Arabic Farmer Chatbot.pdf

The researchers created an Arabic dataset for farmer intent identification to develop a chatbot. The dataset includes 720 records and five intents: pest or disease identification, irrigation, fertilization, weed identification, and plantation date. A Multi-Layers Perceptron (MLP) was applied for intent classification, achieving 97% accuracy with 256 neurons in each hidden layer and 10 epochs. The chatbot aims to provide farmers with quick and easy access to information, mitigating the issue of decreased extension agents in Egypt. Deep Neural Networks (DNNs) were used to classify and detect farmers' intent, with applications in agriculture including leaf disease identification, yield estimation, and weather prediction. The evolution of chatbots is discussed, from early models like ELIZA to modern AI-powered chatbots like Siri and Google Assistant. Related works on chatbots in agriculture are also reviewed, highlighting their potential to provide farmers with rapid and easy access to information. Key findings include the effectiveness of MLP in intent classification and the potential of DNNs in agriculture.