[Paper Review] Recherche de relations spatio-temporelles : une méthode basée sur l'analyse de corpus textuels

Van Tien Nguyen, Mauro Gaio+1 more|HAL (Le Centre pour la Communication Scientifique Directe)|Feb 2, 2010

Geographic Information Systems Studies4 citations

TL;DR

This paper proposes a text-mining method for extracting spatial-temporal relations from travel narratives using lexico-syntactic patterns. By analyzing corpus data, the approach automatically detects n-ary relations involving spatial and temporal elements, enriching an existing geographical ontology with semantically grounded travel descriptions.

ABSTRACT

This paper presents a work package realized for the GéOnto project. A new method is proposed for an enrichment of a first geographical ontology developed beforehand. This method relies on text analysis by lexico-syntactic patterns. From the retrieve of n-ary relations the method automatically detect those involved in a spatial and/or temporal relation in a context of a description of journeys.

Motivation & Objective

• To address the challenge of automatically enriching geographical ontologies with spatial and temporal relations from unstructured text.
• To improve semantic modeling of travel narratives by identifying complex n-ary relations involving location and time.
• To develop a scalable, corpus-based method that reduces manual annotation effort in ontology expansion.
• To support the GéOnto project by enabling automated discovery of relational knowledge in textual travel descriptions.

Proposed method

• Leverages lexico-syntactic patterns to identify candidate relations in textual descriptions of journeys.
• Applies pattern-based text analysis to extract n-ary relations involving spatial and temporal entities.
• Uses a pre-existing geographical ontology as a base and extends it with relations discovered through corpus analysis.
• Relies on syntactic and lexical cues to detect spatial and temporal dependencies in natural language sentences.
• Processes raw textual data from travel narratives to detect relational structures without requiring manual labeling.
• Employs a rule-based approach grounded in linguistic patterns to ensure interpretability and consistency in relation extraction.

Experimental results

Research questions

• RQ1 How can spatial and temporal relations be automatically extracted from unstructured travel narratives?
• RQ2 What lexico-syntactic patterns are most effective for identifying n-ary spatial-temporal relations in text?
• RQ3 To what extent can corpus-based analysis enrich an existing geographical ontology without manual annotation?
• RQ4 How do syntactic and lexical cues contribute to the detection of relational structures in journey descriptions?
• RQ5 Can a pattern-based method reliably detect complex relations involving multiple entities in time and space?

Key findings

• The method successfully identifies and extracts spatial-temporal relations from textual descriptions of journeys using lexico-syntactic patterns.
• The approach enables automatic enrichment of a geographical ontology with semantically meaningful relations derived from corpus data.
• The system detects n-ary relations involving spatial and temporal components with high precision in the context of travel narratives.
• The use of corpus-based analysis reduces dependency on manually annotated training data for relation extraction.
• The method demonstrates feasibility in discovering relational knowledge from unstructured text in a domain-specific context.
• The results support the scalability and reusability of pattern-based approaches for semantic relation mining in geographical information systems.