Rhodes University Department of Computer Science Computer Science Honours Literature Review Principal Investigator: Edison Mukadah Supervisor: Greg Foster Co-Supervisor: Hannah Slay Mobile Interface Design and Predictive Text Algorithm For Improved Text Entry Abstract There has been a substantial growth in interest in mobile text entry over recent years, among both researchers and users. Increasingly mobile devices are being used to perform text-intensive applications, such as text messaging, creating a demand for more efficient and easier to use text entry methods. Unlike for desktop computing, no single standard mobile text entry method has emerged. The diversity of mobile devices makes it unlikely that this will occur in the near future. Thus, mobile text entry remains a very open area of research, providing a fertile environment for the development of innovative text entry methods. A necessary part of the development of a new mobile text entry method is a comparison of its performance with existing ones. A review of current best practice for the qualitative and empirical evaluation of mobile text entry methods is presented, alongside a classification of existing mobile text entry methods and the various ways to optimise text entry. A mobile interface that uses a scroll wheel will then be designed together with a predictive text algorithm by combining or improving the various techniques that will be examined by this literature review for improved text entry. Empirical tests will then be conducted to test the hypothesis that the new interface has a faster text entry input rate compared to the traditional Multi-Tap keypad. 1 Introduction 1.1 The Popularity of Mobile Devices Mobile computing devices popularity has escalated astronomically over the past years. Mobile computing devices include mobile phones, communicators, wireless Personal Digital Assistants (P.D.As) and hybrid phones [Clarkson et al., 2005]. At the start of the last decade there were just over 10 million mobile cellular telephone subscribers around the world. By the beginning of 2001, this figure had grown almost 70 times to over 700 million [isoc., 2007]. The number of mobile phone subscribers worldwide has risen to 2.6 billion as of 2006 and will reach the 4 billion mark by 2010 [itfacts., 2007]. By June 2006, the number of mobile phone subscriptions had exceeded the population size, marking a major mobile milestone of more than a 100% cell phone penetration in 30 countries globally [telecommagazine., 2007]. The ubiquity of mobile devices can be attributed to a myriad of reasons viewing it in perspective of the various stakeholders at play. Owing to advances in robotics and electronics, mobile phone manufacturers have capitalised on the fabrication of the transistor and better automation processes in mobile device manufacture .Cheaper and smaller devices that appeal to various customer needs have thus been made [Forcada, M., 2001]. New technologies — packet data transfer (GPRS), multi-mode terminals, Wireless LAN, 3G, multimedia messaging, voice control, positioning, and Bluetooth have enhanced the Service Providers’ capability to offer data services that are in vogue with the users [Kiljander, H., 2004]. Necessity has irrefutably been the impetus that has driven the mobile device user. The need for a device that links one to the outside world, providing messaging, internet services and gaming among a plethora of other services that are within the digital realm, has earned mobile devices a niche in our everyday life [Kiljander, H., 2004]. The mobile programmer has also become more adept in developing services and applications that can be accessed with mobile handsets, producing optimal solutions in the face of limited processing power and memory capacity of the phones [Kiljander, H., 2004]. Although earnings from voice calls still comprise 80% of worldwide total mobile revenues, operators globally are focussing on data services for increasing their average revenue per user (ARPU). Mobile messaging includes short messaging service (SMS), multimedia messaging service (MMS), mobile e-mail and mobile instant messaging (IM) [portioresearch., 2007]. SMS has been the star of the data services with global SMS revenues expected to reach 67bn USD, driven by 3.7 trillion messages by 2012 [itproportal., 2007].SMS is salient to the Service Provider since short messages can also be used by Service Providers to send binary content such as ring tones or logos, as well as Over-The-Air programming (OTA) or configuration data [wikipedia., 2007]. At a typical length of only 190 bytes (including protocol overhead), more than 350 SMS messages per minute can be transmitted at the same data rate as a usual voice call (9 Kbit/s). SMS has an average global price of 11 cents and maintains a near 90% profit margin, such a service is important for revenue generation [wikipedia., 2007].SMSing is generally cheaper compared to voice calls, making it the subscribers preferred mode of communication [itproportal., 2007]. MMS, IM and mobile e-mail are also fast growing [portioresearch., 2007]. In Japan more email is sent via mobile phones than home PCs [Starner, T., 2004]. Other markets, such as mobile gaming and location-based services, are also largely supported by text entry [Starner, T., 2004]. Emerging companies like Mxit and Yeigo offer far cheaper data services like chatting and SMSing to its subscribers rendering text messaging an even more affordable and convenient means of communication. Their subscription is free and sending a message up to 2000 characters long costs less than 1 South African cent [mxit., 2007][yeigo., 2007]. 1.2 The Problem Common to all mobile devices is that they are small in physical dimensions and weight [Sandnes et al., 2004]. The smaller and lighter a device is, the more mobile it is and thus more desirable to the users [Kiljander, H., 2004]. However, the small size comes at the expense of small displays for visual feedback and less room for interactive controls. The typing rates of today’s text-entry methods using the mobile phone’s key pad are two to five times slower than that expected with a desktop keyboard [Starner, T., 2004] .For example, in 2003 the world’s fastest mobile “texter” typed 29 words per minute [Folkard, C., 2003], more than six times slower than the Guinness record of 192 wpm for the desktop QWERTY keyboard [Young, M., 1998]. 1.3 User Interface Evolution in Some Other Industries - The Lessons to Learn Like mobile phones, automobiles and audio visual equipment can also be classified as smart products. These industries are more mature than the mobile phone industry and the user interface also plays a major role in products of these industries. It is thus imperative for us to look outside the mobile phone domain to learn how user interfaces have evolved over time [Kiljander, H., 2004]. The Mobile industry derives so many invaluable lessons by reviewing the metamorphosis that has taken place in other product domains. The Motor industry’s success mainly lies in the willingness of various competing manufactures to adopt similar standards for the benefit of all [Norman, A., 1988]. If the mobile industry should also mature to the level of the car manufacturing industry, then the influx of proprietary products needs to be quelled. As the number of product features grows larger than physical displays, separate controls and knobs can no longer be introduced for every new feature hence manufacturers are turning to menu based user interfaces [Kiljander, H., 2004]. Use of menus alleviates the problem of spatial arrangement of controls on an interface. However as the menus get more complex by representing a lot of logic, mobile phone manufacturers must also guard against making menus more ostentatious at the expense of usability. 1.4 Towards a permanent Solution Research in text entry is flourishing because of the great need for wireless messaging and mostly because user needs have not currently been met [Starner, T., 2004].” Can entry rates be improved if we arrange the buttons on a keypad in different ways? What is the effect if we use context to guess the next letter or word, or can we apply altogether a new technology?”[MacKenzie, S., Soukoreff, W., 2002].This literature review thus analyzes the important issues that are pertinent to mobile text entry. 2 Current Text Entry Interfaces Taxonomy given in Figure 4 allows us to categorise text input methods seamlessly. Aside from quantitative measures like words per minute or error rates, there are many other ways to characterise mobile text entry methods [Sirisena, A., 2002]. Other important characteristics include whether an input method is operated with one or both hands and the degree of eye-focus it requires. The term focus of attention (FOA) [MacKenzie, S., Soukoreff, W., 2002] refers to the attention demands placed on a user by a text entry method. Voice Input Figure 1: Classification of Mobile Text Entry Methods [Sirisena, A., 2002] 2.1 Key-Based Methods Table 1 : Taxonomy of key-based text entry methods [Wigdor, D., Balakrishnan, R., 2004] The columns indicate how many keys (single or multiple) and subsequently how many presses of those keys are required to enter a single character [Wigdor, D., Balakrishnan, R., 2004]. Of the unpopulated cells, #1, and #2 represent impossible situations and #4 is the regressive case since consecutive multiple presses on multiple keys will be obviously worse than the single key, multi press, Multi-Tap technique. Techniques that fit in cell #5 could be viable but likely difficult to accomplish in practice. Cell #3 suggests a technique that has not been explored for mobile-phone text entry: concurrent chording, where multiple keys are pressed once concurrently to input an unambiguous letter [Wigdor, D., Balakrishnan, R., 2004]. 2.1.1 Telephone Keypad Figure 2: The standard twelve key telephone keypad [MacKenzie, S., Soukoreff, W., 2002] The keypad consists of number keys, 0-9, and two additional keys (* and #).