Product Recommendation System using AI Agents or AgenticChatBot

This project builds a recommendation system for products using LangChain, OpenAI, and Sentence Transformers. The system leverages product features, price, and image data to provide relevant recommendations to users.

Table of Contents

1. Setup and Installation
2. Data Loading and Preprocessing
3. Feature Extraction
4. Recommendation Agents

• Feature-Based Agent
• Price-Based Agent
• Image-Retrieval-Generaion Agent
• Summary Agent

5. Query Routing
6. Text and Image Embeddings
7. Product Recommendation using Sentence Transformers
8. Gradio Interface
9. Additional Testing

1. Setup and Installation

This section installs the necessary libraries using pip:

• langchain, langchain_community, tiktoken, faiss-cpu, gradio, Pillow, transformers It then imports the required libraries for data processing, machine learning, and LangChain functionality.

2. Data Loading and Preprocessing

• Reads the Amazon cell phone dataset (DB_Amazon_cellphone_features.csv).
• Prepares the data for feature extraction and recommendation.
• Sets up OpenAI API key and models (GPT-4, GPT-3.5) HuggingFace SentenceTransformers, DALL-E, StableDiffusion.

3. Feature Extraction

• Defines a function extract_input_features to extract key features from product descriptions using OpenAI's chat models.
• Defines a function recommend_similar_features to recommend products with similar features using TF-IDF vectorization and cosine similarity.

4. Recommendation Agents

Feature-Based Agent

• Creates a features_tool using LangChain's Tool to recommend products based on features.
• Initializes a recommendation_agent using LangChain's initialize_agent with the features_tool.
• Uses GPT-4 as the language model.

Price-Based Agent

• Defines a function create_price_agent to create a price-based recommendation tool.
• Creates a price_tool using create_price_agent and Amazon cell phone data.
• Initializes a price_agent using LangChain's initialize_agent with the price_tool.

5. Query Routing

• Defines a function route_query_to_agent to determine which agent should handle a user query.
• Uses GPT-3.5 to route queries based on keywords related to price or features.

6. Text and Image Embeddings

• Preprocesses image URLs and extracts image data.
• Generates image and text embeddings using Sentence Transformers.
• Defines functions search_feature_text and search_feature_image to search for similar products using text and image embeddings.

7. Product Recommendation using Sentence Transformers

Feature-Based Recommendation

• Defines a function recommend_product_features to recommend products based on feature similarity.
• Creates a LangChain chain using GPT-3.5 and a custom prompt template.

Image-Based Recommendation

• Defines a function recommend_image_features to recommend products based on image similarity.

8. Gradio Interface

• Builds a Gradio interface to interact with the recommendation system.
• Allows users to enter queries and view product information, pros, cons, and images.

9. Additional Testing

• Includes various test queries to showcase the capabilities of the recommendation system. User prompt: "Suggest a smartphone with 16GB RAM under $800." Expected Output: Relevant products extracted from features and reviews, also images. User prompt:: "Show me laptops with 512GB SSD." Expected Output: Laptops meeting criteria with features and ratings and generate missing Images.

Note

This README provides a general overview of the code. For more detailed explanations, refer to the comments within the notebook.