資料挖掘技術應用于闊葉樹材鑑別之可行性探討

Feasibility Study of Data Mining Application on Hardwood Identification

鐘大歡(Da-Huan Chung) ； 王灜生(Ying-Shen Wang) ； 洪國榮(Ko-Jong Hong)

《林產工業》 22卷2期 (2003/06) Pp. 89-97

摘要

木材鑑別為基於木材組織知識的一種技術。從事木材鑑別技術需要大量實務經驗的累積。經驗豐富的研究者對熟悉的木材，能依經驗僅對少數的特徵進行驗證，以進行木材鑑別工作。然而，對具備較少經驗的研究人員而言，仍需要更多的時間及精力進行之。本研究以50種闊葉樹木材特徵，對42種商用闊葉樹木材進行觀察，並將結果以資料庫儲存。使用資料挖掘（Data mining）之決策樹（Decision tree）技術，以ID3（Induction decision tree）演算法對此資料庫進行知識的挖掘。期望能找出對木材鑑別有用的知識（規則），提供不同於傳統經驗累積方式的知識，以補強某些研究人員經驗不足的問題。

關鍵字

資料挖掘 ； 決策樹 ； 木材鑑別

並列摘要

Wood identification is an experience-oriented skill based on knowledge of wood anatomy. A skillful researcher of wood identification could identify wood easily by their experienced. But, it is still costly for other researcher with less experience to identity wood. Data mining was used to retrieve unknown, implicit but useful information or knowledge from wood database consisting 42 entities with 50 wood identification features. Among various methods of data mining, ＂decision tree＂ was adopted to classify wood specimens. Rules of wood identification were extracted by ID3 algorithm to create a decision tree with useful identifying knowledge from database. The system enables researchers who with less experience to identify wood more easily by using those rules which mined from database and may be different from traditionally ones.

並列關鍵字

Data mining ； Decision tree ； Wood identification