The CASMACAT project (2012-2014) built the next generation translator's workbench to improve productivity, quality, and work practices in the translation industry.
We carried out cognitive studies of actual unaltered translator behaviour based on key logging and eye tracking. The acquired data was examined for how interfaces with enriched information were used, to determine translator types and styles, and to build a cognitive model of the translation process.
Based on insights gained in the cognitive studies, we developed novel types of assistance to human translators and integrated them into a new workbench, consisting of an editor, a server, and analysis and visualisation tools. The workbench was designed in a modular fashion and can be combined with existing computer aided translation tools.
We developed new types of assistance along the following lines:
- Interactive translation prediction, where the CASMACAT workbench makes suggestions to the human translator how to complete the translation. We adapted the existing interactive machine translation paradigm by adding input modalities, especially electronic pens and basing the suggestions on better exploitation of novel statistical machine translation models, such as ones based on syntactic structure.
- Interactive editing, where the CASMACAT workbench provides additional information about the confidence of its assistance, integrates translation memories, and assists authoring and reviewing.
- Adaptive translation models, where the CASMACAT workbench learns from the interaction with the human translator by updating and adapting its models instantly based on the translation choices of the user.
We demonstrated the workbench's effectiveness in extensive field tests of real-life practice of a translation agency. In addition, we also reached out to the wider language service industry and online volunteer translation platforms. The outcome of the CASMACAT project is available as open source software to industry, academia, and to individual end users.