For a Human-Centered AI

Google Prize to Carlo Strapparava: the only Italian researcher that won the contest

January 4, 2017

Carlo Strapparava from Trento based FBK has created a software system that can tell emotions from a text. He is the only Italian researcher that has won the international contest organized early this year.

google1200A humorous phrase, or some writing that causes sensations of fear or happiness. We are all able to identify them, but now there is a new IT system that can recognize different types of writings, automatically and with no need of human assistance.

The series of algorithms designed by researcher Carlo Strapparava from the Fondazione Bruno Kessler (FBK) in Trento is able to distinguish some of the nuances and emotions expressed in written language and is able to open a new era in Internet surfing or in automatic generation of texts.
The study attracted the attention of Google, which in an international competition launched earlier this year, awarded the Italian scientist with $ 50,000 as an incentive to continue his research, especially with the participation of young researchers.

The competition launched by Google in several research areas was won by some of the best centers and universities in the world, like Harvard and Columbia University. The official results published this month show that Strapparava, with the HLT Unit (Human Language Technologies) of FBK’s Information Technology Center is the only Italian researcher who has received recognition for the June 2011 edition.
“Dealing with the emotional, persuasive, or other aspects of creative language content in the texts – Strapparava explains- is commonly considered to be off limits for any computational ability. Actually, these features are a key part of communicating, and it is important that research in the field of natural language processing deal with it. The usefulness of automatic recognition of these aspects is nowadays even greater, given the enormous daily production of texts on the web. Through these technologies, it will also be possible to predict the emotional or persuasive content of a text.”


The author/s