There is no shortage of ideas for ways to overcome the limitations of a small screen and the standard numeric phone keypad. These solutions promise faster text entry or quicker navigating to an application, function or service. Zi Corporation is building a portfolio of products to address these issues. We previewed Zi's Decuma handwriting recognition system last year, now it is the turn of Qix and eZiText.

Qix

It is difficult to miss Qix. From the standby screen of a Series 60 phone, such as the Nokia 6600, pressing any key activates the Qix pick list. The principal is simple, one of those why-didn't-I-think-of-that ideas. As keys are pressed Qix looks for matches based on the number or letters associated with that key, so 3 would be matched with 3, d, e, f etc. Using these options Qix searches for records, contacts, applications, help file or web book marks stored on the phone. Qix can theoretically access anything on a phone, but specific features vary with each device implementation.

To illustrate, if a user want to text "John Doe", tapping out 363 pulls up a list of possible matches and, on the test phone, John Doe is listed. From here it is simple list navigation. First to select John Doe, then his mobile phone number from Qix's contact screen and finally "Send Text" from the actions menu.

In this example Qix took 10 clicks to get to the point of entering the text message. Using the standard Series 60 interface it took 12 (via contacts directly from the idle screen) or 14 (through the menu to messaging) clicks. Not a significant saving, but many users could perceive Qix as providing a much shorter route as there is much less context switching.

In some cases Qix is not faster - for example selecting the calendar application. Through the standard interface this takes 4 clicks while in Qix it takes 7 clicks. On the other hand for URLs it is much faster. The Series 60 WAP browser is 2 clicks away from the idle screen (using the feature to access the first 9 menu items with the number keys 1 to 9). Finding a URL is then a case of scrolling through the bookmark list. With Qix most URL bookmarks will be available in far fewer clicks, Tube News is 3 clicks while Yell.com (UK Yellow Pages) is 4 clicks.

Qix also has a learning feature which means that commonly selected items get priority when searching. This assists in reducing the clicks needed for common searches.

Qix presents one real challenge, or perhaps, to be fair, making content accessible through Qix presents content designers with one real challenge. Qix is excellent where the name of the item required is known or at least obvious, "ring tone", "football results" and so forth. Where the name of the content is unknown the usefulness of Qix as a discovery mechanism is less clear cut. If an operator wants to add an on-line bus timetable service then will a users look for Bus or something more genetic like travel or perhaps the name of the local bus company? Even intelligent naming may not be a solution, users still need to know they might find services they do not yet know about. Qix therefore may not be the best solution for service discovery and another mechanism may need to be employed, however clever naming may be.

Zi Corporation is currently trialing Qix with Virgin in the UK (press release link). It will certainly be interesting to see what results this yields on Qix's ability to help user find and activate Virgin's services.

eZiText

Having used Qix to get to the point where the message to John can be entered Zi's second Series 60 tool now comes into play. eZiText is an extended predictive text application. In addition to the familiar prediction, based on all the keys pressed, eZiText predicts complete words based on partially complete entries and can even predict the next word in a sentence.

In a T9 predictive text system typing "station" would required 7 key presses, 7828466. Using eZiText selecting 78274 results in the prediction of the word "station" , as shown below, from just 5 keys. eZiText also has a learning capability which biases prediction towards a users vocabulary, so the next time "station" is entered it is predicted on just 4 letters.

It is not just the users vocabulary that eZiText can learn, it also learns its user's grammar, how they strings those words together. It is a couple of hours later, at the rail station there is no sign of John. So a new text gets send. This time eZiText's next word prediction speeds up the text entry process. As "the" is finished eZiText predicts the next word as "rail" and, after selecting rail, the next word as "station".

As eZiText is used it continues to build information on the user's vocabulary and grammar. As a result it will get better at predicting the more it is used.

Ironically the one thing eZiText is designed to overcome is also its biggest draw back. For those familiar with the UTI keypad, people who can watch the screen as they type, eZiText will undoubtedly be a faster and more efficient system for text entry than systems such as T9. However for any user who still has to hunt out the letters on a keypad the advantages of eKeyStick, might be a much better solution.

Qix and eZiText are both well thought out and executed solutions to the challenges of navigation and text entry on today's smartphones. For the majority of users of smartphone (or indeed virtually any phone) they will offer a faster more effective way of entering text or finding contacts, services or phone functions. Neither Qix or eZiText have yet been implemented as part of any Symbian OS based phone. As Zi's licensees include Nokia, Sony Ericsson, Samsung and Lenovo implementations can only be a matter of time.