Chatbot Diagnosis Penyakit Kulit pada Anjing Menggunakan Algoritma Long Short-Term Memory dan Fuzzy String Matching

Abstract

Dogs are loyal pets, are widely kept by humans, and are prone to skin diseases. The high public interest in maintaining dogs has not been supported by knowledge of how the first treatment can be done if the dog has a skin disease, and there are still limited supporting facilities and infrastructure, such as veterinary clinics that are easily accessible to the community. This problem is the reason for the need for media to help make an initial diagnosis of dog skin diseases and initial treatments that can be used before dog owners can meet directly with a veterinarian. Chatbot is one of the media that can be utilized in the diagnosis process. The chatbot system uses Long Short-Term Memory and Fuzzy String-Matching algorithms to diagnose dog skin diseases and provide first treatment suggestions that can be given by dog owners. The data used is 1197 data and has seven disease classes obtained from local veterinary clinics and veterinary journals. System tests conducted to measure chatbot performance get an accuracy rate of 90.32%, and for model testing, get an accuracy of 0.9375 or 93.75%.