Features/IndicTypingBooster

From FedoraProject

< Features
Revision as of 15:12, 1 February 2011 by Petersen (Talk | contribs)

Jump to: navigation, search


Contents

Indic Typing Booster

Summary

Indic Typing Booster is a predictive input method for ibus and scim. It suggests complete words based on partial input, which can then simply be selected from a list, and boost one's typing speed for more enjoyable input.

Owner

  • Email: psatpute AT redhat DOT com, nkumar AT redhat DOT com

Current status

  • Targeted release: Fedora 15
  • Last updated: 2011-02-01
  • Percentage of completion: 20%

Detailed Description

This is an input method engine using scim-tables or ibus-table to look words in a dictionary table. More information about project can be found at https://fedorahosted.org/indic-typing-booster/

Indic Typing Booster supports currently widely used maps, so the user does not need to switch his existing input maps, like inscirpt, itrans, typewriter. So with this new input method he can type the same way as he did with his older IME's but with prediction.

The mission of this project is to provide user friendly predictive text entry for Indic language users.

Benefit to Fedora

Users can create content in their native language with enhanced speed, accuracy and ease.

Speed:

After pressing a few characters the user can get the expected word in the candidate list and quickly select that.

Example: Entering the word 'भारतीय' with itrans keyboard normally requires 9 keys presses: 'bhaaratiiy'. With Indic Typing Booster it will only require 6 key presses, after inputting "bhaar" user can get 'भारतीय' in suggestion box, and then just need to press corresponding number key and done.


Accuracy:

Since the suggestions come from a validated word dictionary database, the selected words always give 100% accurate spelling. Note: correctly the present generated database is from unreviewed web data, in the long term with community help, they can be reviewed and the correctness checked as per standard. Also infrequent words which may be mis-spellings can be filtered easily from the dictionaries.


Easiness:

Indic scripts are complex since the order of rendering may differ from input order. It has reordering (eg क + ि -> कि ), feature conjunct (eg क + ् + ष -> क्ष), as well as split matras. Often new users are not familiar with these subtleties leading to incorrect or invalid text entry. Indic Typing Booster relaxes this as the user need to enter the complete word.

Adding New Words: One can add new words in to the IME's database, and they will also appear in the suggestion box.

This is easy with scim-tables: type phrase/word and then hit "Ctrl + =", then just enter a corresponding input key, and done.

Scope

1) Create dictionary tables from wikipedia dumps for Marathi, Hindi, and Malayalam. More languages will follow later.

2) Fix ibus-table to work well for input with Typing Booster tables.

3) Test with scim-tables and ibus-table.

How To Test

  • Install packages if required
  • Add input-methods to ibus or scim
  • Activate input method and type with completion in applications.

User Experience

1) predictions for entered words

2) easy typing of complex words

3) improved typing speed

4) often native users type their language's words using latin script (transliteration), so with this IME I think they will be able to type words in there own script easily.

Dependencies

This feature requires some improvements to be made to ibus-table.

It is already working well with scim-tables.

Contingency Plan

As these are new input methods, no contingency plans necessary.

Documentation

See the project website

Release Notes

Indic Typing Booster input methods provide new completing input for Indic languages.

Comments and Discussion