An integrated ontology and multi-objective optimization model for intercropping decision support systems

Intercropping in rubber plantations is a resource allocation problem to achieve Multiobjective Optimisation (MOO), including higher earnings, low investment, and using less water for cultivation. Moreover, the MOO model must be suitable for area conditions. Since each farmer has different conditions, areas, and limitations, therefore the decision process must involve Multiple Criteria Decision-Making (MCDM) approach. This study proposed a new Decision Support System (DSS) model with integrated ontology and Goal Programming model (GP) for solving the MOO intercropping plantation problem by selecting crops that are suitable for each farmer and recommended by experts. The methodology used an ontology with the triangulation method and MOO modelling development. The DSS is designed to work in (i) a plant introduction section and (ii) a multi-purpose resource allocation section. The findings revealed that: (i) the Intercropping in Rubber Plantation Ontology and Recommendation Rules Section provided recommendations with accuracy up to 91.00% and precision equal to 95.79 % based on three rubber planting experts; (ii) the list of plants from the ontology base approved as a decision variable in the GP if the system recommended more than one plant is suitable to allocate resources and constraints of each farmer to achieve the above three objectives; and (iii) to adjust the scale of the values target error from different objectives units of measurement, a
minimum percentage of deviation summation is used. The model has improved decision-making by solving the problem of plant selection. This complex MCDM problem requires consideration of both quantitative and qualitative criteria and solving the problem of area allocation that meets the MOO requirements. The intercropping ontology has covered the modelling weakness in a selection problem through a hierarchical structure alone. In addition, the model can be applied to solve many other
resource allocation problems with MCDM and MOO.