Semantic search of wines

This project's origin is here and the Hackernoon article: "Semantic Search Queries Return More Informed Results".

(TODO: Add demo video)

(TODO: Add Description)

The transformer model sentence-transformers/msmarco-distilroberta-base-v2 is connected to Weaviate as vectorization module (see docker-compose.yaml file).

Prerequisites

(TO DO)

Setup instructions

(TO DO)

Usage instructions

1. Start up Weaviate: docker-compose up -d. Once completed, Weaviate is running on http://localhost:8080.
2. Install requirements: pip install -r requirements.txt
3. Run python import.py to import 2500 wines to Weaviate.
4. Navigate to console.semi.technology, connect to http://localhost:8080, navigate to the query module, and happy querying!

Dataset license

This folder contains Wine review data, retrieved from Kaggle (from WineEnthusiast).