Societies and industries are rapidly changing due to the adoption of artificial intelligence (AI) and will face deep transformations in upcoming years. In this scenario, it becomes critical for under-represented communities in technology, in particular developing countries like Latin America, to foster initiatives that are committed to developing tools for the local adoption of AI. Latin America, as well as many non-English speaking regions, face several problems for the adoption of AI technology, including the lack of diverse and representative resources for automated learning tasks. A highly problematic area in this regard is natural language processing (NLP), which is strongly dependent on labeled datasets for learning. However, most state-of-the-art NLP resources are allocated to English. Therefore, creating efficient NLP tools for diverse languages requires an important investment of time and financial resources. To deal with such issues, our group has worked toward creating language-agnostic approaches as well as adapting and improving existing NLP techniques to local problems. In addition, we have focused on producing new state-of-the-art NLP publicly available data and models in Spanish. Next, we briefly present some of them.
Our group has worked toward creating language-agnostic approaches as well as adapting and improving existing NLP techniques to local problems.a
Twicalli, a social seismograph, and other crisis management tools. Timely detection and accurate description of natural disasters and other crisis situations are crucial for emergency management. This is challenging and important for our region, since one must rely on human observers appointed to specific geographical areas or on advanced infrastructure. In the case of earthquakes, geographically dense sensor networks are expensive. A viable inexpensive alternative to this problem is to detect events through people's reactions in online social networks, particularly on Twitter.a
No entries found