I created a database that contains company profiles and their associated files. My primary requirement is vector search functionality.
Schema:
Company ----[:HAS FILE]----> Files ----[:HAS CHUNK]----> ChunkNode
The database structure is as follows:
- A Company node is connected to File nodes via a
[:HAS FILE]
relationship. - Each File node is further connected to ChunkNode nodes via a
[:HAS CHUNK]
relationship.
I currently have data for 100 companies.
- This results in 100 Company nodes.
- Each company has 3-4 File nodes associated with it.
- These files are divided into chunks, which are embedded and stored as ChunkNode nodes.
This schema is designed to enable efficient vector searches on the embedded data.
In this database, each company forms its own distinct tree.
- A company's nodes, whether File or ChunkNode, will not have any relationships with nodes belonging to another company.
- This ensures that the data for each company is completely isolated within its own subgraph.
I need to perform a vector search where a question is given.
- identify the company node
- navigate to file node
- then perform vector search on the ChunkNode
anyone have idea on this...