Integrating users'online comment texts and evaluation levels to calculate commodity trust value is an important part of building trust mechanisms in social commerce.The LSTM neural network is used to calculate the emotional score value of the user's comment text,and it is combined with the user evaluation grade value to obtain the users'comprehensive evaluation observation value.The hidden Markov model is used to establish the observation state between the evaluation information and the trust degree.The generated probability matrix calculates the degree of trust corresponding to different evaluation observa-tions.The commodity trust value is obtained after calculating the average of the probability values of the most trusted state of the commodity.The experimental results show that the proposed SHMM model can effectively improve the accuracy of trust calculation.
维普
万方数据