| Food safety incidents happened frequently recent years, which attracts the attention of people around the world. The Internet of Agricultural Things was proposed to trace the food information in whole food supply chain. Thanks to this technology, customers are able to get amount of real information about the product they bought. However, the amount of the information about the products is too large for a customer to face with. It is really difficult for an ordinary customer to choose a trustworthy product by analyzing all information about products.Our paper designs a novel trust model to solve this problem. It is data-driven and can run without expert help during trust evaluation. The model will offer the customer a simple trust value which based on amount of information about the product. When we apply the trust model into the Internet of Agricultural Things, we take time, location, signature, chain, coverage and reputation as impact factors. We calculate each factor’s value, which is in the range of [0,1], then use a Bayesian network to get the trust value of the product. Secondly, we realize the trust model with the help of an electronic pedigree system. We get all the electronic pedigrees about the products which need to be evaluated from the electronic pedigree system. Then, we extract and analyze information to evaluate each impact factor’s value. A Bayesian network with parameters is applied to evaluate the trust value of the product. An example of pork is proposed to explain the application of the trust model in the Internet of Agriculture Things. At last, we do several experiments to study the influence of each factor on the trust value and the influence of probabilities in the Bayesian network on the trust value. We also compare our trust model with others. |
