2015gespin.pdf (229.68 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01455315 , version 1 (22-02-2018)

Identifiants

  • HAL Id : hal-01455315 , version 1

Citer

Brigitte Bigi, Jorane Saubesty. Searching and retrieving multi-levels annotated data. Proceedings of Gesture and Speech in Interactioni - 4th edition, 2015, Nantes, France. pp.31-36. ⟨hal-01455315⟩
104 Consultations
53 Téléchargements

Partager

Gmail Facebook X LinkedIn More