Need a Senior Thesis - Design a Computer Modeled Typo Generator
Are you a computer programmer in need of a senior thesis for the upcomming year? I know most people probably already picked theirs long ago and are going to write some great program for a Palm Pilot to control your Tivo from zimbabwe. What use is that? Maybe to Britney Spears of Angelina Jolie, but not the rest of civilization.
Why not do something useful?
If I had a lot of time on my hands and a lot of spare people around me (such as a college campus), and I could find a professor willing to use senior funding to pay said students, or had enough money to tell my professor to piss off and do it anyway than this is what I would do.
I would make a program based on data collected from at least 500 participants typing a representative list of words (probably at least 500 different words and phrases) over and over again. From the lists these people create I would then run some statistical analysis on the results. The data/results could be reduced to split up left and right hand as well as those letters close to one another. This would eliminate outliers like going to type an F when your pinky hits Q, or your right pinky hits [. From this one could then generate an easy alrogithm in which they looked at consectutive letters begin typed as well as location on the keyboard and give probabilities for mistyping. The results could easily be check against a new random sample and random list of words, and if the results were drasitcally different you could group the new results with the old results to get a new algorithm.
What could you use this for? For someone wanting to buy domain names that would generate a lot of typo/type-in traffic. Great internet tool.
If someone decides to do this let me know, I want first crack at seeing how it works with generating some good typos. I realize that more goes into finding a domain name with traffic and click through potential but would something like this be nice for the domaining community?