Abstract:
Agriculture is the basis of the Ethiopian economy. The country’s economic development depends,
in large part on sustainable improvements in agriculture However; its productivity is not kept
pace with population growth. In Ethiopia, there is scarcity of agriculture experts; the available
once may not be accessible to every farmer. By having an agricultural knowledge based system,
the problem of experts in Agriculture can be reduced. It is therefore the aim of this study to
develop a Case-based system that enable to make proper decision in the process of land and
cereal crop matching so as to select suitable cereal crops for the farm unit under cultivation.
This research was conducted following design science approach. Purposive sampling technique
was used to select 10 domain experts for knowledge acquisition. To develop land cereal crop
matching case based reasoning system, important knowledge was acquired through interview
and document analysis. The acquired knowledge was modeled using hierarchical decision tree.
jCOLIBRI and ArcGIS was used for developing case-based system (CBS). The developed CBS
provide a method in the process of land cereal crop matching proposing a solution to a new
problem or providing relevant experiences to the decision maker. GIS tools were used for
preparing, handling and generating spatial and non-spatial information as a tabular form for
CBR tools. The prototype of CBRLCCM system utilizes multiple knowledge to determine
suitable, optimal cereal crops for a farm unit. This knowledge consists of representative cases to
reflect physical, economic, environmental and social factors that affect the choice of land use for
cereal crops. Domain experts’ evaluation by visual interaction with the prototype achieves 84%
user acceptance. In addition, performance of the prototype system was evaluated using case
testing method which scores f-Measure of 76%. This system is promising to develop an
applicable system for improving the productivity of farmers by assisting agricultural expert and
development agents who advise farmers on their daily needs. However, further study should be
done to include inputs from climate prediction model so as to predict future land use choice.
