The human race is spending more and more time inputting information into electronic devices of all types. So it is important that we find easier, faster and more accurate ways of transferring the information from our heads to our electronic beasts.
Using a keyboard has been the way to do this since the beginning of the computer age. More recently, voice recognition has taken off but still accounts for a small percentage of the information entered. Video cameras and audio recorders now account for most of the new content but do not displace much of the text content being produced.
Gestures are the latest input method but are really only used for controlling the device not for input, although we might see some simple gestures for: hello, goodbye, yes, no, etc. Thought transference is in the labs but it will some considerable time before I can think this sentence and then see it on the screen.
All of this suggests that typing is going to remain a major method of input to electronic devices for years to come. To make matters worse, devices are getting smaller so that a full size QWERTY keyboard becomes impractical. Tablets and smartphones with touch screens do not even give any tactile feedback, although this may change in the next few years.
So, as typing is going to remain and the physical interface is not going to improve how can we make it easier, faster and more accurate? Predictive text has been around, especially for 12 key telephone input, for some years but has been of limited use because the predictions were often not right and just got in the way.
KeyPoint Technologies (KPT) have extended the concept of predictive text technology with new methods and greater intelligence; to such a degree that to type the 1700 odd characters above should require less than 500 key presses.
With that increase in speed we should all become more productive and the use of on screen keyboards would become an acceptable input device for more than just a quick note. Hence helping to narrow and bridge the gap, what KPT describe as ‘the chasm of inutility’, between the desires of the users and capabilities of the input devices.
To promote the technology KPT has announced the Open Adaptxt engine; this is an open source version of the engine freely available for a variety of mobile platforms.
What does the engine do that makes it so much more productive than standard predictive text? There are a collection of techniques which include:
- Intelligent prediction. As you type it will predict the word you are typing not just by the letters you typed but also by the context of the sentence and your personal word usage. This greatly increase the chance that the word you are trying to type will be in the prediction list and will require fewer characters to be typed. Further it will predict the next word before you even start typing; it can also predict whole phrases when that would be helpful.
- Intelligent error processing. If you type a word that is not recognised it will provide a list of alternatives. If a QWERTY keyboard is being used these alternatives will include those that would occur because of typical typing errors; for example letters typed in the wrong order, or adjacent letters (‘a’ instead of ‘s’). It can also automatically correct the word when you press space and will deal with capitalisation of proper names and acronyms.
There are further methods for specific issues that complete the engine.
Adaptxt is being marketed as a general purpose solution that should benefit all users by speeding up text entry from a keyboard. However, it should be of particular interest to users with limited dexterity who type slowly and are more likely to hit the wrong key. In fact it was originally developed to help a relative, who had lost an arm, to be able to type more easily.
I am keen to see examples of Adaptxt being built-in to applications and will write about them and hopefully with them soon.