Semantic search allows for the comparison of unstructured data, such as text, images, or media files, by evaluating their semantic similarity. This is achieved by generating a vector that represents the features of the data under analysis because the vector is a convenient data structure to compress information and is easily manageable by a computer.

With vectors, we can represent thousands of features for unstructured data, such as long texts, images, audio files, and more, using lists of floating point numbers.

Check: Vector Embeddings