Machine translation and post-editing 3200-M1-TMP
The aim of the course is to familiarize students with machine translation technology (MT), both in theory and practice, and with the increasingly popular model of the translators’ work: post-editing of machine translation (PEMT). Course participants will use the available MT solutions, they will also try to build their own ones. The possibilities of integrating MT systems with CAT tools will also be presented. Great emphasis will be placed on practical translation work - a post-editing of machine translated texts in accordance with ISO 18587:2017.
During the course students will get acquainted with:
- the history of MT and various types of MT;
- MT engine operation modes, typical errors in their operation;
- rules of picking the right MT solution for a given translation task; ISO 18587:2017 international standard, which regulates the rules of using MT tools in the work of a translator;
- available solutions for building custom MT engines; necessary requirements, data preparation, typical data formats;
- methods of manual and automatic assessment of MT quality;
- ways of integrating MT tools with CAT tools;
- post-editing as a type of translation work; levels of post-editing, ways of doing the post-editing.
They will also perform practical tasks in the field of selection, training, quality assessment of MT engines and in the field of post-editing texts in various domains, in different language pairs, translated using various MT tools.
Forms of work:
- elements of presentation and lecture;
- independent and team work with configuration of MT tools and CAT tools;
- independent post-editing of machine translated texts, performed during classes and at home;
- error analysis in MT and own translations;
- independent and team exercises in quality evaluation of machine translations.
Student’s workload:
30 class hours;
15 hours of preparation for classes;
5 hours of preparation of the final project
Total 50 hours.
Type of course
Mode
Prerequisites (description)
Course coordinators
Term 2023L: | Term 2024L: |
Learning outcomes
Knowledge: types of machine translation, typical problems, language errors in machine translated texts; rules of choosing MT tools for the type of translation work; the possibility of customizing MT tools - training own MT engines; quality evaluation systems for machine translation; post-editing as a type of translation; levels of post-editing; rules for post-editing.
Skills: working with MT tools; integration of MT and CAT tools; working with solutions for creating custom MT engines; post-editing of machine translation; using automated and manual methods to evaluate the quality of machine translated texts.
Social competences: awareness of the responsibility of the translator when including MT tools in the workflow; knowledge of the professional translator's ethics when using digital tools, when processing the entrusted data and when communicating with the client.
Assessment criteria
Requirements:
1. attendance;
2. active work during workshop classes;
3. small “entry” tests from readings;
4. a final project in the field of post-editing and evaluation of own post-editing work or in the field of training and / or evaluation of the quality of available MT solutions.
Criteria for evaluation of the tests and final project:
• 99% - 100 - 5!
• 98% - 91% - 5
• 90% - 86% - 4.5
• 85% - 76% - 4
• 75% - 71% - 3.5
• 70% - 60% - 3
• less than 60% - 2 (failing grade).
Each requirement, if assigned, must be fulfilled independently, hence their share in the final mark is not determined.
Bibliography
Readings provided by the teacher on the didactic platform.
Additional information
Additional information (registration calendar, class conductors, localization and schedules of classes), might be available in the USOSweb system: