UA124691U - METHOD OF INTRODUCING TEXT TO COMMUNICATE THE REDUCED NUMBER OF LEADING COMMANDS - Google Patents

Abstract

A way to enter text to communicate a reduced number of control commands is that the text is entered with a minimum number of controls by grouping letters into blocks. The text input process consists of two steps, where the first step is to select grouped letter controls based on selection methods for continuous and discrete control signal, and in the second step to determine the most probable words using the maximum likelihood method based on the corpus of words and phrases , which is formed on the basis of dialogue texts for everyday communication.

Description

The utility model belongs to the field of computing and is intended for inputting text into digital devices for the purpose of providing alternative communication. Allows you to simplify the text input process by reducing the number of control signals (less than the number of letters of the alphabet), predicts word variants, based on data from the corpus of words and word combinations.

A known method of entering text with a reduced number of commands is used in mobile devices with digital buttons containing several letters of the alphabet (1). The method involves the use of 9 control elements. Another method, the closest in technical essence, is the technology known as T9 (2), which allows you to enter words with one press for each letter of the word (the button also contains several letters).

The technical problem, which the claimed method is aimed at solving, is the problem of ensuring the possibility of entering alphabetic characters and forming words and sentences from them by expanding the TU technology with the help of 1) a minimum number of clicks; 2) different methods of selecting control elements (blocks, letters, candidate words, etc.); 3) the algorithm for predicting the next words of the sentence based on the already entered ones.

The set technical problem is solved by reducing the list of control signals for text input by grouping the letters of the Ukrainian alphabet. That is, the control elements for entering text information are blocks consisting of grouped letters in a certain sequence: alphabetical, keyboard ("jepu"), by frequency of use, etc. The number of such blocks is much smaller than the number of letters and may vary depending on the type of input control signal. When implementing the claimed text input method, the control signal received from the user can be continuous or discrete, have different data transfer rates and different quality.

Features of the control signals require the use of various methods of element selection. they can be grouped as follows: direct selection, scanning, coding. The speed of selecting elements directly affects the speed of text input, and therefore the speed of communication.

Direct selection allows the user to access all blocks with grouped letters directly. This can be achieved when the input control signal has a sufficient amount

From states for selecting these blocks and implementing additional control commands. Direct selection is also possible when a reliable continuous control signal is established. In this case, the selection is made using the cursor.

If the control signal includes only two different states, then the element scanning system is used to make a choice. In this case, elements or groups of elements are shown sequentially, and the user must confirm or reject the proposed choice.

Encoding - is one of the variants of direct selection, where each element in the output alphabet is encoded into a sequence of input commands. Since the number of unique codes is limited, it is advisable to use this option only for a set of the most frequently used words and sentences.

The use of a reduced number of control signals involves attaching more than one letter of the alphabet to each selection element, resulting in ambiguous selection.

The disambiguation algorithm uses a word database to search for candidate words that match a selected sequence of clustered letter blocks. As the number of such blocks decreases, the number of letters belonging to one block increases, which leads to an increase in candidate words and requires the implementation of a text input system using a word prediction algorithm and a method for optimizing the distribution of letters by blocks.

With the help of prediction algorithms, the words that occur most often after the entered words in the sentence are suggested.

To implement the text input system with prediction, a set (corpus) of words (phrases) of the Ukrainian language, which is limited to words for everyday communication, has been formed.

On the basis of this corpus, a suitable model was formed for the possibility of predicting the following words in a phrase.

An expert approach was used to form the corpus of words, which made it possible to select words and phrases used in everyday communication based on sources from the content of Ukrainian-language websites, periodicals, dictionaries, etc.

In order to model the amount of text information obtained for the purpose of further prediction of words in word combinations, the obtained sentences from dialogues are divided into M-grams - unigrams, bigrams and trigrams |ZI.

M-gram models determine the probability of using a given phrase (phrase), formally, as the probability of occurrence of a sequence of words in a certain corpus (set of texts).

To estimate these probabilities, the maximum similarity method of MI E (Tahitit

IKeinosa eviitaitsyop) (ZI.

Grouping a set of letters of the Ukrainian alphabet in a certain sequence (alphabetical, keyboard) allows you to minimize typing efforts for people with different experience of working with digital devices. In the case of frequency distribution, the forecasting efficiency increases.

The implementation of the declared method of entering text into digital devices allows you to achieve the following technical results: - reduces the number of text entry controls; - processes a continuous or discrete input control signal characterized by different data transfer rates; - uses such methods of selecting control elements as direct selection, scanning and coding; - allows you to change the sequence of letters to alphabetical, keyboard ("Zuepu") or by frequency of use; - uses thematic dictionaries to predict words.

The claimed method can be illustrated graphically. In Fig. 1 shows the touch screen of the device in text input mode. In Fig. 2 shows a table with a list of proposed options for the sequence of letters and their distribution for different numbers of control units. In Fig. With the sequence of actions for the implementation of the word set is demonstrated.

The working area of the screen (Fig. 1) consists of the following fields and control elements: 1) a field for displaying the entered text; 2) an area for displaying suggested words corresponding to the current code; 3) an area displaying the current number of control blocks and the selected sequence of letters in them; 4) a control element that allows you to select a block that contains the required letter;

From 5) a control element that allows you to cancel a wrong choice; 6) a control element that makes the transition to the next word; 7) a control element that activates the function of voice reproduction of the text.

For the 8th groups, it is advisable to use the alphabetical sequence of letters (a similar distribution is used in mobile devices), since this order of letters is well-known and does not require much time to adapt.

For the 6th group, a keyboard sequence of following letters is proposed, which corresponds to the standard arrangement of lines and letters, like a normal keyboard. Also, this distribution better predicts word variants, due to the distribution of the most frequent letters concentrated inside the keyboard, into different blocks.

For 4 groups, it is more appropriate to use frequency sequences following letters. For other sequences, the number of words with the same code increases significantly, and their use becomes impractical.

The proposed method involves a certain sequence of actions (Fig. 3), which must be done to type a word. In one action, not a specific letter is selected, but a set of letters, which depends on the chosen arrangement of symbols by blocks.

The word entered in this way forms the code for the current layout of characters on the keyboard.

This code is then compared with an internal dictionary, and a list of matching code words is offered, sorted by the score calculated by the maximum similarity method. The word that receives the highest score is offered as the "default" word and does not require any additional action from the user to select it. The following words are entered in the same way.

The formed sentence is voiced using the Ukrainian language synthesizer.

Sources of information: 1. 5iimetbega M., MasKep?ie I.5., Kotopep R. Rgeedisiipud yehi epigu zreed op tobriye rpopev// Rgoseyedipd5 oi She ASM Sopiegepse op Nytap Rasioiv5 ip Sotrshiyipd Zubzivt5-SNI 2000. - 2000. -

Mem mok: ASM. - R. 9-16. 2. Sygomeg 0.Yi., Kipd M.T., apa Kyvspieg S.A. Raiepi Mo. О055818437, Nedysed Keuroaga disatridtsaiipud sotriieg. Tedis Sottipisaiopv, Ips., Zeanieu, UA (1998).

Z. digaizKku, O. breesi apa I apatsade Rgosezvzipa/ Yu. UigayivKu, u. N. Mapip. - 2gpa-Mem/ Uegzeu: 60 Rgepiise Nai, 2008. - 1024

4. Yu.V. Krak, A.V. Barmak, R.A. Bagryi, I.O. Ceiling. Text input system for alternative speech communication// Problem! management and informatics. - 2017. - Month 1. -

P. 128-137.

Claims (2)

USEFUL MODEL FORMULA 1. The text input method for communication with a reduced number of control commands, which consists in the fact that the text is entered with a minimum number of control elements by grouping letters into blocks, which differs in that the text input process consists of two stages, where in the first stage a selection is made control elements with grouped letters based on selection methods for a continuous and discrete control signal, and at the second stage, the most likely words are determined by the method of maximum similarity, based on the data of the corpus of words and phrases, which is formed on the basis of dialogue texts for everyday communication. 2. The text input method according to claim 1, which differs in that the grouped letters in the blocks are set in alphabetical, keyboard order, following the frequency of use, and the distribution of the set of letters in the control elements is determined depending on the number of such control elements. and She A A M MU VVA Y hi Ya buye) mem ne mh waye Zm ahzhana More : : I WANT TO BE UNDERSTOOD WITH I! The same thing | ZVOBUMSHI 00100 ZeOZUMILE | | I UNDERSTAND ' I would, MU Uh : She | UNDERSTANDING : ; UUUUUUU U U UM UM UM UM UM : FO I ut, I : sha . ; No in N pet Kut yati pen p i i M IBUTENHZH 00 bOUTRMA I; u gi aa aa MO PI PANA VAN : Kiyak schi ho zh REZ o M 'Ayaynpzchsh | ивЕвВсвго |. : Hooyuyuyuyuyuyu yuyu kyu yuyu o ; x sdtstettenkyi Fig. 1 o Kilkyut VBozpolii beech! o block: nm y МВ В В В С С В В Б Б и: и сом и оч ты У г и V. AV i dDEeЖ her yan SCHI e i KAM i ope sh TUFH i i KO V PI V YOU SY V y chni i vya o btsukvn o | gshshaki KO unyuchyauchuuhuummu mu tn teach Km Kuchmy : ovivApe | olazhe ooyaYachami 1 tBu and GE TV tTKDHZH | OUVMYIISH and and her Koru uyu p: and lainizchv || Ikev EP Nosov o.o. MOH A so io Fig. 2 o MNtsUKEN | tshshchah nn nin nyyn nky nyny nyny in A I OKA AAAKK KK i : i Bek Gesnikity of choice BINAYe spyuMmlyazhe 00 deoh VKM i udnei Vus x; ххххх в ШИ a shim kyu yachesmi | What a spirit I see! Leads vvvn j che mch xx Ach as a child is confused, persuade va IN TE hod hod kh Kk her wheelbarrow dyan ! uuooooetoikKKKKnkonkncnkncnosechn vOapit "DI | Puyuenkmkkwana pis Mesuyuyummmh Hymn ee moh YYYYYYYYYYYY YYYY YYYY YYYY YYYY Fig. WITH