A team wants semantic document search but stores only original document text. What additional data is normally required for vector similarity retrieval?

1. A new chart color for each document
2. Suitable vector embeddings for documents and queries
3. A larger browser window
4. A shared database-administrator password