The architecture of NLGbAse allow to build a metadata representation for each
concept represented in Wikipedia. We consider all the
characteristics of the Wikipedia internal structure to extract the metadata. For each encyclopedic
article, a corresponding metadata is extracted. It consists of a
semantic graph of all writing forms, a set of weighted words extracted
from the encyclopedic article description and a taxonomic class label
(such as PERson, ORGanization, LOCation). Finally, each metadata
is associated to an entry point in the LinkedData Network (DBPedia,
CIA-Factbook rdf set, Wikicompanies, etc ...).
This allow you to use this search engine when you look for a specific representation of a concept on the LinkedData Network to point directly on its LinkedData entrypoint.
Each linguistic edition of metadata is extracted originally from the corresponding edition of Wikipedia. Please see Publications section for more detailed view of extraction process.
Simple search : this form explores the metadata to find best fit to a query. It uses terminology graphes and cosine similarity.
Semantic search, basic. It gives you the possibility to retrieve an entity (person, organisation ...) from terms in a query.
This allow you to browse the metadata content with a semantic perspective. Samples :
This allow you to use this search engine when you look for a specific representation of a concept on the LinkedData Network to point directly on its LinkedData entrypoint.
Each linguistic edition of metadata is extracted originally from the corresponding edition of Wikipedia. Please see Publications section for more detailed view of extraction process.
Simple search : this form explores the metadata to find best fit to a query. It uses terminology graphes and cosine similarity.
Semantic search, basic. It gives you the possibility to retrieve an entity (person, organisation ...) from terms in a query.
This allow you to browse the metadata content with a semantic perspective. Samples :
- I search all the car built by the Renault Company : i select prod or prod.vehicule, and type Renault : try it.
- I want a list of all the cities of Texas : i select loc.admi and type Texas: try it.
