Modeling Preference Reasoning for Customizable Biological Greening Material using Bayesian Belief Network and Particle Swarm Optimization


  • Mirwan Ushada Gadjah Mada University
  • Haruhiko Murase Osaka Prefecture University Japan


This paper presents work on modeling preference reasoning regarding the uses of Customizable Biological Greening Material (CBGM). CBGM is a biological material which is produced to fulfill preference for greening technology in a given consumer segment. The objectives of the paper were: 1) To propose modeling preference reasoning for CBGM by predicting its attributes importance using consumer mentality constraints; 2) To develop the modeling by hybridizing Bayesian Belief Network (BBN) model and Particle Swarm Optimization (PSO).  These attributes are used as information sources to support decision for producing biological material in plant factory and applying its functionality in the greening technology. The inputs of modeling were various consumer mentality constraints as different demographic, their prior knowledge, familiarity, agreement to material function and interest. The output was a predicted attribute importance of a preferred material. BBN and PSO were hybridized to take advantage of both methods to identify the probability-based reasoning and maximize the satisfaction using the analogy between the consumer preference and social behavior of animal swarm. The modeling was demonstrated on a case study of moss material (Rhacomitrium canescens). The materials were offered to the respondents using questionnaires we designed. A 24 simple BBN model was used to predict each attribute importance. PSO was used to optimize a 24 simple BBN model using a satisfaction function. Hybrid modeling of BBN and PSO has indicated the performance improvement of reasoning model compared to single modeling of BBN. The improvement was based on satisfied correlation and minimum error between measured and predicted value. It was concluded that consumer mentality constraints are possible to be used as inputs to predict an attribute importance of the preferred moss material. Subsequently, hybrid modeling of BBN and PSO is a feasible method to model reasoning easily and accurately. The modeling and information in this paper are applicable to expand the application of greening technology in different consumer segments and different contexts.



Keywords:   Attribute importance, Bayesian belief network, biological material for greening technology, mentality constraint, particle swarm optimization.

Author Biographies

Mirwan Ushada, Gadjah Mada University

Faculty of Agricultural Technology, Department of Agro-industrial Technology, Laboratory of Production Systems

Haruhiko Murase, Osaka Prefecture University Japan

Graduate School of Life and Environmental Sciences, Department of Applied Life Sciences, Laboratory of Bioinstrumentation, Control and Systems Engineering, Osaka Prefecture University





VII-Information Systems