An Environmental Evaluation Learning System

The environmental impact assessment of projects is one of the most common and complex environmental evaluation tasks. The diversity and depth of expertise required for pose major difficulties for the development of computer systems that can help the process. Knowledge-based systems offer a potential solution by providing ways to represent some evaluation heuristics and to reason with the qualitative and quantitative information involved. However, knowledge acquisition is a major bottleneck in expert system development, especially in problems with a weak domain theory such as environmental evaluation. This paper presents EELS, an environmental evaluation learning apprentice system, which can partly overcome this difficulty by acquiring and improving evaluation heuristics through its normal use. Environmental impact estimation and comparison rules are acquired through an inductive process that classifies impact levels by extracting important commonalities and differences between previous impact assessments stored in a database. The induced rules are generalizations of the past assessments, which can be used for impact assessment predictions in situation not previously encountered by EELS.