Making the news of zientzia.eus smart
2020/11/17 Andonegi Beristain, Garazi - Elhuyar Zientziaren Komunikazioa
Zientzia.eus offers more than 23,000 contents in Basque in multiple formats: articles, videos, audios, photos and agenda. Previously, it has been the showcase of the technologies developed in Elhuyar and has integrated voice synthesis to listen to the news and the search engine of Elhuyar for the Basque language.
On this occasion, the increased reading of news shows users the most significant elements of the text. In addition, these elements are clickable and allow you to explore descriptive tabs brought from Wlikata. The most significant elements are found through entity linking techniques, specifically through deep learning algorithms proposed by Raiman and Raiman (2018) and, as said, using Wlikata as a knowledge base.
Since the news is related to the concepts of the Wliquidata knowledge base, consultations can be conducted based on the structure (taxonomy) of that knowledge base. This allows you to perform analytics based on zientzia.eus news. In this case, the analyses have been carried out based on different perspectives (gender, territory, language, origin, area of knowledge,...) and always with measurable indicators. The first version of this feature focuses on indicators associated with gender.
Artificial Intelligence and Language Technologies will be increasingly present in our daily lives, and the new technologies developed by Elhuyar are a sign of this. Technologies developed by Elhuyar in the call for ICT grants from the Basque Government.