app.py

import gradio as gr
import pandas as pd
from sentence_transformers import SentenceTransformer
from sklearn.metrics.pairwise import cosine_similarity

# Load your trained SentenceTransformer model

import joblib
model = joblib.load("recommender_model.pkl")

# Load the posts
posts = pd.read_csv("posts_cleaned.csv")

# Precompute embeddings for all posts
post_texts = posts["post_text"].astype(str).tolist()
post_embeddings = model.encode(post_texts, convert_to_tensor=False)

# Recommendation function
def recommend(user_input):
    user_embedding = model.encode([user_input], convert_to_tensor=False)
    similarities = cosine_similarity(user_embedding, post_embeddings)[0]
    top_indices = similarities.argsort()[-5:][::-1]
    recommended = posts.iloc[top_indices]["post_text"].tolist()
    return "\n\n".join(recommended)

# Launch Gradio app
gr.Interface(fn=recommend, inputs="text", outputs="text", title="AI Post Recommender").launch()