1 – What is Weaviate search?
“Weaviate is a low-latency vector search engine with out-of-the-box support for different media types (text, images, etc.). It offers Semantic Search, Question-Answer Extraction, Classification, Customizable Models (PyTorch/TensorFlow/Keras), etc.”
Weaviate is therefore an open source search engine using sophisticated machine learning models. WPSOLR sends post types data to Weaviate, which converts and stores text/images as numbers (vectors). During the search, the query text/image is also converted to numbers, and compared to the stored numbers in the database to fetch the results.
Weaviate can use many models, open source or closed, from: HuggingFace, OpenAI, Cohere.
2 – Create the Weaviate index from WPSOLR admin
2.1 The WPSOLR new index form
- (1), (2), then (3) to edit a new index
- (4) Select “Weaviate” among index types
- (5) Select “No hosting service” among Hosting Services
You will now choose a “vectorizer”, a model that will convert your data to vectors. There are several vectorizers available, each one described in its own chapter below.
2.2 Choose a vectorizer for your index