deepparse 0.2.3

A library for parsing multinational street addresses using deep learning.

Here is deepparse.

Deepparse is a state-of-the-art library for parsing multinational street addresses using deep learning.

Use deepparse to:

• Use the pre-trained models to parse multinational addresses.
• Retrain our pre-trained models on new data to parse multinational addresses.

Read the documentation at deepparse.org.

Deepparse is compatible with the latest version of PyTorch and Python >= 3.6.

Countries and Results

The following table presents the accuracy on the 20 countries we used during training for both our models.

Country	Fasttext (%)	BPEmb (%)	Country	Fasttext (%)	BPEmb (%)
Italy	99.66	99.73	United States	99.56	99.53
Germany	99.72	99.84	Austria	99.19	99.03
South Korea	99.96	100.00	Canada	99.76	99.80
Mexico	99.54	99.60	Australia	99.62	99.74
Finland	99.75	99.87	Netherlands	99.50	99.84
France	99.54	99.50	United Kingdom	99.54	99.62
Russia	98.71	99.49	Norway	99.40	98.71
Switzerland	99.48	99.61	Poland	99.64	99.83
Brazil	99.33	99.24	Denmark	99.65	99.84
Spain	99.70	99.79	Czechia	99.46	99.83

We have also made a zero-shot evaluation of our models using data from 41 other countries; the results are shown in the next table.

Country	Fasttext (%)	BPEmb (%)	Country	Fasttext (%)	BPEmb (%)
Philippines	81.56	83.73	South Africa	92.69	95.03
Colombia	85.92	87.50	Venezuela	95.36	89.67
Bermuda	91.30	93.66	Lithuania	89.21	76.60
Moldova	88.51	89.13	India	66.91	77.26
Malaysia	81.31	92.78	Bosnia	88.91	84.33
Belgium	89.57	86.41	Ukraine	91.80	92.73
Greece	83.42	39.82	Algeria	86.93	80.62
Slovakia	81.00	91.28	Bangladesh	74.49	79.29
Latvia	93.80	80.18	Reunion	96.48	93.40
Romania	93.23	91.83	Singapore	84.55	81.68
Indonesia	63.15	67.97	Cyprus	97.69	98.30
Portugal	93.39	93.20	Serbia	95.62	94.69
Croatia	96.63	86.24	Japan	44.33	35.77
New Caledonia	99.42	99.01	New Zealand	97.04	98.86
Uzbekistan	87.63	71.93	Faroe Islands	71.73	85.46
Hungary	47.00	24.05	Slovenia	96.27	97.28
Paraguay	97.00	97.15	Iceland	95.76	98.01
Estonia	90.61	76.45	Argentina	89.47	88.55
Bulgaria	92.70	95.87	Sweden	77.29	87.77
Belarus	88.77	93.00	Kazakhstan	87.24	91.23
Ireland	86.35	87.49

Getting started:

from deepparse.parser import AddressParser

address_parser = AddressParser(model_type="bpemb", device=0)

# you can parse one address
parsed_address = address_parser("350 rue des Lilas Ouest Québec Québec G1L 1B6")

# or multiple addresses
parsed_address = address_parser(
    ["350 rue des Lilas Ouest Québec Québec G1L 1B6", "350 rue des Lilas Ouest Québec Québec G1L 1B6"])

# you can also get the probability of the predicted tags
parsed_address = address_parser("350 rue des Lilas Ouest Québec Québec G1L 1B6", with_prob=True)

Retrain a model

see here for a complete example.

# We will retrain the fasttext version of our pretrained model.
address_parser = AddressParser(model_type="fasttext", device=0)

address_parser.retrain(training_container, 0.8, epochs=5, batch_size=8)

---

Installation

Before installing deepparse, you must have the latest version of PyTorch in your environment.

• Install the stable version of deepparse:

pip install deepparse

• Install the latest development version of deepparse:

pip install -U git+https://github.com/GRAAL-Research/deepparse.git@dev

---

Cite

Use the following for the article;

@misc{yassine2020leveraging,
    title={{Leveraging Subword Embeddings for Multinational Address Parsing}},
    author={Marouane Yassine and David Beauchemin and François Laviolette and Luc Lamontagne},
    year={2020},
    eprint={2006.16152},
    archivePrefix={arXiv}
}

and this one for the package;

@misc{deepparse,
    author = {Marouane Yassine and David Beauchemin},
    title  = {{Deepparse: A state-of-the-art deep learning multinational addresses parser}},
    year   = {2020},
    note   = {\url{https://deepparse.org}}
}

---

Contributing to Deepparse

We welcome user input, whether it is regarding bugs found in the library or feature propositions ! Make sure to have a look at our contributing guidelines for more details on this matter.

License

Deepparse is LGPLv3 licensed, as found in the LICENSE file.

---