Ingredient mining for food recipe production

Recipe creation is a complex problem in many SME production factories. Each product requires several kinds of raw materials and normally the precise ingredients are not known. The scientific process for finding the relationship between raw materials and product qualities increase production costs. Laboratory tests for checking these relationships in any low cost products may not be cost effective. Instead, many SMEs use their experiences in production. So, they generally have a few experts whose experiences are not transferable, mainly because of the recipe secrecy and its complexities. Thus, to help find a recipe, this project has used the decision tree approach to classify and extract knowledge from existing data in a sample factory. We built a prototype and cross checked with the factory expert and found that most recipes from the model are acceptable for production. Keywords. Manufactory, Data mining, Decision Tree, DTree, Knowledge discovery in databases (KDD).