Table of Contents

Search

  1. Preface
  2. Introduction
  3. The Design Issues
  4. Standard Population Choices
  5. Parsing, Standardization and Cleaning
  6. Customer Identification Systems
  7. Fraud and Intelligence Systems
  8. Marketing Systems
  9. Simple Search
  10. Composite Keys
  11. Summary

Application and Database Design Guide

Application and Database Design Guide

Code Pages, Character Sets and other Encoding Issues

Code Pages, Character Sets and other Encoding Issues

This subject is not for the faint hearted; nothing in this area is as simple as we would all like it to be. Massive advances in character display technology, standards, tools and protocols have occurred over time. However the globalization of systems and databases has increased the frequency with which these standards are being mixed together.
Some examples of real world problems will suffice to raise the awareness of important issues.
It is true that accents on characters make them sound different but in most countries the error rate and variation in the use of accented characters is very high.
It is true that today’s keyboard and code pages support accented forms, however many users still key the countries old conventions where two adjacent characters are used instead, or simply leave the special characters out.
We have found that databases in some countries suffer from non-standard versions of the local codepage standard. Fixing this still means that old data has different characters.
Moving data between tools sometimes converts characters without your knowledge. Some tools convert from EBCDIC to BCD and then back losing information. Some processes convert ASCII to EBCDIC and back inconsistently.
One terminal in a network set up with the wrong Code Page can cause database maintenance errors. In a site in Chile we saw a large database where some terminals were using a USA English code page, others with a European Spanish code page, and others with a Latin America code page. This led to users continuously correcting and re-correcting the accented characters in a name and still each user was unable to see a correct form of the data. The net result is a very corrupt customer file.
DBCS encoding for Japan and China suffers from having several standards. This leads to increased complexity when sharing or comparing data from different sources.
The fact that people sharing data around the world can not read the same character sets as each other leads to names and addresses necessarily being recorded twice, once in a local form and also in an international form. In some cases this leads to the wrong form being used in the two fields, or even unrelated names being used in each field.
There are mixed protocols for handling foreign words, such as in Israel where sometimes Hebrew phonetic forms for a foreign name are used rather than the original Roman characters, or in Japan sometimes using Romanji and at other times using Katakana for a foreign word.
Different code pages and data entry conventions involving foreign data increase the complexity and error in identity data and this in turn increases the complexity of the algorithms needed to overcome the error and variation.

0 COMMENTS

We’d like to hear from you!