Menu
Home
People
Places
Arts
History
Plants & Animals
Science
Life & Culture
Technology
Reference.org
Latent semantic analysis
open-in-new
<p>Latent semantic analysis (LSA) is a technique in <a href="/facts/Natural_language_processing/1hjMKsSN">natural language processing</a>, in particular <a href="/facts/Distributional_semantics/klV0l6o7">distributional semantics</a>, of analyzing relationships between a set of documents and the terms they contain by producing a set of concepts related to the documents and terms. LSA assumes that words that are close in meaning will occur in similar pieces of text (the <a href="/facts/Distributional_semantics/klV0l6o7">distributional hypothesis</a>). A matrix containing word counts per document (rows represent unique words and columns represent each document) is constructed from a large piece of text and a mathematical technique called <a href="/facts/Singular_value_decomposition/8t3v6wk3">singular value decomposition</a> (SVD) is used to reduce the number of rows while preserving the similarity structure among columns. Documents are then compared by <a href="/facts/Cosine_similarity/vrbNb6wH">cosine similarity</a> between any two columns. Values close to 1 represent very similar documents while values close to 0 represent very dissimilar documents. </p><p>An information retrieval technique using latent semantic structure was patented in 1988 by <a href="/facts/Scott_Deerwester/1dWF3lFT">Scott Deerwester</a>, <a href="/facts/Susan_Dumais/Tf9KGpdj">Susan Dumais</a>, <a href="/facts/George_Furnas/qPnMNLnY">George Furnas</a>, <a href="/facts/Richard_Harshman/S2NwCPJh">Richard Harshman</a>, <a href="/facts/Thomas_Landauer/nV6Kmde0">Thomas Landauer</a>, Karen Lochbaum and Lynn Streeter. In the context of its application to <a href="/facts/Information_retrieval/MFSmpeNw">information retrieval</a>, it is sometimes called latent semantic indexing (LSI). </p>