A method and apparatus for detecting the occurrence of new ideas in documents
or
communications. The method is comprised of three processes. The first process lexiconizes
all words or symbols in a set of documents. The second process compares all words
in a second set of documents to the words in the lexicon. Words not already in
the lexicon are presented to a user who takes one of two courses of action, 1)
lexiconizes the word, or, 2) declares it a "fad" indicating that the word is to
be further analyzed. The third process measures the spatial and temporal spread
of said fad by searching a third set of documents and computing metrics based on
additional occurrences of said fad, said metrics being used to determine when a
fad has achieved a level of interest denoted as a category. When a category is
detected, a user is notified.