Abstract:
The field of Agriculture is considered as one of the main knowledge intensive domains because it is highly dependent on various aspects like environmental conditions, market movements, pest controlling, and crop management and soil conditions. These aspects are highly dynamic in nature, as a result agricultural domain itself can be considers as a complex system. With the industrial revolution, demand for food was increased rapidly and farmers used latest technological advances to improve the productivity. However when it comes to agricultural knowledge management all traditional ICT approaches have failed because of the complexity of the domain. People always struggled with the complexity of the domain and they came up with various ICT approaches to address the complexity. Finally AI based approaches found to be the promising technology in the areas of this nature because when it comes to real world problem solving, people survive within the complexity by being communicative with each other. This real life problem solving approach could be modelled as a Multi Agent System where each inter connected represent a separate agents such as Environment Agent, Market Agent, Seeds Agent, Soil Agent , Fertilizer Agent, and Pest Agent.
In this research we hypothesize that, by modelling interrelated areas as agents and by means of negotiating among each entity, it is possible to solve queries related to agricultural domain. Users such as Agriculture researchers, “Govi Jana Seva” officers, and farmers can access the system through Internet and mobile phones to place requests. Each agent module has its own ontological model and knowledge base to handle queries with the help of negotiation through the simple message passing. This thesis provides background information related to the problem domain. Based on comprehensive analysis on existing approaches and their limitations, technological and implementation methodology of the proposed multi agent system is presented. At last the technological and implementation details of the proposed multi agent system and an emphasis on the evaluation process concludes the acceptability of Multi Agent based ap