6. INFORMATION & COMMUNICATION TECHNOLOGIES
» ICT applications
» Intelligence systems
» Artificial intelligence/robotics
» Software, Data Processing
GEOLSemantics is an SME specializing in automatic semantic extraction from texts (or transcribed
speech) in different languages (French, English, Arabic (standard and dialects from the Maghreb and
Egypt) and Chinese (Mandarin).
The technology is based on a general purpose deep morphosyntactic analysis recognizing and
normalizing words (simple or compound), recognizing named entities, building dependency relations,
assimilating passive and active forms, treating negation, finding referent of pronouns, detecting
tenses and modalities of actions.
The knowledge extraction is application-dependent. The application is described in an ontology.
Extractions rules are written to extract the knowledge according to the ontology. GEOLSemantics has
developed an ontology and extraction rules for the security.
GEOLSemantics has been the leader of the SAIMSI project (Suivi Adaptatif Interlingue et MultiSources
des Informations), the main French-funded project focused on the extraction of information about
the activities of persons suspected of illegal acts, especially jihadist terrorism. GEOLSemantics is
currently the leader of the ORELO project aiming at detecting the dialectal origin of writers or
speakers. In the ORELO project, GEOLSemantics has developed dictionaries of dialects that extend
the robustness of its treatment of Arabic containing dialectal elements. The project also required
GEOLSemantics to expand its capabilities to include processing Arabic (standard or dialect) written
using latin characters, a format commonly found on the web.