Koko allows users to share their serious thoughts.
American scientists from the University of Cincinnati are developing a neural network that can analyze the results of psychological tests and interviews with students, and then give their verdict about the aggressiveness of a particular student.
During the tests, the accuracy of this algorithm was 90%. In the experiment, the neural network was proanalizirovana data, 103 people aged 12 to 18 from 74 schools. Students were selected on the basis of registered cases of aggression in school.
Each student gave a short interview, after which scientists have identified in these words semantically related to different categories of mental diagnoses, positive and negative emotions, as well as illegal actions.
In the past year, has released a mobile application Kip, which allows you to daily monitor the progress of patients in the conduct of psychotherapy, which makes the treatment process more effective. Sometimes psychological support is available and without recourse to a specialist, for example, using an online service Koko, who created the psychologist and former researcher at the MIT Media Lab Robert Morris. Composure users to share their serious thoughts, and to help other people with similar problems.
The result was an algorithm that indicates correlation between certain words and emotions, as well as test results, aggressive behavior. The main goal of this study is the automatic detection of students with hidden aggression in their interviews. While scientists still do not plan to implement this system in schools, because it still needs improvements and statistical study.