Tilde MT Platform for Developing Client Specific MT Solutions

Tilde MT Platform for Developing Client Specific MT Solutions M¯arcis Pinnis, Andrejs Vasil¸jevs, Rihards Kalnin¸ˇs, Roberts Rozis, Raivis Skadin¸ˇs, Valters ˇSics Tilde Vien¯ıbas gatve 75A, Riga, Latvia, LV-1004 {marcis.pinnis, rihards.kalnins, roberts.rozis, raivis.skadins, valters.sics, andrejs}@tilde.lv Abstract In this paper, we present Tilde MT, a custom machine translation (MT) platform that provides linguistic data storage (parallel, monolingual corpora, multilingual term collections), data cleaning and normalisation, statistical and neural machine translation system training and hosting functionality, as well as wide integration capabilities (a machine user API and popular computer-assisted translation tool plugins). We provide details for the most important features of the platform, as well as elaborate typical MT system training workflows for client-specific MT solution development. Keywords: machine translation, cloud-based platform, data processing workflows 1. Introduction Today, as globalisation and cross-border trade infuse more sectors of the economy, the need for translations and mul- tilingual content is continuously increasing. The demand for translations is surpassing the supply that professional translation services can handle. As Common Sense Advi- sory reported in a 2016 market study, ”enterprises intend to increase their translation volumes by 67% over current levels by 2020.” (Lommel and DePalma, 2016) An obvi- ous alternative to customers who cannot afford professional translation services or who have too much content to trans- late, and also an obvious choice for translation and local- isation service providers who see the potential in increas- ing their productivity, is machine translation (MT). Gen- eral MT providers (such as Google1 and Microsoft2) cater for the masses with general-domain MT systems. However, customers who require purpose-built and highly customised MT systems, particular for complex or low-resourced lan- guages, turn to MT service providers that offer MT sys- tem development and customization capabilities as a full- service package. One such custom MT platform is Tilde MT3, the successor of the LetsMT! platform (Vasil¸jevs et al., 2012) for Sta- tistical Machine Translation (SMT) system development, first launched in 2012. Tilde MT builds upon the LetsMT! platform by providing greater customisation capabilities, smarter data processing workflows, and current state-of- the-art Neural Machine Translation (NMT) system support. The paper is further structured as follows: Section 2. pro- vides an overview of the main features of the Tilde MT plat- form, Section 3. describes the MT system training work- flow with the different tools used for data processing, Sec- tion 4. describes the MT system translation workflow, and Section 5. concludes the paper. 2. Overview of Tilde MT Tilde MT is a cloud-based custom MT platform that al- lows users to store linguistic resources (such as paral- 1https://translate.google.com 2https://www.bing.com/translator 3https://tilde.com/mt lel and monolingual corpora and multilingual term collec- tions); train SMT and NMT systems; integrate the systems through computer assisted translation (CAT) tool plugins or the Tilde MT external API in users’ translation and multi- lingual content creation workflows; and perform translation of text snippets, documents of various popular formats, and websites directly in the Tilde MT graphical user interface or using a widget that can be integrated in any website. 2.1. SMT …

Content truncated to 500 words