Towards a recommender system for the ingredients of traditional ayurveda churna

Ayurveda is an Indian traditional alternative medicine system based on herbal treatment. Various medicinal preparations in the form of tablets, powder, and liquid are used for the treatment of diseases. The powder form, also known as churna is an unprocessed mixture of grinded powder prepared from one or more herbs and minerals. The ingredients of these churna can be represented by a graphical network. In this paper, the author analyses two models, Pointwise Mutual Information (PMI) based model and a graph based Louvain method to find out the complementing and disfavoring drugs from the churna dataset. Another objective is to detect the communities of ingredients from the churna dataset. The outcomes of the proposed model have been evaluated by the domain experts and the results obtained are quite satisfactory.