The long explanation is this: "LSI or Latent Semantic Indexing is a complicated term with a simple explanation. The main concept behind Latent Semantic Indexing is to discover words and phrases that are related in the context of any document or group of documents. Search engines, Internet Marketing Professionals, and Website Designers often use LSI in their day-to-day activities.
Latent Semantic Indexing is the discovery process for finding related terms and phrases. LSI is a mathematical equation that will fold words into a matrix for analysis that will draw out semantically related terms."
The short one is this: It's pretty good for finding out what words people are searching for in a specified geographic area, i.e. a useful journalism tool.