Intro to Predictive Coding: Overview & Interpretation of Terminology June 2014 | Page 26

Glossary
Language modeling—computing a model of the
relationships among words in a collection.
Language modeling is used in speech recognition to
predict what the next word will be based on the
pattern of preceding words. Language modeling is
used in information retrieval and predictive coding
to represent the meaning of words in the context of
other words in a document or paragraph.

Latent Semantic Analysis—(LSA) a statistical
method for finding the underlying dimensions of
correlated terms. For example, words like law,
lawyer, attorney, lawsuit, etc.
All share some meaning. The presence of any one
of them in a document could be recognized as
indicating something consistent about the topic of
the document. Latent Semantic Analysis uses statistics to allow the system to exploit these correlations
for concept searching and clustering.