Colab Word2Vec Using Emoji2Vec dataset

 .

Import the required libraries:

import gensim

from gensim.models import word2vec

from gensim.models import KeyedVectors

from sklearn.metrics.pairwise import cosine_similarity

.

Upload Word2Vec emoji2vec dataset into /content folder:

.

Alternatively, get from Archive.org:

!wget 'https://archive.org/download/word-embeddings/emoji2vec.bin' -P '/content'

.

Load Emoji2Vec model

import gensim.models as gsm

e2v = gensim.models.KeyedVectors.load_word2vec_format('/content/emoji2vec.bin', binary=True)

happy_vector = e2v['😂']    # Produces an embedding vector of length 300

print(happy_vector.shape)

>>>>(300,)

.

Find cosine similarity:

v_king = e2v["🤴"] 

v_queen = e2v["👸"]

print(v_king.shape)

print(v_queen.shape)

cosine_similarity([v_king],[v_queen])

>>>>(300,)

(300,)

array([[0.48802766]], dtype=float32)

.

Reference:

https://github.com/uclnlp/emoji2vec