Background

Semantic search is being read here in the context of machine learning.
Traditional search works on matching keywords in a data source.
Semantic search thinks about search with the meaning and context in consideration.

What it is

Think of this as a concept of plotting ideas and keywords in an n dimension by meaning.
Such a concept becomes important in grouping ideas together, allowing it to be identified by meaning using vectors.

How are these embeddings created

The object is fed through embedding models such as BERT and T5 to generate vector representations.