Aiming at the arithmetic model and implementation technology of authenticity and falsity identification of geographical indication rice origin under the background of large data, a parallel random forest arithmetic model based on MapReduce was constructed by using hadoop distributed cluster technology on the basis of mineral element content data in rice. The results show that the discriminant accuracy of parallel random forest model is 97.55%. Compared with the linear discriminant analysis model and parallel support vector machine under the same conditions, the parallel random forest model has better discriminant accuracy. At the same time, the parallel random forest model has a growing acceleration ratio, which can achieve rapid and accurate identification of unknown area data.