This ticket is a sub compound of implementing sparse vector functionality to Jina. Currently, there is some low-level code implementation to support sparse vectors. Over the course of this epic we will implement the remaining logic. Main goal is to be able to build a fully neural search system where embedding representations are sparse vectors. This will support certain usecase like TD-IDF, and provide the functionitilty for graph embedding in the future.
Acceptance Criteria
We need to be able to store sparse vectors in blob field of Document content
We need to be able to store sparse vectors in embedding field of Document
Feature Request
This ticket is a sub compound of implementing sparse vector functionality to Jina. Currently, there is some low-level code implementation to support sparse vectors. Over the course of this epic we will implement the remaining logic. Main goal is to be able to build a fully neural search system where embedding representations are
sparse vectors
. This will support certain usecase like TD-IDF, and provide the functionitilty for graph embedding in the future.Acceptance Criteria
We need to be able to store
sparse
vectors inblob
field ofDocument content
We need to be able to store
sparse
vectors inembedding
field ofDocument