RNTI

MODULAD
Perdido : librairie Python pour le geoparsing et le geocoding de textes en français
In EGC 2023, vol. RNTI-E-39, pp.573-580
Abstract
This article presents the Perdido Python library for geoparsing and geocoding of French texts. We present the general architecture of the Perdido tool composed of three layers: back-office, API and Python library. We detail the methods used for the development of the processing chain and the different tasks (named entity recognition and classification and toponym resolution). Finally, we present the different functionalities of the Python library and how to use it. The library is developed as an overlay using the API services and allows to manipulate, visualize and export geoparsing and geocoding results. A Jupyter notebook describes, in the form of a tutorial, all the features implemented in the library.