Abstract
Taste is an important aspect for the assessment of medicinal plants as such assessment helps to determine the
therapeutic property and application of medicinal plants. Traditionally, plant taste has been assessed based on human sensory
perception. This project aims at developing a machine learning (ML) model that will quantify and predict plant taste in terms
of their chemical composition. Given the dataset of chemical compounds, the model will relate a specific compound to the
known taste types: sweet, bitter, pungent, sour, salty. The methodology in this encompasses data preprocessing, feature
extraction, and supervised learning techniques over the preparation of the inference model. Major steps include training over
labeled data and validating the accuracy over cross-validation. In addition, techniques of web scraping and API integration
are used in order to expand the dataset with the properties of medicinal plants from a chemical standpoint. The last model
will be a tool for researchers and practitioners within the field of pharmacology and herbal medicine so as to have a more
rapid and more objective evaluation of plant taste, which could thus be useful in identifying new applications of medicinal
interest.