WP1 - Project
WP1 - Project
WP1: Models and tools allowing to specify and to build search engines dedicated to geographic information (spatial, temporal et thematic facets).
WP1 provides other work packages with data (ontologies and domain models) and metadata in order to help indexing, searching and therefore to the further dissemination of these data as such or as parts of others.
It also provides tools for using these data in order to index document collections, and search in the indexed collections.
Objectives and innovative aspects
Regarding indexing and information retrieval, we propose to combine two kinds of techniques:
- indexing and « full-text » IR based on statistical processing of terms (extracted from the Rameau thesaurus),
- Semantic based indexing and search using domain ontology.
We want to integrate and analyze a specific domain ontology dedicated to botany from the research of concepts in the document collection based on the thesaurus Rameau.
In the context of recent work, the LIUPPA team proposes an experimental platform for building geographical indexes following an original method for information extraction from text and for enriching resources like Geographic databases.
We propose to extend this platform and integrate new validation/representation procedures of Geographic Named Entities based on ontological resources that will implement methods and tools to support indexing for expert librarian users.
The objectives are to establish the following results:
- Methods and tools for extracting concepts and relations in textual documents, based on the definition of patterns, on the identification of Geographic Named Entities and on the exploitation of text structure and argumentative relationships in text,
- We will develop sophisticated environments for browsing and search by human users, as well as services for interoperability with external information systems.
- In this project, the application framework is provided by CBN. The aim is to help this structure to disseminate their resources as much as possible.