This is a Python implementation of the method of learning embeddings of words and Wikipedia entities proposed in this paper. The embeddings can be directly learned using a Wikipedia dump retrieved from Wikimedia Downloads.
% pip install Cython numpy
% pip install -r requirements.txt
% python setup.py develop
The pre-trained embedding can be downloaded from the following links. Please note that these files must be placed in the same directory.
Alternatively, the embedding can be built by using the following commands:
% entity-vector build_dictionary <WIKIPEDIA_DUMP_FILE> <DICTIONARY_FILE>
% entity-vector train_embedding <WIKIPEDIA_DUMP_FILE> <DICTIONARY_FILE> <OUT_FILE>
>>> from entity_vector import EntityVector
>>> entvec = EntityVector.load('enwiki_entity_vector_500_20151026.pickle')
>>> word = entvec.get_word(u'c-3po')
>>> entvec[word]
memmap([ 0.05961042, 0.24534572, 0.42090839, -0.01455959, 0.11772038,
0.55437287, -0.62508648, -0.24478671, 0.07838536, 0.27331885,
0.35184374, 0.34113087, 0.11718472, -0.14086614, -0.00730115,
...
>>> entvec.most_similar(word)
[(<Word c-3po>, 1.0000000000000002),
(<Entity C-3PO>, 0.8855517572211461),
(<Word r2-d2>, 0.85768096183067088),
(<Entity R2-D2>, 0.81842535257607718),
(<Word chewbacca>, 0.7771232783769505),
(<Entity Chewbacca>, 0.77412692204846856),
...
>>> entity = entvec.get_entity(u'C-3PO')
>>> entvec[entity]
memmap([ -3.51071961e-03, 4.82281654e-01, 6.72443198e-01,
2.41103170e-01, 1.43198542e-01, 6.44051048e-01,
-5.48925964e-01, -4.64934616e-01, -2.48444133e-01,
...
>>> entvec.most_similar(entity)
[(<Entity C-3PO>, 1.0),
(<Entity R2-D2>, 0.90188752966007535),
(<Word c-3po>, 0.88555175722114643),
(<Entity Chewbacca>, 0.8304708994223623),
(<Word r2-d2>, 0.82777910810169675),
(<Entity Han Solo>, 0.80912814689071744),
...
>>> entvec.get_similarity(word, entity)
0.90466782126690559If you use the code or the pretrained embedding in your research, please cite the following paper:
@InProceedings{yamada-EtAl:2016:CoNLL,
author = {Yamada, Ikuya and Shindo, Hiroyuki and Takeda, Hideaki and Takefuji, Yoshiyasu},
title = {Joint Learning of the Embedding of Words and Entities for Named Entity Disambiguation},
booktitle = {Proceedings of The 20th SIGNLL Conference on Computational Natural Language Learning},
month = {August},
year = {2016},
address = {Berlin, Germany},
publisher = {Association for Computational Linguistics},
pages = {250--259}
}