Microsoft is indeed trying to innovate in several fields, but the highest level of innovation certainly comes from its Research unit, which builds projects that can surprise anyone out there every once a while.
This time, the team working at Microsoft Research created a new technology that can identify humorous content using a complex algorithm that analyzes every single part of a joke or cartoon to determine its score and then calculate how funny it can be for the human being.
Already tried by the New Yorker for its own cartoon contest, the new feature was impressively effective: the favorite cartoons manually picked by editors were included in 55.8 percent of choices made by the computer running Microsoft's new system, according to a Bloomberg report.
Мнение жюри конкурса совпало с выбором алгоритма Microsoft на 55,8%. Технология была протестирована на 16 различных рисунках. Анализировалось текстовое содержание картинки, графическое изображение, положение элементов рисунка.
Microsoft explains that its new technology was tested on 16 different cartoons to determine which is funnier, and analyzed data such as the length of the joke, the location of the punchline, and the language being used.
The shorter the joke, the funnier, the research shows, and the punchline obviously needs to be at the end of the joke. Microsoft's feature, however, is still having a hard time detecting some expressions and unusual language, so there's still room for improvement, but Redmond says it will continue work in this direction.
“Automated generation and recognition of humor could be harnessed to modulate attention, engagement, and retention of concepts, and thus has numerous interesting applications, including use in education, health, engagement, and advertising. Beyond applications, we believe that pursuing information-theoretic models of humor could reveal new insights about one of the most fascinating human behaviors,” Microsoft explains.
Needless to say, developing a feature that can help computers detect humorous content and in their turn become funny could help give a more human feeling to robots and change the world once and for all. Bill Gates has already predicted that in about 20 years robots will be all around us, so what's wrong with trying to make them funny?