A Research And Implementation Of Knowledge Mining And Diet Recommendation Technology For Nutrition Text

With the improvement of living standards,people not only pursue the basic needs of life,but also pay more attention to health issues.Nutrition diet is an auxiliary means of treatment and important method to prevent diseases.If nutrition diet recommendation can be automatically carried out with the aid of computer methods,it can not only reduce the work intensity and improve the work efficiency of nutritionists,but also provide a more convenient way for ordinary residents to identify their own proper eating habits to prevent diseases.The thesis introduces the construction model of a nutrition knowledge graph aimed to provide data support for a rigorous and rich diet recommendation system,defines the entities and relationships,and gives the labeling specifications and labeling examples.Based on the actual situation of target text,a suitable model of entity extraction is selected after analysis,and the nutrition-related text is extracted by LSTM-CRF.Through comparative experiment,it proves that LSTM-CRF is suitable for the entity extraction task of nutrition-related text.With the nutritional pathography data obtained from the First Affiliated Hospital of Harbin Medical University,the preliminary diet recommendation model is trained under random forest algorithm.Through comparative experiment with other classification models,it prove that random forest is suitable for the diet recommendation task based on nutritional pathography data.Based on the actual situation,the needs in dietitian’s daily work are analyzed,the nutrition pathography and diet recommendation system are designed,and the diet recommendation model is used in the nutrition pathography input and diet recommendation system.The cross-language invocation service structure Springboot+Django solves the problem of combining a good website development framework with a scientific computing program.