CPC G10L 25/63 (2013.01) [G10L 15/08 (2013.01); G10L 25/72 (2013.01); G10L 2015/088 (2013.01)] | 9 Claims |
1. A harmful behavior detecting system, comprising:
a computer configured to execute observation and detection regarding harmful behavior among people in work environment, the harmful behavior including power harassment, sexual harassment, and bullying,
wherein the computer is configured to:
obtain data containing voice information, emotion information, or vital data of a target person, date and time information, and location information of the target person, the data being obtained by inputting voice around a target person, the voice information containing words that represent conversation content extracted on a basis of automatic voice recognition from the voice data, the emotion information representing emotion classified on a basis of emotion recognition from the voice data;
calculate an index value regarding the harmful behavior by using five elements on a basis of the obtained data, the five elements including words and an emotion of another person positioned around the target person, and words, emotion, and vital data of the target person;
estimate a state of the harmful behavior on a basis of the index value, the state of the harmful behavior including a victim or an assailant of the harmful behavior, and presence or absence of the harmful behavior or a degree thereof; and
output handling data for handling the harmful behavior in accordance with the estimated state;
wherein the target person is a first user who uses the harmful behavior detecting system and who may be a victim or an assailant of the harmful behavior while the another person may be the assailant or the victim of the harmful behavior;
wherein the voice is received by an apparatus carried by the user or an apparatus installed in the work environment;
wherein the vital data is detected by a sensor of the apparatus carried by the user, and includes a body temperature or a pulse rate;
in the step of obtaining the data, the computer compares an obtained word and a set key word, calculate an evaluation value (W) of the word as a numerical value in accordance with the corresponding key word, compares a value of the obtained vital data with a setting range, calculate an evaluation value (V) of the vital data as a numerical value in accordance with the corresponding range, and calculate an evaluation value (E) of the emotion as a numerical value in accordance with the obtained value of the emotion,
wherein the computer calculates the index value by a calculation reflecting the evaluation value (E) of the emotion of the target person or the another person, the evaluation value (W) of the word of the target person or the other person, and the evaluation value (V) of the vital data of the target person,
when the index value is calculated at each time point in time series, the computer is configured to:
calculate the index value by using a calculation formula of “([the evaluation value (W) of the word of the other person]+[the evaluation value (V) of the vital data of the target person])×[the evaluation value (E) of the emotion of the other person]”, or
calculate the index value by using a calculation formula of “([the evaluation value (W) of the word of the target person]+[the evaluation value (V) of the vital data of the target person])×[the evaluation value (E) of the emotion of the target person]”.
|