Vector data for semantic text search is automatically protected by backup, multiplexing, or replication settings that specify the tablespace, database, or instance in which it is stored.

This section describes the points to keep in mind when using Fujitsu Enterprise Postgres features to protect vector data for semantic text search.

Vectorization data using the semantic vector embedding model is a conversion that corresponds the semantic similarity between the original data to the closeness of the distance between vectors. Because there is a risk that the meaning of the original data can be inferred from vector data, it should be protected at the same level as the original data.

Multiple database objects are created by the vectorization definition. These database objects are the targets of protection.