Skip to content

Instantly share code, notes, and snippets.

@vered1986
Created December 28, 2018 08:59
Show Gist options
  • Save vered1986/1686307861b68348e605c8833b7dd8ec to your computer and use it in GitHub Desktop.
Save vered1986/1686307861b68348e605c8833b7dd8ec to your computer and use it in GitHub Desktop.
This notebook demonstrates gender bias in word embeddings.
Display the source blob
Display the rendered blob
Raw
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Bias in Word Embeddings\n",
"\n",
"In this notebook we'll demonstrate the gender bias in word embeddings. \n",
"\n",
"We will use GloVe embeddings and project \"feminine\" vs. \"masculine\" occupations to show that the vectors indeed capture this bias.\n",
"\n",
"The [professions file](https://raw.githubusercontent.com/tolga-b/debiaswe/master/data/professions.json) is taken from the paper: \n",
"\n",
"**Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings** \n",
"*Tolga Bolukbasi, Kai-Wei Chang, James Zou, Venkatesh Saligrama, and Adam Kalai.* NIPS 2016. "
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"import json\n",
"import gzip\n",
"import random\n",
"import codecs\n",
"\n",
"import numpy as np\n",
"\n",
"%matplotlib inline\n",
"from matplotlib import pyplot as plt"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Load word embeddings:\n",
"\n",
"Download GloVe or open it from the local directory if it was already downloaded."
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Loading embeddings from /home/vered/word_embeddings/glove/glove.6B.50d.txt.gz\n"
]
}
],
"source": [
"def load_embeddings(file_name):\n",
" \"\"\"\n",
" Load the pre-trained embeddings from a file\n",
" :param file_name: the embeddings file\n",
" :return: the vocabulary and the word vectors\n",
" \"\"\"\n",
" with gzip.open(file_name, 'rt', 'utf-8') as f_in:\n",
" lines = [line.strip() for line in f_in]\n",
"\n",
" embedding_dim = len(lines[0].split()) - 1\n",
" index2word, vectors = zip(*[line.strip().split(' ', 1) \n",
" for line in lines \n",
" if len(line.split()) == embedding_dim + 1])\n",
" wv = np.loadtxt(vectors)\n",
" return wv, index2word\n",
"\n",
"\n",
"LOCAL_EMB_PATH = os.path.expanduser('~/word_embeddings/glove/glove.6B.50d.txt.gz')\n",
"REMOTE_EMB_PATH = 'https://s3-us-west-2.amazonaws.com/allennlp/datasets/glove/glove.6B.50d.txt.gz'\n",
"\n",
"if not os.path.exists(LOCAL_EMB_PATH):\n",
" print(f'Downloading embeddings from {REMOTE_EMB_PATH}')\n",
" local_dir = os.path.dirname(LOCAL_EMB_PATH)\n",
" !mkdir -p $local_dir\n",
" !wget $REMOTE_EMB_PATH -O $LOCAL_EMB_PATH\n",
"\n",
"print(f'Loading embeddings from {LOCAL_EMB_PATH}')\n",
"wv, index2word = load_embeddings(LOCAL_EMB_PATH)\n",
"word2index = {w:i for i, w in enumerate(index2word)}"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Load the list of common occupations.\n",
"\n",
"Format: \n",
"- occupation\n",
"- definitional female -1.0 -> definitional male 1.0\n",
"- stereotypical female -1.0 -> stereotypical male 1.0\n",
"\n",
"We take everything that is not defined to one gender only, but is stereotypical to this gender."
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\"Male\" occupations:\n",
"accountant\n",
"adjunct_professor\n",
"administrator\n",
"adventurer\n",
"alderman\n",
"ambassador\n",
"analyst\n",
"anthropologist\n",
"archaeologist\n",
"archbishop\n",
"\n",
"\"Female\" occupations:\n",
"advocate\n",
"aide\n",
"artist\n",
"artiste\n",
"baker\n",
"ballerina\n",
"bookkeeper\n",
"businesswoman\n",
"caretaker\n",
"choreographer\n"
]
}
],
"source": [
"LOCAL_OCC_PATH = 'professions.json'\n",
"REMOTE_OCC_PATH = 'https://raw.githubusercontent.com/tolga-b/debiaswe/master/data/professions.json'\n",
"\n",
"if not os.path.exists(LOCAL_OCC_PATH):\n",
" !wget $REMOTE_OCC_PATH\n",
" \n",
"with open(LOCAL_OCC_PATH) as f_in:\n",
" occupations = json.load(f_in)\n",
" \n",
"female_occupations = [occ for occ, def_f, ste_f in occupations if def_f != -1 and def_f <= 0 and ste_f < 0]\n",
"male_occupations = [occ for occ, def_f, ste_f in occupations if def_f != 1 and def_f >= 0 and ste_f > 0]\n",
"stereotypes = [occ for occ, def_f, ste_f in occupations if abs(def_f) == 1]\n",
"occupations = female_occupations + male_occupations + stereotypes\n",
"\n",
"print('\"Male\" occupations:')\n",
"print('\\n'.join(male_occupations[:10]))\n",
"print('')\n",
"\n",
"print('\"Female\" occupations:')\n",
"print('\\n'.join(female_occupations[:10]))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Limit the word vectors to this vocabulary. "
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [],
"source": [
"def get_vector(w, curr_wv=wv, curr_word2index=word2index):\n",
" if w in curr_word2index:\n",
" return curr_wv[curr_word2index[w], :]\n",
" else:\n",
" return None\n",
" \n",
"vocab = occupations + ['he', 'she']\n",
"word_vectors = [get_vector(w) for w in vocab]\n",
"filtered_index2word, filtered_wv = zip(*[\n",
" (w, v) for v, w in zip(word_vectors, vocab) if v is not None])\n",
"filtered_word2index = {w:i for i, w in enumerate(filtered_index2word)}\n",
"filtered_wv = np.vstack(filtered_wv)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Define the code to solve analogies as in the word2vec paper."
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"he : doctor :: she : nurse\n"
]
}
],
"source": [
"all_norms = [np.linalg.norm(filtered_wv[i, :]) \n",
" for i in range(len(filtered_index2word))]\n",
"\n",
"\n",
"def most_similar_word(v, index_to_exclude=None):\n",
" \"\"\"\n",
" Returns the most similar word to a vector v, using cosine similarity\n",
" \"\"\"\n",
" # Apply matrix-vector dot product to get the dot product of w to all the other vectors\n",
" dot_product = np.dot(filtered_wv, v)\n",
"\n",
" # Compute cosine similarity. We don't need to divide by the norm of v because\n",
" # it is constant across all words. We only need to divide by the norm of each word vector.\n",
" norm_v = np.linalg.norm(v)\n",
" cosine = dot_product / all_norms\n",
" \n",
" if index_to_exclude is not None:\n",
" best_index = [i for i in np.argsort(cosine)[::-1] if i != index_to_exclude][0] \n",
" else:\n",
" best_index = np.argmax(cosine)\n",
" \n",
" return (filtered_index2word[best_index], cosine[best_index])\n",
"\n",
"\n",
"def solve_analogy(a, b, c, threshold=-np.inf):\n",
" \"\"\"\n",
" d* = argmax_d(cos(b - a + c, d))\n",
" \"\"\"\n",
" vectors = []\n",
" for x in [a, b, c]:\n",
" curr = get_vector(x, curr_wv=filtered_wv, curr_word2index=filtered_word2index)\n",
" if curr is None:\n",
" return None, -np.inf\n",
" vectors.append(curr)\n",
" \n",
" v_a, v_b, v_c = vectors\n",
" d, score = most_similar_word(v_b - v_a + v_c, index_to_exclude=filtered_word2index[b])\n",
" \n",
" if score >= threshold:\n",
" return d, score\n",
" else:\n",
" return None, -np.inf\n",
" \n",
"\n",
"solution, score = solve_analogy('he', 'doctor', 'she')\n",
"print(f'he : doctor :: she : {solution}')"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"First demonstration of bias. \n",
"\n",
"Now, let's follow Bolukbasi et al. and define the analogies on the \"gender direction\". \n",
"\n",
"We will find analogies of type: *he : occupation_1 :: she : occupation_2* for a select number of \"masculine\" occupations."
]
},
{
"cell_type": "code",
"execution_count": 14,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"he:she\n",
"composer:pianist\n",
"guitarist:vocalist\n",
"envoy:ambassador\n",
"painter:sculptor\n",
"professor:researcher\n",
"minister:deputy\n",
"drummer:vocalist\n",
"comedian:actress\n",
"playwright:filmmaker\n",
"doctor:nurse\n",
"economist:analyst\n",
"columnist:editor\n",
"scientist:researcher\n",
"screenwriter:filmmaker\n",
"saxophonist:vocalist\n",
"editor:columnist\n",
"attorney:lawyer\n",
"narrator:protagonist\n",
"filmmaker:actress\n",
"surgeon:nurse\n",
"legislator:lawmaker\n",
"strategist:analyst\n",
"physicist:scientist\n",
"cardiologist:pediatrician\n"
]
}
],
"source": [
"word_pairs = [(('he', 'she'), np.inf)]\n",
"\n",
"for b in male_occupations:\n",
" if b in {'he', 'she'}:\n",
" continue\n",
" \n",
" d, score = solve_analogy('he', b, 'she')\n",
" if d is not None:\n",
" word_pairs.append(((b, d), score))\n",
"\n",
" \n",
"# Sort by score and take the 25 highest scoring pairs\n",
"word_pairs = [(b, d) for (b, d), score in sorted(word_pairs, key=lambda x: x[1], reverse=True)][:25]\n",
"print('\\n'.join([':'.join(pair) for pair in word_pairs]))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now let's plot a t-SNE projection, including lines between the word pairs. "
]
},
{
"cell_type": "code",
"execution_count": 15,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x7f1911485c18>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"import matplotlib.pyplot as plt\n",
"\n",
"from sklearn.manifold import TSNE\n",
"\n",
"\n",
"def plot_tsne(word_pairs):\n",
" \"\"\"\n",
" Plot a t-SNE projection of the word embeddings,\n",
" and draw lines between the word pairs (pairs of indices).\n",
" \"\"\"\n",
" projector = TSNE(n_components=2, random_state=0)\n",
" np.set_printoptions(suppress=True)\n",
" \n",
" # Limit the vocabulary \n",
" w1s, w2s = zip(*word_pairs)\n",
" limited_vocab = list(set(w1s).union(w2s)) \n",
" curr_word2index = {w:i for i, w in enumerate(limited_vocab)}\n",
" curr_wv = np.vstack([get_vector(w) for w in limited_vocab])\n",
" Y = projector.fit_transform(curr_wv)\n",
"\n",
" fig = plt.figure(figsize=(12, 12))\n",
" ax = plt.axes()\n",
" ax.scatter(Y[:, 0], Y[:, 1])\n",
" for w, x, y in zip(limited_vocab, Y[:, 0], Y[:, 1]):\n",
" ax.annotate(w, xy=(x, y), xytext=(0, 0), textcoords='offset points', fontsize=12)\n",
"\n",
" # Draw lines\n",
" for b, d in word_pairs:\n",
" indices = [curr_word2index[b], curr_word2index[d]]\n",
" i, j = min(indices), max(indices) \n",
" plt.plot([Y[i, 0], Y[j, 0]], [Y[i, 1], Y[j, 1]])\n",
" \n",
" plt.show()\n",
"\n",
"\n",
"plot_tsne(word_pairs)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"It's not a very strong signal, but we can see that the projection of some word pairs has a similar direction as *he:she*: *doctor:nurse*, *professor:researcher*, *minister:deputy*, etc. "
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.4"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment