Wine Recommendation System

Current major wine websites are not doing a good job on recommending wines to users. They either just provide simple tag based filtering, or use human agent to provide recommendations mainly based on wine color and price. As wine enthusiasts ourselves, we'd like to have a more powerful and user-friendly wine recommendation system to help us find our ideal wines.

We thus decided to build a wine recommender system by ourselves! Our system provides wine drinkers with useful recommendations to narrow down their choices based on description of their expected wines, by combining NLP model with other additional models in the pipeline based on wine categories (color, price, style, expert review score and etc.). It also provides context, along with useful information that may help them to enjoy the experience of a certain wine. The target audience of this product is wine enthusiasts, but also novices who are interested in learning about wines.

User will interact with the recommender using either the website version, or the mobile app version which allows user to enter their wine descriptions using audio-to-text methods to save their time.