Decision Support Method Based On Emotion Perception

Serving as a basic human cognitive activity,ubiquitous decision-making is closely linked to human life.According to the different processes performed,decision-making is generally categorized into divergent discussion decision-making and convergent discussion decision-making.It has been observed in previous studies that decision-makers tend to take risks and underestimate the possibility of risks in positive emotions while tending to avoid risks and overestimate the possibility of risks.However,there is still no effective means of acquiring decision-maker's emotion in real time at the present.Since EEG signals are difficult to disguise and of strong objectivity,it can truly express the characteristics of human internal emotional states,making EEG signals promising for more accurate emotional state perception.In this thesis,emotion perception is first conducted according to EEG signals,which is then used for decision making support in divergent discussion and convergent discussion separately.The main contributions of this thesis are as follows:(1)In order to obtain the probability distribution of the emotional states of decision makers in real time,a method of emotion perception based on SVM probability output is proposed,which first processes the EEG signals by extracting statistical characteristics.Afterwards,a three-class emotional probability output model based on SVM is established,based on which the probability distribution of the emotional states of decision-makers can be obtained.Experimental results validate the effectiveness of the method,which can obtain the probability of decision-makers' emotional states in real time.(2)According to the probability of the emotional states of the decision-makers,a divergent discussion decision support method based on emotion perception is proposed,which leverages the PageRank algorithm to design a method to calculate the degree of viewpoint support with the emotional states of decision-makers according to the decision-maker's speech information.Finally,the decision-maker's viewpoints are sorted according to the PageRank values.Experimental examples validate the effectiveness and practicability of this method.(3)According to the probability of the emotional states of the decision-makers,a convergent discussion decision support method based on emotion perception is proposed.Taking into account the emotion of decision-makers and the randomness and fuzziness of linguistic assessment,the cloud model is used to quantify linguistic assessment,and then the schemes are sorted according to prospect theory based on emotion generalization hypothesis.Experimental example demonstrates that the method can improve the effectiveness and accuracy of the convergent discussion decision.