app.py

import gradio as gr
import pandas as pd
import nltk
from nltk.corpus import stopwords
from umap import UMAP
from hdbscan import HDBSCAN
from sentence_transformers import SentenceTransformer
from sklearn.feature_extraction.text import CountVectorizer
from bertopic import BERTopic
from bertopic.representation import MaximalMarginalRelevance
from bertopic.vectorizers import ClassTfidfTransformer

# Charger les stopwords
try:
    stop_words = stopwords.words('english')
except LookupError:
    nltk.download('stopwords')
    stop_words = stopwords.words('english')

# Stopwords personnalisés
custom_stopwords = ["made", "sure"]
stop_words.extend(custom_stopwords)

# Pipeline BERTopic personnalisé
def generate_topics(file):
    docs = pd.read_csv(file.name)

    embedding_model = SentenceTransformer("all-MiniLM-L6-v2")
    embeddings = embedding_model.encode(docs['text'].tolist(), show_progress_bar=True)

    umap_model = UMAP(
        n_neighbors=20,
        n_components=5,
        min_dist=0.0,
        metric='cosine',
        random_state=42
    )

    hdbscan_model = HDBSCAN(
        min_cluster_size=60,
        min_samples=1,
        metric='euclidean',
        cluster_selection_method='eom',
        prediction_data=True
    )

    vectorizer_model = CountVectorizer(
        stop_words=stop_words,
        min_df=1,
        ngram_range=(1, 3)
    )

    ctfidf_model = ClassTfidfTransformer()
    representation_model = MaximalMarginalRelevance(diversity=0.7)

    BERT_model = BERTopic(
        embedding_model=embedding_model,
        umap_model=umap_model,
        hdbscan_model=hdbscan_model,
        vectorizer_model=vectorizer_model,
        ctfidf_model=ctfidf_model,
        representation_model=representation_model,
        verbose=True
    )

    topics, _ = BERT_model.fit_transform(docs['text'].tolist())
    BERT_model.reduce_outliers(docs['text'].tolist(), topics)

    fig = BERT_model.visualize_documents(docs['text'].tolist())
    return fig

# Interface Gradio
demo = gr.Interface(
    fn=generate_topics,
    inputs=gr.File(label="Upload bbc-text.csv"),
    outputs=gr.Plot(label="Topic Map"),
    title="Topic Modeling avec BERTopic",
    description="Téléversez un fichier CSV avec une colonne 'text' pour générer une visualisation thématique interactive."
)

if __name__ == "__main__":
    demo.launch()