Purpose

Jovian Tempest, credit NASA/JPL-Caltech/SwRI/MSSS

Scholar information at large is growing at a very rapid rate. In addition to traditional scholar publications and patents, the scientists need to cope with a massive increase of primary and experimental scientific data. This evolution is not limited to natural sciences and social sciences, the Digital Humanities face similar issues with numerous ongoing massive digitalization programs.

This evolution is often presented as an issue for the scientists, as an information overload. However, this growing amount of information is also a fantastic opportunity for accelerating and strengthening science. After the revolutions of rational thinking in ancient Greece, in experimental and observation methods from Galileo and Bacon, after the practical and mass applications of scientific thought afforded by the industrial revolution, science is poised for a fourth scientific revolution using computers to create new concepts, ideas, models and simulations,  bringing a new scientific renaissance.

This new research paradigm of data-intensive science requires new methods, tools and technologies. The goal of science-miner is to contribute, at its (very) modest scale, to make a fourth scientific revolution a reality, by developing new, high performance, scalable Open Source softwares based on Machine Learning to support probabilistic scientific knowledge engineering. For us, one of the main challenges of the future of eScience is the ability to learn and integrate continuously new entities and contradictory facts, in order to generate automatically new hypotheses and knowledge of interest for the scientists.