Recommendation engine based on user behavior and product data for the Food & Beverage category
Overview
- Provide a single platform for users to discover similar and complementary products.
- Improve customer engagement and satisfaction by offering personalized recommendations.
- Leverage AI to analyze product data and generate actionable recommendations.
- Ensure the system can handle large datasets and adapt to new product additions.
Solution
- Leveraged Azure OpenAI to generate product descriptions and tags, ensuring rich and consistent metadata for analysis.
- Integrated product data from client app, including OpenAI-generated descriptions, tags, and embeddings, to create a unified dataset.
- Applied TF-IDF for similar item identification and Sentence Transformers for complementary item recommendations, generating 1000+ dynamic product pairings based on business rules.
- Developed a recommendation engine using cosine similarity and semantic embeddings to compare products and rank them by relevance.
- Applied category-based filtering (e.g., Main Course → Beverages, Snacks → Desserts) to ensure logical and personalized recommendations.
- Scalable architecture designed to handle growing product catalogs and user bases, ensuring long-term performance and adaptability.
Impacts
Personalized recommendations improve customer engagement and satisfaction.
The system provides real-time recommendations based on user behavior and product data.
The recommendation system is designed to scale with the growing product catalog.
