A case-based reasoning approach to fuzzy soil mapping.

摘要

This dissertation presents a case-based reasoning (CBR) approach to knowledge acquisition and knowledge-based inference for soil mapping under fuzzy logic. CBR is a technique in the artificial intelligence discipline. It uses the knowledge represented in specific cases to solve a new problem. The solution to the new problem is based on the similarities between the new problem and the available cases. With the CBR method, the soil scientist can express his or her knowledge by providing locations (cases) to indicate the association between a certain soil type and a specific landscape. To perform soil inference using these cases, the CBR inference engine first computes the similarity between the environmental configuration at a given location in the mapping area and the environmental configuration associated with each case, then use the similarity value to approximate the fuzzy membership value of the local soil at that location for the soil type represented by that case. A case study in the Pleasant Valley study area, southern Wisconsin, demonstrates the advantages of the inference results generated from this methodology over the published soil survey map produced from the conventional soil mapping process.