Frontend for a RETAIN Function

1

• Front-end for a RETAIN Function
• Senior Project
• Melissa Chávez
• Client IBM
• Advisor Prof. Fernandes

Abstract IBM is trying to provide their support
group with a better user interface in their
primary database, RETAIN. This primary tool was
created 30 years ago therefore, it has a command
line user interface (with complicated and
unintuitive commands), based on technology at the
time. When problems reported by customers are
caused by an error in IBM's operating system
code, an APAR (Authorized Program Analysis
Report) is created so the problem can be fixed.
My primary goal was the design and implementation
of a front-end for a portion of RETAIN, including
creating an APAR. A second goal was to implement
a spell checker, which is used to identify
misspelled acronyms, in a detailed problem
description that is written in one of the
windows. I implemented the spell checker using
the Approximate String Matching Algorithm which
compares the misspelled word with the word in the
dictionary in a character by character basis it
gives the comparison a weight. Using the lowest
weights, it makes the suggestions for the
misspelled word.

• RETAIN
• RETAIN is a legacy database used to record of
  IBMs Operating Systems Problems by corporate
  customers. IBM does not want to throw this
  database away and recreate it because it contains
  all problems worldwide and it has been very
  reliable and available.

• GUI Requirements
• Creates APAR
• Limits user not to enter unwanted
• information
• -Obliges user to enter required
• information

Spell Checker Step 1 Construct a list of all
different words in the input field. Step
2 Look up each word in the list in the
dictionary. ? If word exists, delete from
list. ? If word is not found, leave it in
the list. Step 3 Use approximate string
matching algorithm to find suggestions for
misspelled words.

• Spell Checker Requirements
• -Used when creating APAR
• English only
• Regular and custom dictionaries
• Suggests words

The weight of each vertex is Weight of (i, j)
min vertex (i-1, j) 25 (deletion)
vertex (i, j-1) 25 (insertion) vertex
(i-1, j-1) 0 (exact match)
20 (substitution)
Approximate String Matching Algorithm The string
matching algorithm works by calculating the cost
of transforming the misspelled word into a word
in the dictionary. This can be achieved by
inserting, deleting or substituting a character
from the misspelled word. Each of these has a
different cost. Then, the misspelled word is
compared to all the words in the dictionary and
the words that will be suggested are the ones
that have the lowest cost of transformation.
? H O U S E
? H U S E
Horizontal lines insertion Vertical lines
deletion Diagonal lines a substitution
The path with the minimum cost is highlighted
Results Misspelled word ? Suggestions PAR ?
APAR MS ? QMS MS ? PMS PSR ? PMR PSR ?
PTR PROCC ? PROGCK ICOROUT ? INCORROUT