Learn to optimize your KDB.AI vector searches by leveraging metadata filtering. Each vector stored in the KDB.AI vector database can have associated metadata attached. This metadata can be used to enhance similarity search by specifying filters on the attached metadata. By filtering our search, we effectively shrink the search space and reduce the number of vectors that need to be searched.
In this demo, we will use metadata filtering to enhance search results over a dataset of movie plots. We will filter metadata such as ‘genre’, ‘release year’, ‘director’, and more to narrow our searches, and increase their speed and accuracy.
Download the Jupyter notebook to try it today!