Automatic learning through modulable supervision for automatic estimation of translation quality

The main objective of this project is the research, development, and validation of automatic translation quality estimation systems using supervised and unsupervised machine learning methods.

This main objective is broken down into the following scientific and technological goals:

1. Research and development of methods and tools for automatic translation quality estimation using supervised machine learning.

2. Research and development of methods for automatic translation quality estimation using unsupervised machine learning.

3. Creation and preparation of suitable datasets for training supervised estimators.

4. Evaluation of the developed systems for estimating the quality of machine and human translations.

The following scope and impact objectives have also been defined:

1. Definition of a pilot project for applying the project results, which will determine the requirements for the techniques to be developed and the linguistic resources to be used.

2. Validation of the prototypes against the state of the art in terms of adaptability and accuracy of the quality estimations.

3. Academic dissemination of the project's results at international conferences.

4. Transfer of results to industry and analysis of commercial exploitation possibilities.

To achieve these objectives, QUALES has a strong consortium with extensive experience in the field of language technologies, which aspires to become a benchmark in the field. This is a strategic project for all participating entities and for Eiken, MondragonLingua, Eleka, and Argia, which have expressed their support for the project through letters of endorsement.