FarmFriend: Your crop companion

Faculty Mentor

Sanmeet Kaur

Presentation Type

Oral Presentation

Start Date

5-7-2024 9:05 AM

End Date

5-7-2024 9:25 AM

Location

PAT 348

Primary Discipline of Presentation

Computer Science

Abstract

­­Across the world, agriculture is an important business that employs a large percentage of the workforce and is essential to the maintenance of international food supply chains. Despite its significance, many farmers find it difficult to get the vital information they need regarding their crops; these problems include communication difficulties, a lack of crop inspectors, and other issues that prevent their questions from being answered. This work presents a novel software solution that is implemented using a chatbot interface and is intended to bridge the information gap that exists between farmers and the vital data they need. This solution grants farmers easy and immediate access to vital agriculture related queries in their native language

This methodology encompasses thorough data collection, processing, and storage strategies, coupled with an innovative retrieval system that applies the Maximum Marginal Relevance (MMR) technique to balance the relevance and diversity of the information provided. We utilize advanced Natural Language Processing (NLP) techniques, including tokenization and chunking, to preprocess data, significantly enhancing the efficiency of the data retrieval and response generation process. Specifically, the framework that was used employs the Langchain library for text segmentation and the OpenAI Embeddings API for generating semantically rich vector embeddings of agricultural content. These embeddings are stored and managed using the Chroma vector store, facilitating efficient retrieval based on query relevance.

To make this chatbot more accessible and user-friendly, we have integrated text-to-speech (TTS) and speech-to-text (STT) technologies. This enables farmers to interact with the chatbot through voice commands and receive audible responses, breaking down literacy barriers and making critical agricultural knowledge more accessible to a broader audience. The integration of these technologies into a pre-trained Language Model (LLM) allows for the generation of precise, actionable responses to farmer inquiries, thereby promising to transform agricultural practices worldwide. This approach allows the farmers to access valuable information, elevating crop quality and yields, and illustrates the revolutionary potential of AI in agriculture. By proposing a scalable and innovative model, this research highlights how AI can bridge gaps in critical sectors, making essential information accessible to those who need it the most.

This document is currently not available here.

Share

COinS
 
May 7th, 9:05 AM May 7th, 9:25 AM

FarmFriend: Your crop companion

PAT 348

­­Across the world, agriculture is an important business that employs a large percentage of the workforce and is essential to the maintenance of international food supply chains. Despite its significance, many farmers find it difficult to get the vital information they need regarding their crops; these problems include communication difficulties, a lack of crop inspectors, and other issues that prevent their questions from being answered. This work presents a novel software solution that is implemented using a chatbot interface and is intended to bridge the information gap that exists between farmers and the vital data they need. This solution grants farmers easy and immediate access to vital agriculture related queries in their native language

This methodology encompasses thorough data collection, processing, and storage strategies, coupled with an innovative retrieval system that applies the Maximum Marginal Relevance (MMR) technique to balance the relevance and diversity of the information provided. We utilize advanced Natural Language Processing (NLP) techniques, including tokenization and chunking, to preprocess data, significantly enhancing the efficiency of the data retrieval and response generation process. Specifically, the framework that was used employs the Langchain library for text segmentation and the OpenAI Embeddings API for generating semantically rich vector embeddings of agricultural content. These embeddings are stored and managed using the Chroma vector store, facilitating efficient retrieval based on query relevance.

To make this chatbot more accessible and user-friendly, we have integrated text-to-speech (TTS) and speech-to-text (STT) technologies. This enables farmers to interact with the chatbot through voice commands and receive audible responses, breaking down literacy barriers and making critical agricultural knowledge more accessible to a broader audience. The integration of these technologies into a pre-trained Language Model (LLM) allows for the generation of precise, actionable responses to farmer inquiries, thereby promising to transform agricultural practices worldwide. This approach allows the farmers to access valuable information, elevating crop quality and yields, and illustrates the revolutionary potential of AI in agriculture. By proposing a scalable and innovative model, this research highlights how AI can bridge gaps in critical sectors, making essential information accessible to those who need it the most.