In semantic search, both the query and the search target objects (such as text, images, videos, and audio data) are represented as vectors. The search results are returned in the order of proximity to the query vector, enabling the retrieval of objects that are semantically closest to the query. This approach generally eliminates the need for maintaining a synonym dictionary, which is common in keyword search.
Furthermore, because both the query and the search target objects are represented as vectors, semantic search is not limited to text-based searches like keyword search is; it can handle searches across different media types including text, images, videos, and audio data.
