alBERTUM: A Portuguese Search Engine for Scientific and Academic Language

alBERTUM is a Portuguese search engine designed for scientific and academic language. This paper provides an overview of alBERTUM’s data sources, model, architecture, and two core functions. The search engine uses a vast collection of scientific open data available in national repositories to provide bilingual terminology searches and offer a Portuguese academic writing assistant. Utilizing the deep learning model BERT, alBERTUM converts the available data and user queries into semantic vectors, enabling semantic searches for precise and contextually relevant results. The bilingual terminology search function is particularly valuable for professionals, researchers, and those working in fields that require cross-linguistic understanding, such as translation studies, linguistics, or international business. alBERTUM identifies translation equivalents in scientific documents that are not translations of each other. The academic writing assistant function seamlessly incorporates an European Portuguese academic phrase bank, which has also been developed by our team, into the search engine architecture. This integration empowers users, particularly students with limited academic vocabulary, to search for academic expressions based on their intended communicative function. The assistant not only provides contextual examples but also shows the communicative functions of these expressions, enhancing users’ understanding of their usage within an academic setting.