{ "cells": [ { "cell_type": "code", "execution_count": 1, "id": "3f943bb9", "metadata": {}, "outputs": [], "source": [ "import pandas as pd\n", "import os" ] }, { "cell_type": "code", "execution_count": 5, "id": "e1997b1a", "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
comment_idannotator_idplatformsentimentrespectinsulthumiliatestatusdehumanizeviolence...annotator_religion_jewishannotator_religion_mormonannotator_religion_muslimannotator_religion_nothingannotator_religion_otherannotator_sexuality_bisexualannotator_sexuality_gayannotator_sexuality_straightannotator_sexuality_othertranslated
0477771087330.00.00.00.02.00.00.0...FalseFalseFalseFalseFalseFalseFalseTrueFalseOui, en effet. Elle me rappelle en quelque sor...
139773279020.00.00.00.02.00.00.0...FalseFalseFalseFalseFalseFalseFalseTrueFalseLa femme trans en train de lire ce message est...
247101337934.04.04.04.04.04.00.0...FalseFalseFalseTrueFalseFalseFalseTrueFalseQuestion & #160;: ces 4 filles qui critiquent ...
343625736532.03.02.01.02.00.00.0...FalseFalseFalseFalseFalseFalseFalseTrueFalseIl est temps pour tous les illégaux de retourn...
41253848804.04.04.04.04.04.04.0...FalseFalseFalseFalseFalseFalseFalseTrueFalsePour commencer, repliez -vous sur celui en ros...
..................................................................
13555137080859021.01.00.00.02.00.00.0...FalseFalseFalseFalseFalseFalseFalseTrueFalseéquilibréealphilippines. nicolas. ternisien. t...
13555222986830322.00.00.00.02.00.00.0...FalseFalseFalseFalseTrueTrueFalseFalseFalseDes millions de #Yémen-is ont participé à des ...
13555321008620721.01.01.01.01.00.00.0...FalseFalseFalseFalseFalseFalseFalseTrueFalse@AbeShinzo @realDonaldTump @shinzoabe Ðţţţţţţţ...
13555422986788622.00.00.00.02.00.00.0...FalseFalseFalseTrueFalseFalseFalseTrueFalseDes millions de #Yémen-is ont participé à des ...
13555514785689704.04.04.02.02.02.03.0...FalseFalseFalseFalseFalseFalseFalseTrueFalseOCTROR DECIDENTIBILITÉ OCTRORIPERIPERIPERISERI...
\n", "

135388 rows × 132 columns

\n", "
" ], "text/plain": [ " comment_id annotator_id platform sentiment respect insult \\\n", "0 47777 10873 3 0.0 0.0 0.0 \n", "1 39773 2790 2 0.0 0.0 0.0 \n", "2 47101 3379 3 4.0 4.0 4.0 \n", "3 43625 7365 3 2.0 3.0 2.0 \n", "4 12538 488 0 4.0 4.0 4.0 \n", "... ... ... ... ... ... ... \n", "135551 37080 8590 2 1.0 1.0 0.0 \n", "135552 22986 8303 2 2.0 0.0 0.0 \n", "135553 21008 6207 2 1.0 1.0 1.0 \n", "135554 22986 7886 2 2.0 0.0 0.0 \n", "135555 14785 6897 0 4.0 4.0 4.0 \n", "\n", " humiliate status dehumanize violence ... \\\n", "0 0.0 2.0 0.0 0.0 ... \n", "1 0.0 2.0 0.0 0.0 ... \n", "2 4.0 4.0 4.0 0.0 ... \n", "3 1.0 2.0 0.0 0.0 ... \n", "4 4.0 4.0 4.0 4.0 ... \n", "... ... ... ... ... ... \n", "135551 0.0 2.0 0.0 0.0 ... \n", "135552 0.0 2.0 0.0 0.0 ... \n", "135553 1.0 1.0 0.0 0.0 ... \n", "135554 0.0 2.0 0.0 0.0 ... \n", "135555 2.0 2.0 2.0 3.0 ... \n", "\n", " annotator_religion_jewish annotator_religion_mormon \\\n", "0 False False \n", "1 False False \n", "2 False False \n", "3 False False \n", "4 False False \n", "... ... ... \n", "135551 False False \n", "135552 False False \n", "135553 False False \n", "135554 False False \n", "135555 False False \n", "\n", " annotator_religion_muslim annotator_religion_nothing \\\n", "0 False False \n", "1 False False \n", "2 False True \n", "3 False False \n", "4 False False \n", "... ... ... \n", "135551 False False \n", "135552 False False \n", "135553 False False \n", "135554 False True \n", "135555 False False \n", "\n", " annotator_religion_other annotator_sexuality_bisexual \\\n", "0 False False \n", "1 False False \n", "2 False False \n", "3 False False \n", "4 False False \n", "... ... ... \n", "135551 False False \n", "135552 True True \n", "135553 False False \n", "135554 False False \n", "135555 False False \n", "\n", " annotator_sexuality_gay annotator_sexuality_straight \\\n", "0 False True \n", "1 False True \n", "2 False True \n", "3 False True \n", "4 False True \n", "... ... ... \n", "135551 False True \n", "135552 False False \n", "135553 False True \n", "135554 False True \n", "135555 False True \n", "\n", " annotator_sexuality_other \\\n", "0 False \n", "1 False \n", "2 False \n", "3 False \n", "4 False \n", "... ... \n", "135551 False \n", "135552 False \n", "135553 False \n", "135554 False \n", "135555 False \n", "\n", " translated \n", "0 Oui, en effet. Elle me rappelle en quelque sor... \n", "1 La femme trans en train de lire ce message est... \n", "2 Question & #160;: ces 4 filles qui critiquent ... \n", "3 Il est temps pour tous les illégaux de retourn... \n", "4 Pour commencer, repliez -vous sur celui en ros... \n", "... ... \n", "135551 équilibréealphilippines. nicolas. ternisien. t... \n", "135552 Des millions de #Yémen-is ont participé à des ... \n", "135553 @AbeShinzo @realDonaldTump @shinzoabe Ðţţţţţţţ... \n", "135554 Des millions de #Yémen-is ont participé à des ... \n", "135555 OCTROR DECIDENTIBILITÉ OCTRORIPERIPERIPERISERI... \n", "\n", "[135388 rows x 132 columns]" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "dataset_path = \"../../data/translated/en/ucberkeley-dlab_measuring-hate-speech_translated_positive_en_translated.csv\"\n", "\n", "df = pd.read_csv(dataset_path, index_col=[0])\n", "df" ] }, { "cell_type": "code", "execution_count": 6, "id": "272a8a64", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Worst : 14206\n", "Best : 14206\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "/tmp/ipykernel_18064/1870029499.py:38: SettingWithCopyWarning: \n", "A value is trying to be set on a copy of a slice from a DataFrame.\n", "Try using .loc[row_indexer,col_indexer] = value instead\n", "\n", "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n", " worst_df[\"label\"] = 1\n" ] } ], "source": [ "# Preprocessing\n", "import os\n", "import re\n", "\n", "def clean_text(text):\n", " \n", " patterns = [\"(\", \")\", \"Name\", \"URL\", \"NAME\", \"probably does not need a translation\", \"(optional\", \"optional,\"]\n", " out = text\n", " for p in patterns:\n", " out = out.replace(p, \"\")\n", " \n", " #out = \"\".join([x for x in out if x.isascii() or x == \" \"])\n", " \n", " out = re.sub(\"#[0-9]\", \"\", out)\n", " out = re.sub(r\"& 60;\", \"\", out)\n", " out = out.replace(\"#\", \"\") # supression des hashtag après erreur de decoding\n", " out = out.replace(\"@\", \"\")\n", " \n", " return out\n", "\n", "OUTPUT_DIR = \"../data/processed/\"\n", "if not os.path.exists(OUTPUT_DIR):\n", " os.makedirs(OUTPUT_DIR)\n", "\n", "worst_threshold = 0.9\n", "best_threshold = 0.1\n", "\n", "avg_score = pd.DataFrame(df.groupby(\"translated\").hate_speech_score.mean())\n", "avg_score = avg_score.sort_values(\"hate_speech_score\", ascending=False).reset_index()\n", "avg_score[\"translated_clean\"] = avg_score.translated.apply(clean_text)\n", "\n", "worst_df = avg_score.loc[avg_score.hate_speech_score > 0]\n", "n_worst = worst_df.shape[0]\n", "\n", "best_df = avg_score.loc[avg_score.hate_speech_score < 0]\n", "best_df = best_df.iloc[:n_worst] # keep balancing\n", "\n", "worst_df[\"label\"] = 1\n", "best_df[\"label\"] = 0\n", "\n", "print(\"Worst :\", worst_df.shape[0])\n", "print(\"Best :\", best_df.shape[0])\n", "\n", "worst_df.to_csv(os.path.join(OUTPUT_DIR, \"uc-berkeley-measuring-hate-speech-positive-translated.csv\"))\n", "best_df.to_csv(os.path.join(OUTPUT_DIR, \"uc-berkeley-measuring-hate-speech-negative-translated.csv\"))" ] } ], "metadata": { "kernelspec": { "display_name": "sexism_detection", "language": "python", "name": "sexism_detection" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.9.15" } }, "nbformat": 4, "nbformat_minor": 5 }