app.py

# ==============================
# Explainable Recommendation System
# Group 15 Project
# ==============================

import os
import json
import numpy as np
from fastapi import FastAPI
from pydantic import BaseModel
from sklearn.metrics.pairwise import cosine_similarity
from openai import OpenAI

# ------------------------------
# Initialize OpenAI client
# ------------------------------
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))

# ------------------------------
# Small Local Dataset
# ------------------------------
dataset = [
    {
        "id": 1,
        "title": "Introduction to Machine Learning",
        "description": "Learn supervised and unsupervised learning, regression, and classification.",
        "tags": ["machine learning", "ai", "beginner"]
    },
    {
        "id": 2,
        "title": "Deep Learning with Neural Networks",
        "description": "Advanced deep learning concepts including CNNs and RNNs.",
        "tags": ["deep learning", "neural networks", "ai"]
    },
    {
        "id": 3,
        "title": "Data Science with Python",
        "description": "Data analysis, visualization, and machine learning using Python.",
        "tags": ["python", "data science"]
    },
    {
        "id": 4,
        "title": "Natural Language Processing",
        "description": "Text processing, embeddings, and transformer models.",
        "tags": ["nlp", "transformers", "ai"]
    }
]

# ------------------------------
# Generate Embedding
# ------------------------------
def get_embedding(text: str):
    response = client.embeddings.create(
        model="text-embedding-3-small",
        input=text
    )
    return np.array(response.data[0].embedding)

# ------------------------------
# Precompute Dataset Embeddings
# ------------------------------
for item in dataset:
    combined_text = (
        item["title"] + " " +
        item["description"] + " " +
        " ".join(item["tags"])
    )
    item["embedding"] = get_embedding(combined_text)

# ------------------------------
# Recommendation Function
# ------------------------------
def recommend(user_query: str, top_k: int = 2):
    query_embedding = get_embedding(user_query)

    similarities = []
    for item in dataset:
        score = cosine_similarity(
            [query_embedding],
            [item["embedding"]]
        )[0][0]
        similarities.append((item, score))

    similarities.sort(key=lambda x: x[1], reverse=True)
    return similarities[:top_k]

# ------------------------------
# LLM Explanation Function
# ------------------------------
def generate_explanation(user_query, recommended_items):

    items_text = "\n".join([
        f"- {item['title']}: {item['description']}"
        for item, score in recommended_items
    ])

    prompt = f"""
User interest: {user_query}

Recommended items:
{items_text}

Explain clearly why these recommendations match the user's interest.
Make it personalized and easy to understand.
"""

    response = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[{"role": "user", "content": prompt}]
    )

    return response.choices[0].message.content

# ------------------------------
# FastAPI App
# ------------------------------
app = FastAPI(title="Explainable Recommendation System")

class QueryRequest(BaseModel):
    user_query: str

@app.post("/recommend")
def get_recommendation(request: QueryRequest):

    recommended = recommend(request.user_query)
    explanation = generate_explanation(request.user_query, recommended)

    return {
        "query": request.user_query,
        "recommendations": [
            {
                "title": item["title"],
                "description": item["description"],
                "score": float(score)
            }
            for item, score in recommended
        ],
        "explanation": explanation
    }

# ------------------------------
# Root Endpoint
# ------------------------------
@app.get("/")
def home():
    return {"message": "Explainable Recommendation System is running 🚀"}