511Big Textual Data Analytics and KnowledgeManagement
methods of knowledge extraction are used to populate ontologies on the
basis of given fact schemes from documents. The main distinction of such
approaches against information extraction is that shifts in the ontology
should be accounted for by the extraction methods without a full reprogram-
ming of entity or relationship patterns.
Analysis on a single document scope with unsupervised methods is used
on the information retrieval level either for document term prole genera-
tion (document vector representation in a term space, as detailed below in
section “Document Vector Models”) or for a representation of a document
in topic space.
Unsupervised methods for information extraction