visualize.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

# script to visualize word embeddings of given model with PCA dimensionality reduction
# creates image with matplotlib
# 
# @author: Andreas Mueller
# @see: Bachelor Thesis 'Analyse von Wort-Vektoren deutscher Textkorpora'
#
# Contributors:
#  Michael Egger <michael.egger@tsn.at>
#
# @example: python visualize.py
import argparse

import gensim
import matplotlib.pyplot as plt
from sklearn.decomposition import PCA
from sklearn.manifold import TSNE

parser = argparse.ArgumentParser(description='Script for visualizing word vector models')
parser.add_argument('model', type=str, help='source file with trained model')
args = parser.parse_args()

# configuration
currency = [
    'Schweiz', 'Franken', 'Deutschland', 'Euro', 'Grossbritannien', 'britische_Pfund',
    'Japan', 'Yen', 'Russland', 'Rubel', 'USA', 'US-Dollar', 'Kroatien', 'Kuna'
]
capital = [
    'Athen', 'Griechenland', 'Berlin', 'Deutschland', 'Ankara', 'Tuerkei', 'Bern', 'Schweiz', 'Hanoi', 'Vietnam',
    'Lissabon', 'Portugal', 'Moskau', 'Russland', 'Stockholm', 'Schweden', 'Tokio', 'Japan', 'Washington', 'USA'
]
language = [
    'Deutschland', 'Deutsch', 'USA', 'Englisch', 'Frankreich', 'Franzoesisch', 'Griechenland', 'Griechisch',
    'Norwegen', 'Norwegisch', 'Schweden', 'Schwedisch', 'Polen', 'Polnisch', 'Ungarn', 'Ungarisch'
]

# matches = model.most_similar(positive=["Frau"], negative=[], topn=30)
# words = [match[0] for match in matches]


def draw_words(model, words, pca=False, alternate=True, arrows=True, x1=3, x2=3, y1=3, y2=3, title=''):
    """
    Reduces dimensionality of vectors of given words either with PCA or with t-SNE and draws the words into a diagram.

    :param model: to visualize vectors from
    :param words: list of word strings to visualize
    :param pca: use PCA (True) or t-SNE (False) to reduce dimensionality
    :param alternate: use different color and label align for every second word
    :param arrows: use arrows to connect related words (items that are next to each other in list)
    :param x1: x axis range (from)
    :param x2: x axis range (to)
    :param y1: y axis range (from)
    :param y2: y axis range (to)
    :param title: for diagram
    :return: None
    """
    # get vectors for given words from model
    vectors = [model[word] for word in words]

    if pca:
        pca = PCA(n_components=2, whiten=True)
        vectors2d = pca.fit(vectors).transform(vectors)
    else:
        tsne = TSNE(n_components=2, random_state=0)
        vectors2d = tsne.fit_transform(vectors)

    # draw image
    plt.figure(figsize=(6, 6))
    if pca:
        plt.axis([x1, x2, y1, y2])

    first = True  # color alternation to divide given groups
    for point, word in zip(vectors2d, words):
        # plot points
        plt.scatter(point[0], point[1], c='r' if first else 'g')
        # plot word annotations
        plt.annotate(
            word, 
            xy=(point[0], point[1]),
            xytext=(-7, -6) if first else (7, -6),
            textcoords='offset points',
            ha='right' if first else 'left',
            va='bottom',
            size="x-large"
        )
        first = not first if alternate else first

    # draw arrows
    if arrows:
        for i in range(0, len(words)-1, 2):
            a = vectors2d[i][0] + 0.04
            b = vectors2d[i][1]
            c = vectors2d[i+1][0] - 0.04
            d = vectors2d[i+1][1]
            plt.arrow(
                a, b, c-a, d-b,
                shape='full',
                lw=0.1,
                edgecolor='#bbbbbb',
                facecolor='#bbbbbb',
                length_includes_head=True,
                head_width=0.08,
                width=0.01
            )

    # draw diagram title
    if title:
        plt.title(title)

    plt.tight_layout()
    plt.show()

# get trained model
trained_model = gensim.models.KeyedVectors.load_word2vec_format(args.model, binary=True)
# draw pca plots
draw_words(trained_model, currency, True, True, True, -3, 3, -2, 2, r'$PCA\ Visualisierung:\ W\ddot{a}hrung$')
draw_words(trained_model, capital, True, True, True, -3, 3, -2, 2.2, r'$PCA\ Visualisierung:\ Hauptstadt$')
draw_words(trained_model, language, True, True, True, -3, 3, -2, 1.7, r'$PCA\ Visualisierung:\ Sprache$')