Login

Publications  •  Project Statistics

Glossary  •  Schools  •  Disciplines
People Search: 
   
Title/Abstract Search: 

Dissertation Information for Donald R. King

NAME:
- Donald R. King
- (Alias) Donald Ross King

DEGREE:
- Ph.D.

DISCIPLINE:
- Library and Information Science

SCHOOL:
- Rutgers University (USA) (1971)

ADVISORS:
- Thomas H. Mott Jr.

COMMITTEE MEMBERS:
- Susan Artandi
- Henry Voos
- Daniel H. Lehrman

MPACT Status: Fully Complete

Title: AN INVERTED FILE STRUCTURE FOR AN INTERACTIVE DOCUMENT RETRIEVAL SYSTEM

Abstract:
In an automated document retrieval system, a computer may be used to store and manipulate the index to the document collection. The form or structure of the index has a great influence on the efficiency of the operation of the system and on the amount of space required to store the index. The standard form of the inverted file utilizes digital document numbers for the storage of the document numbers. This study proposes the use of binary vectors for the storage of the document number references. In the case of the inverted index file, the length of the vector for each index term is equal to the expected number of documents in the system. The position of the vector element corresponds to a document number such as that ones are stored in the positions corresponding to the numbers of the documents which contain information about the term; zeroes are stored as other elements. Two model systems were prepared to contrast the use of the binary vectors with the standard inverted file. The following results or conclusions are drawn from the study:
1. It is feasible to apply the binary vector in a working document retrieval system
2. Predictive equations were developed for computing
a. the ration of the binary vector file memory size to the standard inverted file size, and
b. the average number of physical records per index term in the standard file
3. The size of the binary vector file can be reduced significantly by the use of a compression algorithm, which is described in detail. In the case of the model systems, the binary vector file size was reduced from 2.7 to .41 times the size of the standard inverted file.
4. In forty searches of the model systems, search times for the binary vector file ranged from .79 seconds to 9.72 seconds. The corresponding range for the standard file was .15 seconds to 202.98 seconds.
5. Analysis of the search times and parameters describing the individual queries shows that
a. Search times for the binary vector file and the number of index terms in the queries have a correlation coefficient of .96 with a significance of .001.
b. Search times for the standard inverted file and the number of postings in the terms in the queries have a correlation coefficient of .98 with a significance level of .001.
6. The narrower range of search terms for the binary vector file and the greater case of predicting search times for the binary vector file provide an advantage over the standard inverted file as the basis for on-line interactive document retrieval systems.

MPACT Scores for Donald R. King

A = 0
C = 2
A+C = 2
T = 0
G = 0
W = 0
TD = 0
TA = 0
calculated 2008-01-31 06:09:19

Advisors and Advisees Graph

generating graph, please reload

Students under Donald R. King

ADVISEES:
- None

COMMITTEESHIPS:
- Joy Kaiser Moll - Rutgers University (1975)
- Elizabeth Futas - Rutgers University (1980)