Semantic Search, as opposed to lexical search, stands for searching by understanding the overall meaning of query terms rather than looking at literal matches of search words entered by the user. The relevance of the search results is refined by looking at the general contextual meaning of the input phrase typed in by the user as they search for a particular topic. Categorization of text is also more coherent when done based on the meaning of the text than looking at matches for individual words.
Advanced Natural Language Processing engines can extract the meaning out of text, facilitating categorization of text based on the meaning it contains. The semantic based approach has the advantage of being able to find similar sentences, even when the search phrase has words the NLP engine has not seen previously, as the engine figures out the overall meaning of the sentence and finds those sentences which have the closest meanings.