How to detect if a user query might be an address and requires to launch a map answer.
The project implements 3 classifiers, using an a la scikit template.
The first classifier is a simple scorer classifier, based on the parsing result of the address parser libpostal (https://github.com/openvenues/libpostal) According to how the parser manage to work, and which fields are parsed, we make a score and decide if an address or not.
The second classifier is based on the FastText classifier trained on address data. The fasttext makes an embedding of the differents address it sees and therefore when a new address is submitted if it’s in a close spaceto what have been learned. The classifier is pre-trained, and the Fasttext zip model is store within the package.
The third classifier is a voting classifier combining the results of the two previous classifiers.
you have the following prerequisites:
sudo apt-get install curl autoconf automake libtool pkg-config
Then to install the C library:
git clone https://github.com/openvenues/libpostal
cd libpostal
./bootstrap.sh
./configure --datadir=[...some dir with a few GB of space...]
make
sudo make install
# On Linux it's probably a good idea to run
sudo ldconfig
In order to build fastText, use the following:
$ git clone https://github.com/facebookresearch/fastText.git
$ cd fastText
$ make
The previous step must have been installed before.
Package is on the Pipy : pip install addr_detector
Or directly after forking this one, run pip install -r requirements.txt
from addr_detector.Fasttext_clf import Fasttext_clf
from addr_detector.Postal_clf import Postal_clfclf1 = Fasttext_clf()
clf2 = Postal_clf()addresse = ['7 rue Spontini Paris']
print(clf1.predict(addresse))
print(clf2.predict(addresse))[True]
[True]
addresses= ['7 rue Spontini Paris', '27 rue de l université Paris','Comment se faire cuire un oeuf','Google\n']
print(clf1.predict(addresses))
print(clf2.predict(addresses))[True, True, False, False]
[True, True, False, False]