Latent Semantic Analysis with scikit-learn

Searching for keywords has become part of our information lives. We use keyword searches to find articles. When we type in words, we do not mean to retrieve articles with the exact words but the concepts. For example, when we search for “Tesla” and “electric car,” we want to get articles about the company Tesla and its electric cars, rather than historical information about the scientist Nicola Tesla. This is called a “semantic search,” which means interpreting a word for its intent and contextual meaning. We use keywords as cues to get the concept and the documents that are associated with that concept, rather than using the keywords at their face values.

Latent Semantic Analysis (LSA) is a milestone solution that was developed in the 1990s. When words are entered for a keyword search, LSA finds the underlying topics that the words in documents associate with and retrieves those documents. LSA...