An innovative methodology combining professional translation, translation memories and artificial intelligence to translate a selection of the online encyclopaedia into English, making Italian knowledge accessible to a global audience.

The hybrid approach ensures linguistic accuracy, terminological consistency and respect for the scientific rigour of the content, enhancing human experience and empowering it through advanced technologies.

The methodology was divided into three complementary stages that guarantee the quality and consistency of the translation.

In the first stage, a selected corpus of entries was professionally translated by expert translators. Based on this translation, a Translation Memory (TM) was created and a customised Machine Translation (MT) model based on Lara was trained.

In the second stage, the model automatically translates the remaining entries, ensuring terminological and stylistic consistency.

In the third phase, the Human-in-the-Loop system identifies the most critical strings for human review, ensuring maximum accuracy of the final result.

Artificial Intelligence translates the entire corpus while maintaining terminological consistency. Professional translators intervene on critical passages identified by the Human-in-the-Loop system, ensuring accuracy and fidelity to the original text.

This approach allows large volumes of content to be processed without compromising quality; technology handles large-scale translation while human expertise refines the translation work and intervenes where the system identifies critical issues or ambiguities.