|
Title:
|
A NEURAL NETWORK FOR DISCOVERY OF RECORD LAYOUTS |
|
Author(s):
|
Azita Bahrami , Ray R. Hashemi , John R. Talburt , Zhebnya Burachevsky |
|
ISBN:
|
978-972-8924-93-5 |
|
Editors:
|
Pedro Isaías, Bebo White and Miguel Baptista Nunes |
|
Year:
|
2009 |
|
Edition:
|
2 |
|
Keywords:
|
Record Layout, Knowledge Discovery, Neural Network, and Record Layout Discovery. |
|
Type:
|
Short Paper |
|
First Page:
|
319 |
|
Last Page:
|
323 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
In order for automated systems to interact with ASCII files, they must be able to discover the layout of records. There are
three formats of record layouts: fixed, delimited, and mixed. The goal of this paper is to discover the record layout of
fixed format. Each record of such files is considered to be a character string in which the fields startings and endings are
unknown and two adjacent fields may or may not be separated by a space. A new neural network was devised to accept a
random sample of the files records as a working set to discover the record layout for the file. The validity of the
methodology was established through using 20 different synthesized files with regard to number of fields, length of fields,
order of fields, content of fields, or any combination of them. The methodologys ability to discover the record layouts
has 95% accuracy. |
|
|
|
|
|
|
