ISR10
Scientific Report No. ISR-10 Information Storage and Retrieval
Evaluation of Document Retrieval Systems
chapter
Joseph John Rocchio
Harvard University
Gerard Salton
Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government.
5-1
CHAPTER 5
EVALUATIOI[OCRerr] OF DOCITh[OCRerr][OCRerr] RETRIEVAL SYS[OCRerr]S
1. The General Problem
An operationally effective automatic document retrieval
system must satisfy the requirements of a diverse class of users. The
functional model discussed in this thesis considers only a subset of
these requirements, principally those related to the methodological
aspects of the system operation. The ability to establish system
requirements is directly related to the notion of performance evalua-
tion, and in recent years the investigation of evaluation measures for
document retrieval systems has rec[OCRerr]ed considerable attention (refer-
ences 1 through 8). The purpose of this chapter is to consider in
general the problems associated with the design of evaluation
experiments and the collection of performance statistics, and to
discuss in particular the implications of the system model on these
problems.
The general aim of a document retrieval system is to
mechanize the deduction of the attribute 11relevance1,' which is a
subjective relation between the state of a user's informatiQn need and
the information content 6f the "docume[OCRerr]nts'[OCRerr] in some collection, as
perceived' by the user. It is clear that since the' system operates
The references cited are only representative of the literature
on this topic.