PERDIDO Geoparser python library


Keywords
geoparsing, named-entity-recognition, geographic-information-retrieval, toponym-resolution, toponym-disambiguation
License
MulanPSL-2.0
Install
pip install perdido==0.1.50

Documentation

Perdido Geoparser Python library

PyPI PyPI - License PyPI - Python Version

Installation

To install the latest stable version, you can use:

pip install --upgrade perdido

Quick start

Geoparsing

Binder Open In Colab

Import

from perdido.geoparser import Geoparser

Run geoparser

text = "J'ai rendez-vous proche de la place Bellecour, de la place des CĂ©lestins, au sud de la fontaine des Jacobins et prĂšs du pont Bonaparte."
geoparser = Geoparser()
doc = geoparser(text)

Some parameters can be set when initializing the Geoparser object:

  • version: Standard (default), Encyclopedie
  • pos_tagger: spacy (default), stanza, and treetagger

Get tokens

  • Access token attributes (text, lemma and UPOS part-of-speech tag):
for token in doc:
    print(f'{token.text}\tlemma: {token.lemma}\tpos: {token.pos}')
  • Get the IOB format:
for token in doc:
    print(token.iob_format())
  • Get a TSV-IOB format:
for token in doc:
    print(token.tsv_format())

Print the XML-TEI output

print(doc.tei)

Print the XML-TEI output with XML syntax highlighting

from display_xml import XML
XML(doc.tei, style='lovelace')

Print the GeoJSON output

print(doc.geojson)

Get the list of named entities

for entity in doc.named_entities:
    print(f'entity: {entity.text}\ttag: {entity.tag}')
    if entity.tag == 'place':
        for t in entity.toponym_candidates:
            print(f' latitude: {t.lat}\tlongitude: {t.lng}\tsource {t.source}')

Get the list of nested named entities

for nested_entity in doc.nested_named_entities:
    print(f'entity: {nested_entity.text}\ttag: {nested_entity.tag}')
    if nested_entity.tag == 'place':
        for t in nested_entity.toponym_candidates:
            print(f' latitude: {t.lat}\tlongitude: {t.lng}\tsource {t.source}')

Get the list of spatial relations

for sp_relation in doc.sp_relations:
    print(f'spatial relation: {sp_relation.text}\ttag: {sp_relation.tag}')

Shows named entities and nested named entities using the displacy library from spaCy

displacy.render(doc.to_spacy_doc(), style="ent", jupyter=True)
displacy.render(doc.to_spacy_doc(), style="span", jupyter=True)

Display the map (using folium library)

doc.get_folium_map()

Saving results

doc.to_xml('filename.xml')
doc.to_geojson('filename.geojson')
doc.to_iob('filename.tsv')
doc.to_csv('filename.csv')

Geocoding

Binder Open In Colab

Import

from perdido.geocoder import Geocoder

Geocode a single place name

geocoder = Geocoder()
doc = geocoder('Lyon')

Geocode a list of place names

geocoder = Geocoder()
doc = geocoder(['Lyon', 'la place des CĂ©lestins', 'la fontaine des Jacobins'])

Get the geojson result

print(doc.geojson)

Get the list of toponym candidates

for t in doc.toponyms: 
    print(f'lat: {t.lat}\tlng: {t.lng}\tsource {t.source}\tsourceName {t.source_name}')

Get the toponym candidates as a GeoDataframe

print(doc.to_geodataframe())

Perdido Geoparser REST APIs

http://choucas.univ-pau.fr/docs#

Example: call REST API in Python

import requests

url = 'http://choucas.univ-pau.fr/PERDIDO/api/'
service = 'geoparsing'
data = {'content': 'Je visite la ville de Lyon, Annecy et le Mont-Blanc.'}
parameters = {'api_key': 'demo'}

r = requests.post(url+service, params=parameters, json=data)

print(r.text)

Tutorials

Cite this work

Moncla, L. and Gaio, M. (2023). Perdido: Python library for geoparsing and geocoding French texts. In proceedings of the First International Workshop on Geographic Information Extraction from Texts (GeoExT'23), ECIR Conference, Dublin, Ireland.

Acknowledgements

Perdido is an active project still under developpement.

This work was partially supported by the following projects: