A method, program and information processing system filters and secures
data (security sensitive words-characters-data objects) in a source
document. The adaptive filter uses a compilation of additional data
(typically networked) and identifies the sensitive words/objects in the
compilation of additional data, and retrieves contextual, semiotic and
taxonomic words/objects from the compilation related to the sensitive
words/objects. The resulting compiled filter is used to extract sensitive
words/objects and retrieved data (words/objects) from the source document
to obtain extracted data and remainder data therefrom. Contextual words,
related to the security sensitive words/objects, are based upon
statistical analysis of the additional data compilation. Semiotic words
related words are synonyms, antonyms, and pseudonyms, syntactics relative
to the target words and retrieved words, and pragmatics relative to the
sensitive words and retrieved words.