Edible Mushroom Expert System
This project was made for Expert System course during my masters study. The project was accessible at https://canieatthemushroom.reflex.run/.
Although the system was supposed to supervised by the “Expert”, at the time I don’t have mushroom expert. So what I did was using a public dataset that has mushroom’s attributes as well as its status (edible or not).
Since my goal was to create an expert system (despite the classic supervised machine learning approach more reasonable), I need to somehow extract the rules from the dataset. Come PRISM, a rule-inducing algorithm, better than decision tree. Thankfully someone has already made the script for me to use.
Now that I have the rules, with small syntax adjustment to match CLIPS syntax, I have a working CLIPS-based edible mushroom expert system. Yes, we use CLIPS. It’s a rule-based programming language, C-based, intended to creating expert systems.
I also integrate a multi-modal-capable LLM (Gemini) to infer the visual features from the mushroom. The inferred feature will automatically submitted to the rule-based inference engine, and will decide whether the mushroom is edible.