Principal Investigator Supervisor Co-Supervisor Edison Mukadah Proff Greg Foster Dr Hannah Slay.

Principal Investigator Supervisor Co-Supervisor

Edison Mukadah Proff Greg Foster Dr Hannah Slay

Background

• Exponential increase of mobile messaging

SMS MMS Mobile email Chatting …mxit

Problem Statement

• Mobile texting is slow

Guinness world record:192 wpm vs 29 wpm

Problem Statement Cont…

mini EV-K130Xun Chi 138F88 wrist watch Mobile Phone

•Miniaturisation of Mobile Device Faceplate

Problem in fitting all the buttons

Objective of Project

• Design a New Interface and predictive Algorithm

For improved text entry speed

To cater for miniaturised devices

To use a scroll wheel

Why choose a scroll wheel

• Saves Space

• To use agility of the thumb

• More feasible alternative to: Voice input

Stylus based input

Virtual Keyboards

Approach to Project

Literature Survey

Voice Input

Developing The Interfaces

Nokia 5300 Nokia 7380New Interface

Design and Implementation

• 5 Interfaces developed in flash

“T9” (Nokia 5300)

“Multi-Tap” (Nokia 5300)

“scroll Wheel Non-Predictive” (Nokia 7380)

“Scroll Wheel Predictive” (Nokia 7380)

Scroll Wheel Enhanced Prediction (New Algorithm)

Why Use Simulator

• No current phones with hardware capability

• To standardise experiment

User Testing

• To test interfaces in terms of :

Text entry accuracy

Text entry speed

Set-up

Laptop ComputerConnection Cables

Mouse

Speaker

Speaker

LCD Monitor

Procedure

• 10 users recruited

• 5 Interfaces tested

• 5 phrases per interface

Readability Scores

Phrases Used

Phrase Flesch Reading Ease as %

Do you like bad people? 100

Our cause will be for all. 100

I left you side cover. 100

Did you like the rain? 100

I like to ask young people a question. 92.9

Average Score 98.6

ResultsText Entry Speed

8.09

11.85

2.25

4.09

6.04

0

2

4

6

8

10

12

14

Traditional Multi-tap

TraditionalPredictive

Scroll WheelNon-Predictive

Scroll WheelPredictive

Scroll WheelEnhancedPrediction

Interface

Wo

rds

per

Min

ute

(wp

m)

Results ContAverage Error Rates

4.89

2.85

9.06

6.6947.34

0

1

2

3

4

5

6

7

8

9

10

Traditional Multi-tap

TraditionalPredictive

Scroll WheelNon-Predictive

Scroll WheelPredictive

Scroll WheelEnhancedPrediction

Interface

Err

or

Rate

(%

)

Interpretations Speed

• Scroll wheel interfaces generally slower

• “T9” gave the fastest text entry speed

• New Interface gave a 50% speed up compared to Nokia 7300

predictive algorithm

• New Interface gave a 300 % speed up compared to Nokia 7300

non-predictive algorithm

Interpretations

Accuracy

• Scroll wheel interfaces generally more error prone

• “T9” gave lowest error rate

• New Interface is slightly more error prone compared to Nokia 7300

predictive algorithm

• Nokia 7300 non-predictive was more difficult to use hence it had more errors.

Future Work

• Use a real hardware device

• Implement Contextual prediction

• Use a pressure sensing touch-pad to implement scrolling

Demo:

ANYONE WITH A