|
import streamlit as st |
|
import pandas as pd |
|
import os |
|
from sacrebleu import corpus_bleu |
|
if st.session_state.Cloud == 0: |
|
from sklearn.cluster import KMeans |
|
from sklearn.neighbors import KNeighborsClassifier |
|
from sklearn.ensemble import RandomForestClassifier |
|
from translate_app import tr |
|
|
|
title = "Traduction mot à mot" |
|
sidebar_name = "Traduction mot à mot" |
|
dataPath = st.session_state.DataPath |
|
|
|
@st.cache_data |
|
def load_corpus(path): |
|
input_file = os.path.join(path) |
|
with open(input_file, "r", encoding="utf-8") as f: |
|
data = f.read() |
|
data = data.split('\n') |
|
data=data[:-1] |
|
return pd.DataFrame(data) |
|
|
|
@st.cache_data |
|
def load_BOW(path, l): |
|
input_file = os.path.join(path) |
|
df1 = pd.read_csv(input_file+'1_'+l, encoding="utf-8", index_col=0) |
|
df2 = pd.read_csv(input_file+'2_'+l, encoding="utf-8", index_col=0) |
|
df_count_word = pd.concat([df1, df2]) |
|
return df_count_word |
|
|
|
df_data_en = load_corpus(dataPath+'/preprocess_txt_en') |
|
df_data_fr = load_corpus(dataPath+'/preprocess_txt_fr') |
|
df_count_word_en = load_BOW(dataPath+'/preprocess_df_count_word', 'en') |
|
df_count_word_fr = load_BOW(dataPath+'/preprocess_df_count_word', 'fr') |
|
n1 = 0 |
|
|
|
def accuracy(dict_ref,dict): |
|
correct_words = 0 |
|
|
|
for t in dict.columns: |
|
if t in dict_ref.columns: |
|
if str(dict[t]) == str(dict_ref[t]): |
|
correct_words +=1 |
|
else: print("dict ref: manque:",t) |
|
print(correct_words," mots corrects / ",min(dict.shape[1],dict_ref.shape[1])) |
|
return correct_words/min(dict.shape[1],dict_ref.shape[1]) |
|
|
|
if st.session_state.reCalcule: |
|
nb_mots_en = 199 |
|
nb_mots_fr = 330 |
|
|
|
|
|
df_count_word_en = df_count_word_en[df_count_word_en==0].fillna(1) |
|
df_count_word_fr = df_count_word_fr[df_count_word_fr==0].fillna(1) |
|
|
|
|
|
if ('new' in df_count_word_en.columns): |
|
df_count_word_en['new']=df_count_word_en['new']*2 |
|
df_count_word_fr['new']=df_count_word_fr['new']*2 |
|
|
|
def calc_kmeans(l_src,l_tgt): |
|
global df_count_word_src, df_count_word_tgt, nb_mots_src, nb_mots_tgt |
|
|
|
|
|
init_centroids = df_count_word_tgt.T |
|
kmeans = KMeans(n_clusters = nb_mots_tgt, n_init=1, max_iter=1, init=init_centroids, verbose=0) |
|
|
|
kmeans.fit(df_count_word_tgt.T) |
|
|
|
|
|
centroids= kmeans.cluster_centers_ |
|
labels = kmeans.labels_ |
|
|
|
|
|
df_dic = pd.DataFrame(data=df_count_word_tgt.columns[kmeans.predict(df_count_word_src.T)],index=df_count_word_src.T.index,columns=[l_tgt]) |
|
df_dic.index.name= l_src |
|
df_dic = df_dic.T |
|
|
|
|
|
|
|
|
|
return df_dic |
|
|
|
def calc_knn(l_src,l_tgt, metric): |
|
global df_count_word_src, df_count_word_tgt, nb_mots_src, nb_mots_tgt |
|
|
|
|
|
knn_metric = metric |
|
|
|
|
|
X_train = df_count_word_tgt.T |
|
y_train = range(nb_mots_tgt) |
|
|
|
|
|
knn = KNeighborsClassifier(n_neighbors=1, metric=knn_metric) |
|
knn.fit(X_train, y_train) |
|
|
|
|
|
df_dic = pd.DataFrame(data=df_count_word_tgt.columns[knn.predict(df_count_word_src.T)],index=df_count_word_src.T.index,columns=[l_tgt]) |
|
df_dic.index.name = l_src |
|
df_dic = df_dic.T |
|
|
|
|
|
|
|
|
|
|
|
return df_dic |
|
|
|
def calc_rf(l_src,l_tgt): |
|
|
|
|
|
X_train = df_count_word_tgt.T |
|
y_train = range(nb_mots_tgt) |
|
|
|
|
|
rf = RandomForestClassifier(n_jobs=-1, random_state=321) |
|
rf.fit(X_train, y_train) |
|
|
|
|
|
df_dic = pd.DataFrame(data=df_count_word_tgt.columns[rf.predict(df_count_word_src.T)],index=df_count_word_src.T.index,columns=[l_tgt]) |
|
df_dic.index.name= l_src |
|
df_dic = df_dic.T |
|
|
|
|
|
|
|
|
|
|
|
return df_dic |
|
|
|
def calcul_dic(Lang,Algo,Metrique): |
|
|
|
if Lang[:2]=='en': |
|
l_src = 'Anglais' |
|
l_tgt = 'Francais' |
|
else: |
|
l_src = 'Francais' |
|
l_tgt = 'Anglais' |
|
|
|
if Algo=='Manuel': |
|
df_dic = pd.read_csv('../data/dict_ref_'+Lang+'.csv',header=0,index_col=0, encoding ="utf-8", sep=';',keep_default_na=False).T.sort_index(axis=1) |
|
elif Algo=='KMeans': |
|
df_dic = calc_kmeans(l_src,l_tgt) |
|
elif Algo=='KNN': |
|
df_dic = calc_knn(l_src,l_tgt, Metrique) |
|
elif Algo=='Random Forest': |
|
df_dic = calc_rf(l_src,l_tgt) |
|
else: |
|
df_dic = pd.read_csv('../data/dict_we_'+Lang,header=0,index_col=0, encoding ="utf-8", keep_default_na=False).T.sort_index(axis=1) |
|
|
|
return df_dic |
|
else: |
|
def load_dic(Lang,Algo,Metrique): |
|
|
|
Algo = Algo.lower() |
|
if Algo=='random forest' : Algo = "rf" |
|
else: |
|
if Algo=='word embedding' : Algo = "we" |
|
else: |
|
if Algo!='knn': Metrique = '' |
|
else: Metrique = Metrique+'_' |
|
input_file = os.path.join(dataPath+'/dict_'+Algo+'_'+Metrique+Lang) |
|
return pd.read_csv(input_file, encoding="utf-8", index_col=0).T.sort_index(axis=1) |
|
|
|
|
|
def display_translation(n1,dict, Lang): |
|
global df_data_src, df_data_tgt, placeholder |
|
|
|
s = df_data_src.iloc[n1:n1+5][0].tolist() |
|
s_trad = [] |
|
s_trad_ref = df_data_tgt.iloc[n1:n1+5][0].tolist() |
|
source = Lang[:2] |
|
target = Lang[-2:] |
|
for i in range(5): |
|
|
|
|
|
|
|
s_trad.append((' '.join(dict[col].iloc[0] for col in s[i].split()))) |
|
st.write("**"+source+" :** :blue["+ s[i]+"]") |
|
st.write("**"+target+" :** "+s_trad[-1]) |
|
st.write("**ref. :** "+s_trad_ref[i]) |
|
st.write("") |
|
with placeholder: |
|
st.write("<p style='text-align:center;background-color:red; color:white')>"+"Score Bleu = "+str(int(round(corpus_bleu(s_trad,[s_trad_ref]).score,0)))+"%</p>", \ |
|
unsafe_allow_html=True) |
|
|
|
def display_dic(df_dic): |
|
st.dataframe(df_dic.T, height=600) |
|
|
|
def save_dic(path, df_dic): |
|
output_file = os.path.join(path) |
|
df_dic.T.to_csv(output_file, encoding="utf-8") |
|
return |
|
|
|
def run(): |
|
global df_data_src, df_data_tgt, df_count_word_src, df_count_word_tgt, nb_mots_src, nb_mots_tgt, n1, placeholder |
|
global df_data_en, df_data_fr, nb_mots_en, df_count_word_en, df_count_word_fr, nb_mots_en, nb_mots_fr |
|
|
|
st.write("") |
|
st.title(tr(title)) |
|
|
|
|
|
st.write("## **"+tr("Explications")+" :**\n") |
|
st.markdown(tr( |
|
""" |
|
Dans une première approche naïve, nous avons implémenté un système de traduction mot à mot. |
|
Cette traduction est réalisée grâce à un dictionnaire qui associe un mot de la langue source à un mot de la langue cible, dans small_vocab |
|
Ce dictionnaire est calculé de 3 manières: |
|
""") |
|
, unsafe_allow_html=True) |
|
st.markdown( |
|
"* "+tr(":red[**Manuellement**] en choisissant pour chaque mot source le mot cible. Ceci nous a permis de définir un dictionnaire de référence")+"\n"+ \ |
|
"* "+tr("Avec le :red[**Bag Of World**] (chaque mot dans la langue cible = une classe, BOW = features)") |
|
, unsafe_allow_html=True) |
|
st.image("assets/BOW.jpg",use_column_width=True) |
|
st.markdown( |
|
"* "+tr("Avec le :red[**Word Embedding**], c'est à dire en associant chaque mot à un vecteur \"sémantique\" de dimensions=300, et en selectionnant le vecteur de langue cible " |
|
"le plus proche du vecteur de langue source.")+" \n\n"+ |
|
tr("Enfin nous calculons :")+"\n"+ \ |
|
"* "+tr("la :red[**précision**] du dictionnaire par rapport à notre dictionnaire de réference (manuel)")+"\n"+ \ |
|
"* "+tr("le ")+" :red[**score BLEU**] (\"BiLingual Evaluation Understudy\")"+tr(", qui mesure la précision de notre traduction par rapport à celle de notre corpus référence. ") |
|
, unsafe_allow_html=True) |
|
|
|
st.write("## **"+tr("Paramètres ")+" :**\n") |
|
Sens = st.radio(tr('Sens')+' :',('Anglais -> Français','Français -> Anglais'), horizontal=True) |
|
Lang = ('en_fr' if Sens=='Anglais -> Français' else 'fr_en') |
|
Algo = st.radio(tr('Algorithme')+' :',('Manuel', 'KMeans','KNN','Random Forest','Word Embedding'), horizontal=True) |
|
Metrique = '' |
|
if (Algo == 'KNN'): |
|
Metrique = st.radio(tr('Metrique')+':',('minkowski', 'cosine', 'chebyshev', 'manhattan', 'euclidean'), horizontal=True) |
|
|
|
if (Lang=='en_fr'): |
|
df_data_src = df_data_en |
|
df_data_tgt = df_data_fr |
|
if st.session_state.reCalcule: |
|
df_count_word_src = df_count_word_en |
|
df_count_word_tgt = df_count_word_fr |
|
nb_mots_src = nb_mots_en |
|
nb_mots_tgt = nb_mots_fr |
|
else: |
|
df_data_src = df_data_fr |
|
df_data_tgt = df_data_en |
|
if st.session_state.reCalcule: |
|
df_count_word_src = df_count_word_fr |
|
df_count_word_tgt = df_count_word_en |
|
nb_mots_src = nb_mots_fr |
|
nb_mots_tgt = nb_mots_en |
|
|
|
|
|
sentence1 = st.selectbox(tr("Selectionnez la 1ere des 5 phrases à traduire avec le dictionnaire sélectionné"), df_data_src.iloc[:-4],index=int(n1) ) |
|
n1 = df_data_src[df_data_src[0]==sentence1].index.values[0] |
|
|
|
if st.session_state.reCalcule: |
|
df_dic = calcul_dic(Lang,Algo,Metrique) |
|
df_dic_ref = calcul_dic(Lang,'Manuel',Metrique) |
|
else: |
|
df_dic = load_dic(Lang,Algo,Metrique) |
|
df_dic_ref = load_dic(Lang,'Manuel',Metrique) |
|
|
|
""" |
|
save_dico = st.checkbox('Save dic ?') |
|
if save_dico: |
|
dic_name = st.text_input('Nom du fichier :',dataPath+'/dict_') |
|
save_dic(dic_name, df_dic) |
|
""" |
|
|
|
st.write("## **"+tr("Dictionnaire calculé et traduction mot à mot")+" :**\n") |
|
col1, col2 = st.columns([0.25, 0.75]) |
|
with col1: |
|
st.write("#### **"+tr("Dictionnaire")+"**") |
|
precision = int(round(accuracy(df_dic_ref,df_dic)*100, 0)) |
|
st.write("<p style='text-align:center;background-color:red; color:white')>"+tr("Précision")+" = {:2d}%</p>".format(precision), unsafe_allow_html=True) |
|
display_dic(df_dic) |
|
with col2: |
|
st.write("#### **"+tr("Traduction")+"**") |
|
placeholder = st.empty() |
|
display_translation(n1, df_dic, Lang) |
|
|