Developing A Case Based Reasoning System For Urban Land Development Control In Ethiopia: The Case Of Jimma City.

Abstract

Ethiopia is one of the most rapidly urbanizing countries in Africa and the challenges that come with this, especially in the provision of adequate vacant urban lands for housing is a major challenge that government faces. In Ethiopia the options available to acquire urban land to be held by leasehold include tendering and allotment. However; the focus of the study is on land acquisition through the modality of tender. Despite the various efforts of government, individuals and agencies both regionally and nationally to improve urban development and particularly those associated with urban land use conversion and management, land use problems particularly service delivery from the administration side still persist. So Case-Based Reasoning is promising to build the decision support system for urban land development control. It uses previous similar case(s) to help solve, evaluate, or interpret a new problem. The aim of this research was therefore, investigating how to develop a prototype case based reasoning system that can give decision support in urban land development control that are acquired through a modality of tender. For the development of the prototype system design science research method was performed by collecting 65 successful and unsuccessful previous cases from Jimma city land development and administration agency. The main attributes and values for the cases were identified and selected with the consultation of domain experts. After the acquired knowledge is modeled using hierarchical conceptual modeling method, cases were generated and represented with feature-value format. For the development of the prototype system, jCOLIBRI implementation tool and nearest neighbor retrieval algorithm were used. Evaluation of the system was done for both system performance and user acceptance. For testing of the prototype seven test cases and five domain experts were used. Based on the performance of the system, the average precision and recall values achieved are 70% and 83% respectively. User acceptance testing also performed by involving domain experts and an average of 83% acceptance is achieved. Although the results of this study are promising, there are challenges that need further investigation for future work. Therefore based on this challenge, efficient machine learning approach that can learn from the data after training and investigation on hybrid approach such as rule based reasoning is recommended.