MHP operates Google, Twitter and Facebook emotion recognition technology

The agro-industrial holding Myronivsky Hliboproduct (MHP) is currently testing an innovative personnel emotion recognition technology which is already being actively operated by Google, Twitter, Facebook, Flickr, YouTube, Ricoh, Fujifilm, Canon and Nikon.

This was announced by the founder, CEO of the MHP holding Yury Kosyuk in the framework of the 5th annual Forum “Conductors of Changes”, which was held in Kyiv on November 13. According to him, the research confirms the thesis that happy people are more productive.

This opinion is also supported by Ksenia Prozhogina, HR and Communications Director at MHP. She clarified that currently a pilot project is being implemented in the company's central office, in which a team of psychologists and experts “trains” the machinery to recognize emotions, work with this data, signal employees' problems.

“This is a computer simulation project. Its goal is to understand whether a person is in a borderline state in which he continuously destroys himself. I’m not talking about finding out the cause of a short-term bad mood, but about identifying a long-term “unhappiness,” Ksenia Prozhogina stresses.

The test program uses data from several dozen employees who themselves volunteered to become participants in the experiment. According to its results, the project will be finalized and extended to a larger number of respondents.

It is significant that the system notifies of the problem, which allows the HR department to help the employee if necessary.

“We are trying to find out what is happening to a person: either something is going on in his family, and he needs help, or something is not working inside the team, or a person is given a project which he is not interested in,” Ksenia Prozhogina adds.

Note: Emotions analytics software development requires vast amounts of labelled emotions data. The emotions data comes from video cameras that capture facial expressions and microphones that collect data on tones of voice, for example. This data is fed into machine learning algorithms, which learn to recognize expressions, tones and other characteristics that correlate to specific emotions. Today's emotions recognition technology typically categorizes emotions as either anger, contempt, confusion, disgust, fear, frustration, joy, sadness or surprise.