Browse Prior Art Database

A method to tolerate URL mistyping

IP.com Disclosure Number: IPCOM000015095D
Original Publication Date: 2002-Oct-29
Included in the Prior Art Database: 2003-Jun-20
Document File: 2 page(s) / 57K

Publishing Venue

IBM

Abstract

The other day, I forgot my VM Computer system password. I called the system administrator, who reset it and sent me the new password by Lotus Notes Mail. It was Hwy-1O1 . I tried this password with Upper case H, lower case w and y, hyphen, then the numbers one, zero and one. The system complained that I was typing an invalid password. I thought I had mis-typed and repeated two more times, surprise, I found myself locked out of the system. I walked over to the system administrator, asked him to reset my password one more time, and to write down the new password. The password written on a piece of paper was: W1NNER . I walked back to my office, tried to log on thrice unsuccessfully, got locked out again. My colleague found "cool stuff" at a web site wrote down the url for me: www.trouble.com.0l045559O9567010 . I typed it in. DNS complained that there was no such url. I battled with the browser for 15 minutes, then gave up. If this has happened to you before, you will understand how much time was wasted and productivity lost. The problems described arises due to the fact that visual discrimination errors occur when discriminating between 0 and O; and l and 1. Variable size fonts and the use of multiple fonts, automatic font conversion and human error are the principal cause factors. The different types of errors could be categorised as follows.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 50% of the total text.

Page 1 of 2

A method to tolerate URL mistyping

    The other day, I forgot my VM Computer system password. I called the system administrator, who reset it and sent me the new password by Lotus Notes Mail. It was Hwy-1O1 . I tried this password with Upper case H, lower case w and y, hyphen, then the numbers one, zero and one. The system complained that I was typing an invalid password. I thought I had mis-typed and repeated two more times, surprise, I found myself locked out of the system.

I walked over to the system administrator, asked him to reset my password one more time, and to write down the new password. The password written on a piece of paper was: W1NNER . I walked back to my office, tried to log on thrice unsuccessfully, got locked out again.

My colleague found "cool stuff" at a web site wrote down the url for me: www.trouble.com.0l045559O9567010 . I typed it in. DNS complained that there was no such url. I battled with the browser for 15 minutes, then gave up. If this has happened to you before, you will understand how much time was wasted and productivity lost.

The problems described arises due to the fact that visual discrimination errors occur when discriminating between 0 and O; and l and 1. Variable size fonts and the use of multiple fonts, automatic font conversion and human error are the principal cause factors. The different types of errors could be categorised as follows.

Visual, i.e.; the letters looks similar; This problem may be stated as the problem of visual disambiguation. The solution is to simply cluster all symbols using any well known "shape differentiator" metric. Many are known in the art of signature analysis to expand the search term into a set of terms in the same visual cluster as a particular phrase. When a phrase is typed into a computer system, for computer response, the phrase is expanded into a multiplicity of phrases, each belonging to the same visual cluster as the phrase typed in. Each such phrase is provided as an alternate. The alternates are submitted to the system sequentially, if the essential system behaviour is unchanged. The alternatives may also be displayed to the user, with or without highlighting the differences from the phrase typed in, and with or without sorting on a numerical measure of visual distance. Thus suggestions are provided to the user, as alternatives. The necessary changes to the GUI APIs would also be made. As an example, for the phrase Hwy-1O1 , the phrases Hwy-1Ol, Hwy-101, Hwy-10l, Hwy-lO1, Hwy-lOl, Hwy-l01, Hwy-l0l are suggested.

An efficient method to apply the multiple options, like Boolean minimisation of the bit strings representing the 8 patterns, may or may not be deployed.

Aural, i.e.; the words sounds similar (the word newt could typed in as nute); and Similar to visual disambiguation we could apply aural disambiguation methods for pin pointing the problem. Algorithms for doing this are widely available and are known as soundex. Errors arising due to transposing c...