Development of an intelligent virtual assistant for digitalization of Moroccan agriculture

EMSI Marrakesh, LAMIGEP, Marrakesh, Morocco

^* Corresponding author: j.iounousse@emsi.ma

Abstract

This paper presents the design, development, and implementation of an innovative text-to-text chatbot system aimed at digitalizing the agriculture sector in Morocco, with a focus on supporting Darija-speaking farmers. The project also encompasses the curation of a comprehensive database to facilitate future fine-tuning of Speech-to-Text (STT) and Text-to-Speech (TTS) models in Darija. The project’s primary objective is the development of chatbot capable of responding to farmers’ text queries in Darija, providing them with instant access to critical agricultural information and support. Concurrently, we have curated an extensive database of Darija agricultural terminology, phrases, and dialogues, laying the groundwork for future voice-enabled interactions. Throughout the project, we have conducted thorough research into Darija linguistics and agricultural practices, followed by an in-depth development phase of the chatbot system. This included natural language processing, intent recognition, and response generation tailored to the nuances of Darija. The database curation involved extensive collaboration with agricultural experts to ensure authenticity and relevance. Looking forward, this project serves as a crucial stepping stone towards a fully voice-enabled agricultural support system in Darija. The curated database will be instrumental in fine-tuning STT and TTS models, potentially revolutionizing how Moroccan farmers access and interact with digital agricultural resources.