User:Ote/proposal
Open Translation Engine (OTE) Version 2.0
Proposal for NLnet[1]
Contents
Abstract
With this funding, I will produce a web-based system to allow a community of users to cooperatively build translation dictionaries.
While there are many Open Source projects in the field of Machine Translation, there is a lack of tools for the community creation of translation dictionaries. The goal of the Open Translation Engine (OTE)[2] Version 2.0 is to create a robust tool for this space.
Use Cases
OTE Version 2.0 will provide additional tools to allow the system to be used as a dictionary building and management tool for Apertium.
School / Language Learning
Bootstrapping
License
Currently the OTE is under the BSD License. This includes both the source code and the translation dictionaries.
During this project, I will re-evaluate which Open Source license is the best choice, with particular attention to possible differences in needs between the source code and translation dictionaries.
Project
Core System
Conversion to Unicode - The current prototype is not unicode aware. Unicode is an absolute necessity to continue the project with all possible word languages.
Genericization of code base - The OTE will be built to be as generic as possible, thus allowing for ease of future enhancements. Currently the prototype code is hard coded with the Dutch/English language pair.
Install Procedure - Installation will be made as user-friendy as possible.
Documentation
User System
User accounts
User Permissions
User administration
Core Tools
Word Viewer
Dictionary Viewer
Dictionary Export: Export of Apertium XML dictionaries. This will allow OTE to be used as a dictionary building and management tool for Apertium.
Dictionary Import
Word-2-Word Translator
Classroom Tools: Random Word, Flash Cards, Word Lists
Community Tools
Add / Delete / Modify: Individual Word
Add / Delete / Modify: Tagging: Parts of Speech for a Word
Add / Delete / Modify: Translation Word Pairs
Add / Delete / Modify: Languages
Versioning - all words, translation pairs, and tags are versioned, allowing reversion to previous states.
