Checking for Accuracy

Input: Data Accuracy

OCR software

O ptical C haracter R ecognition: This software takes a scanned image and converts the characters in the image into computer characters. The document can now be edited with a word processor. This is a very tricky process. Documents must be carefully checked for wrong conversions. If the original print was not very crisp and clean, errors are very likely. Manually checking for proper translation is necessary. These programs are getting really good if they have a clear scan to work with.

A famous slogan in computing sums up the importance of accurate data:

GIGO = Garbage In, Garbage Out

Conclusions are no better than the data they are based on.

A major task for any program that accepts data is to try to guarantee the accuracy of the input. Some kinds of errors cannot be caught but many of the most common kinds of mistakes can be spotted by a well-designed program.

A program should attempt to do the following:

1.	test data type and format	ex.	2/a/96 is not a date
		ex.	If a phone number should have exactly 10 digits with the area code, then 555-123 is not acceptable.
2.	test data reasonableness	ex.	231 should not be a person's age, at least not since the time of Noah and the Flood
		ex.	A sale of $50,000 worth of chewing gum at the corner market is probably missing a decimal point somewhere!
3.	test data consistency	ex.	A man's death date should be later than his birth date!
		ex.	The sum of the monthly paychecks should be the same as the total paid for the year.
4.	test for transcription and transposition errors	ex.	Typing 7754 instead of 7154 is a transcription error, typing the wrong character.
		ex.	Typing 7754 instead of 7745 is a transposition error, interchanging two correct characters. Both are very hard to check for!

Дата добавления: 2014-12-23