Predictive text input

Predictive text input (through the English predictive “suggesting” from Latin praedictīvus ) is a system of accelerated text input into digital devices, in which the device software during typing offers options for ending words and phrases based on existing in its dictionary , and can also suggest correcting common mistakes.

Content

Application

Mobile Devices

Predictive text input is common on mobile devices (primarily phones), on which text input is difficult, for example, when writing SMS messages. Two predictive input systems are mainly used: AutoCorrect and T9

Editors for programmers

The editor analyzes the text and displays a menu that suggests the end of the entered keyword or identifier . Or, for example, when entering a tag, HTML offers a choice of all its parameters.

In most editors, the predictive input menu is called up either with a short pause of the set, or by pressing the Ctrl-Space button.

Document Editors

When entering text, the system suggests the end of the word entered. For example, this can help when entering tables in which the contents of cells are often repeated. In document editors, developers use predictive input very carefully, since prediction errors are especially common here.

Problems

General

In languages with a developed ending system (as in Russian ), a system that offers ending a word by its beginning does not work well - there is no guarantee that the word will turn out in the case in which it is necessary.

For mobile devices

In languages with many short words ( English ), the predictive set offers too many alternatives. Yes, and in other languages, the same key combination can produce several words, and if you do not look at the screen, you can send an error message. The most entertaining phone errors are collectibles , and sometimes they are included in teenage slang , for example:

the name of the vodka Smirnoff T9 turned into Poisoned - "poisoned";
the English antonyms select and reject (“select” and “reject”) are on the same phone buttons;
Russian words the body and dinner on most phones are on the same buttons, which may cause the error “ come home - dinner is in the refrigerator ” → “ come home - the body is in the refrigerator ”
“mobile number - grave number” also match;
but the most common and unpleasantly ambiguous mistake of T9 is the substitution of the word “you” with “yourself”. Especially in the phrase "I love you";
on Nokia, Sony Ericsson phones, the phrase “kiss on the neck, scratch behind the ear” turns into “kiss on the crack, scratch behind the snout”.
On some phones Nokia T9 knows the strange word "obscurantism"

When entering in languages with agglutinative elements, for example, German , you have to enter the word in parts. The text input system should be prepared for this. For example, the implementation of T9 in Sony Ericsson phones offers this solution: after typing the first part of a word, the user clicks on the “OK” button and the second part is entered.

You should also consider entering words that are not in the dictionary. This is most elegantly done in iTAP : when the user sees that the phone is mistaken, he presses the “Erase” button once and fixes the beginning with the arrow buttons. After that, he enters the remainder one letter at a time. In T9, you have to erase the word completely (or almost completely) and type it one letter at a time.

For touch phones Swype technology has been developed, which allows you to type text without taking your hands off.

In programmers editors

The predictive text input system used by programmer editors must be error resistant. For example:

  rObject .  x = aObject .  x + someVector .  x // programmer skipped semicolon
 rObject .  y = aObject .  y + so_ // the cursor is here

In this case, the predictive input system should, despite the missing semicolon, enable the programmer to select someVector identifier from the menu.

Sometimes an error in the code can become a prerequisite for predictive input. For example.

  Vector so_ // the cursor is here
 .  .  .
 rObject .  x = aObject .  x + someVector .  x ;  // someVector id not defined

In this case, the system may offer to enter the identifier someVector .

Curiosities

In Motorola phones (in particular, in firmware R365—2.2F for Motorola C650 and its analogs) until about 2005 there were obscene words in the Russian dictionary. In subsequent models (for example, Motorola KRZR K1 ) they were removed.
In the phones of Siemens (C65, M65, M75, S65, S75, etc.), the Russian dictionary also included obscene language.