3.4.1: The Model

Representation of a document or a query: vector Rn

This means that methods from the vector space can be used like:

ZUGANG3.4.1.1: Vector Space Model of IR:

The document vector could in general as well be defined directly by real valued attributes: Ak:D->R For simplicity reasons and to be consistent with most of the literature we will assume for the future wi,k=Ak(di) :

The similarity measure can be used to compare document and query vectors i. e. find the most similar documents for a query.


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