衝突回避における知識構造の推定 : 状況判断教師データを用いたデータマイニング

概要

論文の詳細を見る
This paper deals with knowledge acquisition from ship operation data base to be used for ship collision avoidance. We used supervised training data that was obtained from an expert ship operator and developed for a circumstantial judgement neural network. We adopted a data mining approach, which has the advantage of finding hidden rules in the data base, and developed a new algorithm based on a linear search algorithm. We applied the new algorithm to the data base and extracted four kinds of meta-knowledge structures expressed in simple directed graphs, which can be used as a knowledge base for ship operation expert systems, intelligent interface systems, etc. The validity of these structures and possible future applications of the data mining approach to the ship operation data base are discussed.
社団法人日本航海学会の論文
1998-03-25