KDD的プロセスの提示と壁要素発見アルゴリズムの検証 : 建築形態要素の発見に関する研究 (その2)

概要

論文の詳細を見る
In this paper, I define "shape element" just like a element of naural language. I propose a processing how to discover the architectural shape element based on conceptual KDD. That processing have some steps. Those steps are plan selection, making pre-data, using bipolar-scopically converting method, discovering algorythm of architectural shape element, interpretation/evaluation. There for, I explain about how to make the pre-data, about bipolar-scopically converting method, about how to discover the wall element. For get reliability to discovering algorythm, I verify by using conceptual N-gram. N-gram is used at study of natura language analysis. I verified as comparing with "static" and "dynamic". I propose a expression for calicuration perprexity. It is value of decision difficulity. perplexity=[summation the AND area of N-area]/[summation all N-area] I verified that perplexity by using 8 houses designed by Marcel Breuer. I got a result that "dynamic" is better than "static". I got good result from case study. I'll think the algorithm is to need to add script somewhat case of wall gathering status.
社団法人日本建築学会の論文
2007-12-30