The semantic text search and automatic vectorization feature utilizes a vector representation that preserves the semantic similarity between text data for the semantic similarity search of text. When you add text data, the worker process automatically generates the vector data in the background according to a predefined vectorization definition (vectorizer) and stores it in a table (an embedded table) containing the vector data corresponding to the table containing the text data.

When you define a vectorization of text data, a view (an embedded view) that combines text data with a column of vector data is also created automatically.

When you perform a semantic text search on an embedded view, a vector representation of the input text is generated internally, and a semantic text search is performed as a similarity search between the vector data.