The EU-funded LT-BRIDGE project is working to bridge the gap in language technology tools and resources for Maltese. Researchers from the project have developed an AI-based language model, BERTu, which is trained on Maltese textual data. BERTu is able to perform advanced natural language processing tasks, such as sentiment analysis and named-entity recognition, with a high level of accuracy. The model is also found to outperform other language models, occasionally by more than 20%. With BERTu, researchers are now able to explore more complex language understanding tasks in Maltese.
This is a major step forward in advancing AI-based language technologies for digital communication in all European languages, including Maltese. The Maltese language has been under-resourced when it comes to language technology tools, but with the development of BERTu and the work of the LT-BRIDGE project, this gap is being filled. The project’s goal is to help integrate Malta into European research and innovation efforts for AI-based language technologies, and with BERTu, they are well on their way to achieving that goal.
This will have a significant impact on the field of natural language processing and will pave the way for new and exciting applications in Maltese, such as automated translation, language-based search, and text generation. With BERTu, researchers have the ability to analyze, understand and generate Maltese language text with a level of accuracy and fluency that was not possible before.
Link to article: https://cordis.europa.eu/article/id/442665-computers-learn-maltese?WT.mc_id=exp