Food material compatibility and inter-restriction relation classification method based on neural network

The invention discloses a food material compatibility and inter-restriction relation classification method based on a neural network. The method comprises the steps that data of constitutional science of Chinese medicine is acquired to serve as text corpus; overall modelling is conducted on the acquired text corpus to generate word vectors, so that each non-stop word in the text corpus correspond to one word vector fixed in length; the cosine similarity between two word vectors is used as the similarity between entities corresponding to the two word vectors; for two given food materials, the relation characteristics of the two food materials are represented as a matrix consisting of the word vectors of representation words of food material relation; a cyclic convolution neural network is used, and the characteristics of food material relation serve as inputs of the cyclic convolution neural network to train data of manually annotated food material compatibility and inter-restriction relation. By adopting the method, the food material compatibility and inter-restriction relation can be accurately and rapidly judged, and further a food therapy recommendation system is assisted to enrich food varieties recommended by the therapy recommendation system.