An automatic extraction method of static and dynamic spatial contexts from texts - Archive ouverte HAL
Communication Dans Un Congrès Année : 2018

An automatic extraction method of static and dynamic spatial contexts from texts

Résumé

Spatial descriptions, with or without motion, are the main issues addressed by this paper. We describe construction grammars implemented in the PERDIDO platform with cascaded finite-state transducers which aims at marking and formalizing relations between extended named entities, geographical terms, spatial relations and motion verbs. These grammars can be seen as a computational synthesis of the work on the expression of space and motion in natural language. The proposed method for geographical information extraction has been tested for three different projects within the digital humanities using specific corpora. The first task deals with the extraction of place names from French novels, the second task deals with the extraction of motion events from hiking descriptions written in Romance languages (French, Spanish and Italian) and the third task aims at identifying fictive motion expressions in English alpine journals.
Fichier principal
Vignette du fichier
EGC_digital_humanities_moncla.pdf (773.62 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01695643 , version 1 (29-01-2018)

Identifiants

  • HAL Id : hal-01695643 , version 1

Citer

Ludovic Moncla, Mauro Gaio, Ekaterina Egorova, Christophe Claramunt. An automatic extraction method of static and dynamic spatial contexts from texts. Atelier Science des Données et Humanités Numériques (SDHN), Conférence Internationale Francophone sur l'Extraction et la Gestion de Connaissance (EGC 2018), Jan 2018, Paris, France. ⟨hal-01695643⟩
178 Consultations
93 Téléchargements

Partager

More