PHP + pgSQL + pgvector webapp to combine words using vector embeddings and display images of vector embeddings.
- Uses spaCy to generate full vocab and corresponding wordvectors
- Word embedding vectors are stored in a PostgreSQL database with pgvector which allows fast semantic search
- Given two words, look up their wordcvectors and take the average
- Find the nearest five words to the averaged vector
A REST-style API is available. Documentation: https://docs.wordchef.app/
The public API key is wordchef-public-api-key and can be passed as a header.
There are five endpoints available from the base https://wordchef.app/api/v1/:
| Endpoint | Method(s) | Description |
|---|---|---|
health |
GET | Returns the status of the API |
embedding |
GET / POST | Returns a word embedding for query param word |
nearest |
GET / POST | Returns nearest words to query param words |
image |
GET / POST | Returns an image for query param words |
bulk_image |
POST | Returns multiple images for array words in body |