Level: Introductory Christopher Cring (ctcring@us.ibm.com), Software Engineer, IBM Systems Technology Group
28 Dec 2004 Differentiating between the number 0 and the uppercase letter O or between the number 1 and the uppercase letter I in search engines is a major cause of the no match found error message. Author Christopher Cring digs up a simple, pre-existing method designed to quickly and easily solve this common search problem of characters that cannot be easily identified.
In working with customers for z/OS Software for many years, my experience is that when customers use currently available search tools to determine if an encountered software or hardware problem has a known fix, they often receive a "no match found" message.
While sometimes no match exists for these diligent searchers, the problem might be one or more of the characters in their search string. If a character is not easy to identify, it can spawn a no match scenario.
In this concise guide, I'll further identify the reasons this problem occurs and the history behind its pervasive nature, plus serve up a relatively simple solution.
More on the problem
I want to focus on one of the more common scenarios that I continue to see. It occurs when a user searches based on an error message that contains any of the following characters:
-
The uppercase alpha character O
-
The number zero (0)
-
The uppercase alpha character I
-
The number one (1)
Since these characters are difficult to differentiate, a user can type the wrong character during the search and get a no match found result.
Why is the problem so pervasive today?
Technology availability and usage trends in the last few years have yielded three changes that make this problem so common today.
History is being ignored today
In the past, when a character string was used to identify any type of information (such as an address, confirmation number, building aisle location, name, error message) the uppercase letters O and I were often avoided to bypass the problem.
For example, in my building, the numbering scheme for aisles skips the letters O and I. The hardware layout for most logic cards excludes these letters as did old error messages. I understand that even Carnegie Hall skips aisle I in its seating arrangement.
This practice was developed to avoid confusion; but it has also been ignored to the point that it could be considered an abandoned practice.
More fonts are in use now
First, have you noticed how many fonts are available lately (including the free-to-download versions)? Couple that with the unfortunate fact that many people choose their fonts based on their attractiveness rather than practicality, and you can see how this further compromises overall clarity.
Thousands of fonts are now available and it is easy to select and customize your text layout. With many of these fonts, it is practically impossible to distinguish between an O and 0 or between 1 and I. (At least with more traditional fonts, the zero had a small dot in the center -- or it was narrow while the capital O was wide.
The explosion of searches
With the Internet, people take for granted that they can search for everything and expect to find a match. Couple that with the growth of Internet users (not just the numbers, but in the range of people with different experience levels), the increase in online transactions, and the population explosion in disparate search engines and search-engine technologies, and it's no wonder that more users get no matches found messages.
Try this simple solution
This problem is not going away; the misleading characters are not going to change. So what's the simplest way to head off the no match scenario?
Include a variant alphanumeric search. Historically, the most common problems are with zeros and ones. IBM patent number 6,459,810 describes a "method and apparatus for forming variant search strings." This method delivers the combinations of 0s and Os and Is and 1s. The search method does not exclude this common mistake of character identification because all combinations are used for the search input.
You can use the search either as an option if no match is found or as a from-the-start variant search. If more than one match is found (and in my experience, this is a rare occurrence!), you can offer the selection of choices or isolate the results with another level of information based on specific-user or specific-request usage.
The method is simple. It includes the combinations and is transparent to the user. In the following example, the message string MSGISOB85 has the mix of characters mentioned so when you select Variant Search it generates a search input of the four combinations.
Figure 1. Variant search example
The following common uses of character strings are those where you're most likely to encounter the problem:
- Airline confirmation numbers
- Order numbers
- Car VIN numbers
- Server numbers
- Computer error messages
- Serial numbers
- Registration numbers
- Insurance ID numbers
- Building location aisles and rows
Couldn't be more pervasive than this list, right?
Is that I/O or the old 1/2?
I've described an existing, quick, and painless method and means to let you implement variant search strings to search engine technology, a design destined to eradicate (or at least make less frustrating) the most common user-input variations that generate the no match found message. For many users, this means that they will find their query answer in the first search (albeit, they may have to parse a returned list of possible matches).
Resources
- In the tongue-in-cheek article, "How not to make your site accessible," pick up tips to make a Web site harder to use (developerWorks, October 2002).
- Discover a host of considerations for designing a portal, including the ability to search it in "Usability for component-based portals" (developerWorks, June 2002).
- In "Why language isn't easy, " explore the language gap and help combat errors in Web design (developerWorks, September 2002).
- Visit the IBM Ease of Use site to address the challenge of creating great user experiences through the discipline of user engineering, supported by design guidelines, tools, services and other relevant materials.
- Visit these valuable resources on developerWorks:
About the author  | | | Christopher Cring has more than 28 years experience working at IBM in many aspects of the computer software area, including designing and programming for the following projects: defect correlation and analysis for logic circuit testing; robotics systems development; mainframe processor controller development; 3174 control unit micro-code development; mainframe performance evaluation; and z/OS language environment development. He is also the inventor of patents relating to character identification, patents that include Patent No. US 6,459,810 (METHOD AND APPARATUS FOR FORMING VARIANT SEARCH STRINGS) and a pending patent (METHOD, SYSTEM, AND STORAGE MEDIUM FOR PROVIDING INSTANT TEXT CHARACTER IDENTIFICATION). You can contact Christopher at ctcring@us.ibm.com. |
Rate this page
| 
