APPLICATION OF ALGORITHM OF HUMAN BEHAVIOUR PREDICTION FOR ESTIMATION OF COMPUTER INFLUENCE ON PERSON BEHAVIOUR TRITENESS
G.N. Sankin
sankin@hydro.nsc.ru
key words: data mining, machine learning, software for AI
The universal replacement by computers of a traditional labour instruments as "pencil and paper" puts a question about influence of digital devices not only on physical, but also on mental health of the user, especially engaged in creative activity in this or that area of the art.
Algorithm developed by the author, realised in a machine code FUTURE, permits to estimate quantitatively a triteness degree of human behaviour. It means, that one are chosing some event randomly and the program try to predict what event one will chose in the next time. With the help of the code it was supposed to find out distinction of behaviour in the work with the computer and with traditional writing-materials. The people under investigation were offered to make a sequence of pressing on the designated set of keys of the keyboard by a casual way. In addition it was offered to write by the ball-pen on the paper a casual sequence of symbols. Each symbol corresponded to one key from a set. The number of symbol was equal to number of keys in a set. After than the sequence of written symbols was entered into the computer with the help of the same keys. Test experiments were lead, in which the random generator was used as the source of outcome events. Then the results of the code's work were analysed and compared.
|
|
Fig.1 | Fig.2 |
On Fig. 1 the typical results of the code's work with the keys "Left" and "Right" of the keyboard is shown, and on Fig. 2 - with sequence of two symbols early written down on the paper, in particular "0" and "1". From the figures it is visible, that on the early stage (about 100 events) there is the self-training of a code and hence the overall performance of the code is low. At the subsequent events the number of guessed events is stabilised and changes poorly. The value for event with number 400 was taken into account. In the test experiments with the random generator the average of guessed events has not exceeded 51%. From the results of analysis of more than 10000 events for 15 patients we have concluded that in both cases percent of guessed events has appeared higher, than it was in test experiments, even with the account of non-equal possibility of a pressing/spelling of symbols. In the works with the keyboard the number of guessed events has appeared higher than with pen and paper for 80% of the patients.
Thus, influence of chosen way of information record on the triteness degree of human behaviour was shown. It is clear to accept the essential difference between the digital way of information transfer in the work with the computer and the analogue way in the work, for example, with a brush and canvas. Namely, in the first case required amount of various movements, needed for achievement of result decreases both for general number of movements and movements on fulfilment of particular action. The different feedback with the computer also brings an additional definition to behaviour of the user. It means the negative role of computer on some people in his art activity. Generalising, it is possible to approve with the help of this code, that the digital labour instruments more than traditional suppress creative ability of untraining human mind.