一個以 BIM 為基礎的建築法條本體論自動建構

Abstract

隨著建築資訊模型(Building Information Modeling)發展逐漸成熟,為了加速 建築案件審查及減少人為審查的失誤,建構審核系統成為各個國家建築資訊系 統的重要研究方向,目前相關研究多以人工分析建築法律並建置審查系統。為 了能幫助建構自動審查系統,本研究提出一個整合 BIM 與本體論的建構方法。 本方法第一步將中文句子進行斷詞與剖析;第二步,針對第一步所產生出來的 樹狀結構進行分析;第三步則運用第二步所產生的結果用來建構法律本體論。 本方法可以達到兩個效果,第一個是解決因為建築物關鍵字的分析失誤,第二 個是可以用來分析句型結構複雜的建築法律條文。經過實驗分析,解析建築法 條的正確率可以由 42.85%提升至 87.5%,因此本研究應能有效地建構自動審查 系統。

In recent year, Building Information Modeling (BIM) is gaining wide acceptance. In order to reduce human errors during building examination developing an automatic examining system becomes an important research issue. To dates, most developed examination systems manually analyze building laws, followed by the construction of the system. In order to facilitate the construction of an automatic examination system, this thesis proposes a method which integrates BIM with ontology. This method has three steps. First, Taiwan’s building laws are processed by using the Chinese part-of-speech tagger and Chinese sentence parser. Secondly, the tree-structure produced in the first step is analyzed according to the structure of Chinese sentence. Finally, the ontology of building laws is generated based on the output of the second step. The proposed method has two merits: reducing the mistakes which incurred in keyword analysis; being able to analyze complicated Chinese sentence. Experiments show that the rate of accuracy of analyzing sentences is improved from 42.85% to 87.5%.