3.4.3: Term Weighting

Document vectors have been invented to give terms weights according to their importance for a document. Issues are:


Intellectual methods are expensive and not very reliable.

Two kinds of influence can be distinguished in weighting methods: local or context sensitive influences and global or context insensitive influences

ZUGANG3.4.3.1: Local Weighting Strategies

ZUGANG3.4.3.2: Word Frequencies in Language

ZUGANG3.4.3.3: Global Weighting Strategies

Global and local strategies can be combined:


© 1998 / HTML-Version 17. 11. 1998: R. Ferber