Installing Joshua Decoder for use in iPReS

Introduction

A critical component within the iPReS middleware framework is the actual machine translation required to do language translation from source to target language(s). From an early stage it was defined that language translation would be delegated to the popular Apache Tika content analysis and metadata extraction toolkit. As of Tika 1.6 release, Tika now maintains the functionality to Language Translation via a number of methods including both WebServices-based e.g. Microsoft/Bing Translation API, Google Translate API, Lingo24 API, and also external systems such as Joshua Decoder and Moses Machine Translation. This document provides commentary on how the Joshua Decoder Tika Translation Implementation has been implemented within the iPReS framework to translate PO.DAAC product metadata from source (English) to target (*foreign) language.

Prerequisites

• iPReS
• Joshua Decoder
• Apache Tika

Procedure

Gotchas