Project leader: Philipp Koehn

Members: Philipp Koehn, Lane Schwartz and Mercedes Garcia

Goals

Two recently started EU-funded projects, MateCat and CASMACAT begun developing open source workbenches for human translators that can take advantage of machine translation. This project will involve some of the project members to come together, reaching out to other interested developers and merge the two ongoing efforts. Since the projects are still in a very early stage this is a good opportunity to get familiar with a open source CAT tool that will see significant development over then next two and and half years and beyond.

Progress

Day 1

Install the MateCat prototype at Edinburgh and on laptops

Day 2

Get familiar with existing code base (PHPTAL, JavaScript, Ajax, JQuery, SQL, ...).

Consideration of features to be integrated:

• word-level confidence measures (started working on this)
• highlighting word alignment in source
• translation options
• interactive machine translation

Day 3

Attempting to modify existing HTML elements. We'd like to give low-confidence words a red background.

Day 4

• Add a new class style in CSS with red background to indicate low-confidence words ( < 50 ).
• Change PHP code to add the functionality with the confidence measures.
• Change <textarea> for <div contenteditable="true"> to interpret the html tags inside the text.
• Changes in the JavaScript code to make it works with <div> instead of <textarea>.

Day 5

• Add a new class style in CSS for the <div contenteditable="true"> because we had some issues.
• Changes in the JavaScript code to make it works with <div> instead of <textarea>.
• Fully functional prototype (on the GUI side) of visualizing word-level confidence measures