Abstract The idea of pseudo-classification based on external relevance is introduced and compared with the more usual classifications derived by associative techniques. A general model for an information retrieval system using term classification is described. The derivation of a set of operators, or perturbations, for adjusting pseudo-classifications and preventing their deterioration is given for a particular match function conforming with this model. The use of pseudo-classifications both for the prediction of relevant documents and for the evaluation of retrieval systems with respect to their theoretical optimum is discussed. The concept of the improvability of a retrieval model with respect to its constituent submodels is introduced and elaborated upon. This report is the result of research conducted on classification techniques for informational retrieval systems supported in part by a grant from the Office of Scientific Information Service of the National Science Foundation to the Computer and Information Science Research Center, The Ohio State University.
[1]
John A. Swets,et al.
Effectiveness of information retrieval methods
,
1969
.
[2]
Michael E. Lesk,et al.
Computer Evaluation of Indexing and Text Processing
,
1968,
JACM.
[3]
M. E. Maron,et al.
On Relevance, Probabilistic Indexing and Information Retrieval
,
1960,
JACM.
[4]
David M. Jackson.
A note on a set of functions for information retrieval
,
1969,
Inf. Storage Retr..
[5]
Karen Spärck Jones,et al.
Current approaches to classification and clump-finding at the Cambridge Language Research Unit
,
1967,
Comput. J..
[6]
Lauren B. Doyle.
Is Automatic Classification a Reasonable Application of Statistical Analysis of Text?
,
1965,
JACM.
[7]
A. Resnick,et al.
The consistency of human judgments of relevance
,
1964
.