From: Bob Wiggins, Cura Consortium Ltd UK e-mail -
robert.wiggins@hemscott.net
Date: 24 Nov 1999
Time: 05:01:13
I am to establish and manage a quality assessment system for a project which involves scanning handwritted household survey forms, and then having the data in the viewed images transcribed into a database. This images and the database will eventually go on the Internet.
Acceptance criteria have been pre-established for the images (e.g. completeness of scan, skew etc) and for the transcribed data (e.g. one transcribed entry per name of individual, home address, age, place of birth) and referentai integrity (i.e. the image viewed relates to the data being displayed).Tolerances have been pre-defined e.g. <5% errors for other transcribed data.
About 100 microfilms per week are being scanned amounting to about 60,000 to 70,000 images per week. A sample size of not less than 1,000 images per 50,000 (2%) in a lot has been predefined (but I could change it if necessary). In all the scanning will last for around 10 months at that average weekly rate (to begin with the rate is expected to be slower until the process beds down).
A set of specially designed PCS will give online access to the images and data for the purpose of sampling.
I wish to review what I have been given to ensure that a sensible and practical sampling regime can be established.
For the image sampling I am considering use of procedures in the ANSI/AIIM TR34-1996 (Sampling Procedures for Inspection by Attributes of Images in electronic Image Management (EIM) and Micrographics Systems. This allows for tightened and reduced inspection depending on the results of the sampling.
I am less sure as to how to approach the assessment of the transcribed data. Could I use the same ANSI/AIIM sampling procedures?
I would be interested in using suitable software programs if I could be confident that I shall be impementing the correct sampling regime(s).