app.py

import streamlit as st
from bertopic import BERTopic
import streamlit.components.v1 as components
from sentence_transformers import SentenceTransformer
from umap import UMAP
from hdbscan import HDBSCAN

# Initialize BERTopic model
model = BERTopic()

st.subheader("Topic Modeling with Topic-Wizard")
uploaded_file = st.file_uploader("Choose a text file", type=["txt"])

if uploaded_file is not None:
    st.session_state["text"] = uploaded_file.getvalue().decode("utf-8")

st.write("OR")

input_text = st.text_area(
    label="Enter text separated by newlines",
    value="",
    key="text",
    height=150,
)

button = st.button("Get Segments")

if button and (uploaded_file is not None or input_text != ""):
    if uploaded_file is not None:
        texts = st.session_state["text"].split("\n")
    else:
        texts = input_text.split("\n")

    # Fit BERTopic model
    topics, probabilities = model.fit_transform(texts)

    # Create embeddings
    embeddings_model = SentenceTransformer("distilbert-base-nli-mean-tokens")
    embeddings = embeddings_model.encode(texts)

    # Reduce dimensionality of embeddings using UMAP
    umap_model = UMAP(n_neighbors=15, n_components=2, metric="cosine")
    umap_embeddings = umap_model.fit_transform(embeddings)

    # Cluster topics using HDBSCAN
    cluster = HDBSCAN(
        min_cluster_size=15, metric="euclidean", cluster_selection_method="eom"
    ).fit(umap_embeddings)

    # Visualize BERTopic results with Streamlit
    st.title("BERTopic Visualization")

    # Display top N most representative topics and their documents
    num_topics = st.sidebar.slider("Select number of topics to display", 1, 20, 5, 1)
    topic_words, topic_docs = model.get_topics(with_documents=True)
    for i, topic in enumerate(topic_words.items()):
        if i >= num_topics:
            break
        st.write(f"## Topic {topic[0]}")
        st.write("Keywords:", ", ".join(topic[1]))
        st.write("Documents:")
        for doc in topic_docs[topic[0]][:5]:
            st.write("-", texts[doc])

    # Display topic clusters
    st.write("## Topic Clusters")
    components.html(cluster.labels_.tolist(), height=500, width=800)