As it is well known, at present, the most frequently used input systems are the predictive input systems, such as “T9™” and “iTap™”, as well as the method of linear search of the characters assigned to a key by pressing the key several times - the so-called “Multitap” method. However, the apparent disadvantages of these methods resulted in the need for creation of a more flexible and convenient method - the “ARONETIS™” method.

In comparison with the standard “МultiTap” input method, the “ARONETIS™” method ensures significant, mathematically provable, benefit in reducing an average number of keypresses per input of one character.

The average number of keypresses per input of one character with the “ARONETIS™” method, counting ALL keypresses, (i.e. input of letters and punctuation marks, case shifting, layout shifting, and switching between input languages) is in the range of 1.6 to 2.3 keypresses per character for different languages. The maximum efficiency of the novel method for the eight major European languages is given in table.

The systems “ARONETIS™” and “T9™” are comparable in the average number of keypresses for input of one character if we consider all keypresses, including those for inserting punctuation marks or shifting the case. At the same time, as compared with the “T9™”-like systems, the “ARONETIS™” will have the following advantages:

• it is possible to input words, which are not stored in the vocabulary, such as personal names, addresses, abbreviations, technical terms, passwords, etc;
• it is possible to input a message comprising words from different languages (any number of languages) at the same time;
• it is possible to enter various special symbols (for example mathematical symbols or transcription symbols);
• it is easy to add layouts for new languages and it is easy to switch between layouts for different languages;
• it is possible to use for input any of the supported languages, independently of the characters marked on the keys of the mobile phone;
• the system is not sensitive to spelling errors;
• it is possible to input text based on non-Roman alphabets with Roman letters (transliteration) or automatic conversion of Cyrillic text in transliterated text.

• implementation of the “ARONETIS™” on the system level of a mobile phone will require several times less memory and computational resource as compared to the “T9™”-like systems;
• it will be possible to input any Unicode symbols;
• it will be possible to input messages in any language with an alphabet of up to 50 letters;
• it will be possible to create special layouts for input of strong passwords;
• it will be possible to ensure input of text on almost any programmable device having a display and a reduced keyboard (keypad) with at least 6 keys, such as ATM, TV set with a remote control, etc.  

ARONETIS™ - How it works? A little of theory!

The initial goal of this project was to create a method of input of text in different languages independent of the characters marked on the keypad of the mobile device. It was necessary to ensure that the application would be independent of the interface languages pre-installed in the mobile phone, as well as to provide an efficiency of text input comparable to the efficiency of predictive input methods (such as “Т9™”, “iTap™”) and higher than that of the method of linear search of the characters associated with the pressed key- the so-called “Multitap” method. However, the results of the project exceeded our expectations. The process of project development and the logic of operation of “ARONETIS™” are described below.

Keys on the Screen

As it was already mentioned above, the main idea of “ARONETIS” was to depart from the characters on the keypad by using the characters displayed on the screen of the mobile phone. During project development, various alternative embodiments have been considered.

The simplest solution may be realized so that a mobile phone keypad is displayed on the screen and, for each key on the keypad, only those characters of the used language, which may be inputted by this key, are displayed on the key.

The positive effect achieved with this solution will be only in that the characters on the keypad will be somewhat larger (due to the fact that characters of the currently non-used language(s) are not displayed) and in that there will be a possibility to input text in languages, for which the characters are not marked on the keypad.

However, such solution does not ensure any significant advantage from the point of view of user convenience, because the input principle itself remains the same “Multitap”. At that, the keypad image will occupy almost the whole screen and, consequently, the user will only be able to see one or two lines of the input text on the screen. Therefore, this solution may not be considered user-friendly.

Another solution is to display on each key of the virtual keypad only one character at any given time point. This will enable reducing the key image size while maintaining the character image size. In this case the screen will have a significantly larger space, which is free i.e. not occupied by the keypad. However, in this case another problem arises – that is how we can ensure that only one character is displayed on each key if the number of characters in a language is several times that of keys on a mobile device keypad.

The History of Characters in Telephone Communication

In order to solve the above-mentioned problem (how to display one character on a key) we will now take a look at the history and discuss why the Roman letters appeared on the phone keys and why, for example, on the key “ 1” there is no character and why the keys “ 7” and “ 9” each have four Roman characters assigned while the other keys bear only three keys each.

These questions are extensively covered in the e-book “Mandership” by Artemy Lebedev. We will not now retell this whole e-book on the history of telephone communication but we will highlight some points, which may be of interest.

As the telephone was invented by Alexander Graham Bell in the USA, the telephone was originally designed to serve the needs of the Americans and other nations had to adapt American innovations to their needs.

Initially, in the telephone communication, the Americans started to use alphabet letters solely for the purpose to facilitate learning of multi-digit telephone numbers. At that, because of technical reasons, not a single telephone number in the USA started with “one”. Again, historically, “zero” was always used to call the operator. Therefore, a telephone number in the USA could start with any digit except “ 1” and “ 0”.

...The first telephone sets with rotary dials had only numerals on them. In the 1920s letters standing beside numerals came into use. In the USA on a telephone dial three letters of the English alphabet used to stand beside each numeral (but 1 and 0). The letters Q and Z were not in use owing to their visual similarity to  0 and 2.

The English used to place the letter O near zero—they looked the same anyway. The Swedish and the German used one letter per each numeral: A, B, C, D, E, F, G, H, J, K, thus taking up all of the ten numerals. The letter I wasn’t used to avoid its confusion with 1.

In the USSR there were also ten letters standing beside numerals on a dial: А, Б, В, Г, Д, Е, Ж, И, К, Л. The letter З wasn’t used not to be confused with three. In today’s  Russia no telephones with Russian letters are produced. Many European telephones don’t have any letters at all. All American telephones have letters standing beside numerals.

All modern mobile phones (no exceptions) sport the full Latin alphabet (some models also have the Russian one). Here’s where the problem with two ways of using the letters starts. If you are driving on a highway and see the words “ How am I driving? Call 1-800-EAT-SHIT” on a huge truck, it means you are suggested to call 1-800-328-7448. But when you type an SMS, letters work in another mode: to choose a letter, pushing the relevant button once is sometimes not enough—you’ll have to push it several times to get the right letter.

As the process of using letters to designate numbers and the process of the using mobile phone keypad to compose text messages differ significantly, it is evident that the system useful to designate numbers is not efficient for text input. From the above discussion it obvious that the “Multitap” input method may be somewhat improved by simple re-assigning Roman characters to all numeric keys and not only to “2”-“ 9” keys. However, the advantage in decreasing of an average key-stroke number per input of a character will not be so significant to justify change of the historical layout of Roman characters on the keypad, which will inevitably bring confusion and mess in the use of mobile phones for at least several years.

Character frequency

The issue of improving the system of text input with a reduced keyboard i.e. with a keypad having only 12 keys appeared long before the era of mobile phones. Already in the 80s, they carried out studies on efficient distribution of characters on the telephone keypad for transmission of text messages through telephone communication channels, being in that time analog ones. The results of such studies are summarized in the article by J. Arnott and M. Javed ( J. L. Arnott and Muhammad Y. Javed. Probabilistic character disambiguation for reduced keyboards using small text samples. AAC Augmentative and Alternative Communication, 8(1), 1992, pp.215–223 ).

In the experiments described in this article, the layouts of the Roman alphabet were analyzed both from the point of view of key-stroke number optimization and from the point of view of decreasing error rate, which is important for reading a message.

The studies showed that, for input of text, a layout based on the alphabet is unpractical and each language will have its own optimal layout designed with consideration of character frequency in the words of this language. That is, the letters in the character sequence on a key shall be rearranged so that the most frequent letters would be inputted by a smaller number of strokes than the more rare letters. However, as the same letters, for example Roman letters, are used in different languages with different frequency, the layout should be different for each language. Therefore, an optimal layout for German language will be different from the optimal layout for French or Italian languages though their alphabets are quite similar.

The existing keypad character layout based on the English alphabet, which is a de-facto standard for all modern mobile phones, was recognized to be a most inefficient for text input. Aside from its historical significance, the only advantage of this layout is that such layout is universal for all languages based on the Roman alphabet. In the result, this advantage has overcome all numerous disadvantages of this layout, but for English, French, German and other languages this layout is equally inconvenient. However, it was not necessary to redesign the layout depending on the place of sale of a phone and this overweighed all inconveniences.

Returning to the question of how to ensure that only one character is displayed on each key of the virtual keypad, now it is evident that character frequency shall be taken into account in every case. And this, in turn, implies that it will be necessary to design a separate layout for each language. Character frequency for some European languages is given in table, and the principles of designing layouts for the languages are discussed in the second part of this document.

Part 2. Theory and practice of “ARONETIS™”