LSI: LSI, or Latent Semantic Indexing, allows search engines and other search systems to have a mathematical and language-based understanding of the similarities between indexed pages and search terms. This explains why a search may pull up a result that doesn't have a particular keyword within the content but will still be relevant to that search because the content contains similar words.
Because Boolean searches rely on exact matches for a search, which can produce irrelevant results, LSI helps to overcome this by incorporating other, related words into searches. It does this through two different methods: by showing words with similar meanings (synonymy) and by looking for words that have more than one meaning (polysemy).
For example, searching for "pop" on Google could bring up potentially different results depending upon the context. This word brings up information on pop (popular) music and on soda pop, the soft drink. Narrowing the search to "pop music" then eliminates the results that center around soda pop because the word "music" gives the search more context.
LSI is also popular in other specific search areas and for the use of different databases. Government and intelligence organizations are known for having computing systems that use LSI in order to find relevant information and to automatically categorize documents or other information. Other industries, such as education, customer service, and software engineering also use it. Social networking sites may rely on LSI searches for Relationship Discovery purposes. For instance, the Friend Finder feature on Facebook may recommend friends based on whom a user already knows.