Something New Under The Sun
Researchers never cease to surprise us. This time around, scientists from Carnegie Melon University and the Hebrew University of Jerusalem have come up with a new Big Data Analysis method that revises large databases of patents, inventions, research papers and proof-of-concept diagrams in order to identify ideas that can be put to use in other applications, or even create new products altogether.
Their approach was twofold: they engaged crowdsourcing to obtain the initial (raw) comparative data, and a method of artificial intelligence called “deep learning”. So the first stage was used in order to register how people made analogies, and those were used by the deep learning model in the second stage so that computers would learn how to make even more analogies based on the observations made by human observations.
The results have been overwhelmingly positive. The power of improvisation has been greatly magnified by this new method and we can now surely “expect the unexpected”. We can now say this is the first time that the analogy problem has gained traction in the Computer Science field. We can henceforth expect that the rate of innovation will take off, as new and better solutions for existing problems will arise from such a vastly improved effort.
As one may already suspect, the history of technology is littered with examples of innovations brought about by analogies. Just to mention a few, Brazilian inventor and aviation pioneer Santos Dumont patented the world’s first wristwatch after comparing timekeeping devices used aboard ships. Italian biological scientist Salvatore Luria came up with an experiment to determine mutations in bacteria (which earned him a Nobel Prize) by watching slot machines.
What we have here is yet another clear demonstration that the rate of solutions found will be multiplied (by orders of magnitude) when one combines the computing power of machines with the subtle and deep capabilities of the human mind. Computers do not understand things on a deep semantic level the same way humans do. Earlier attempts to create data structures by hand turned out to be ineffective and very time-consuming, and therefore improper for databases containing millions of patents and scientific research papers. Approaches strictly using text matching also did not work to satisfaction because they were limited to identifying superficial similarities, and missing out altogether the deep understanding that problem-solving requires.
This remarkable development will be presented on August 17, 2017 at KDD 2017, the Conference on Knowledge Discovery and Data Mining in Halifax, Nova Scotia, Canada. The research paper has already won awards for Best Student Paper and Best Paper.
Here at Tenex Developers we fully embrace the methods and technologies that enable innovation and the enhancement of productivity. As staunch believers in the powers of Big Data Analysis and Artificial Intelligence, we would like to invite you to contact us. Write us an email or give us a call and we will be happy to demonstrate how we can boost your business and help you reach your aims while making you stand out in your market segment.