BIMTag: Concept-based automatic semantic annotation of online BIM product resources

Abstract

With the rapid popularity of Building Information Modeling (BIM) technologies, BIM resources such as building product libraries are growing rapidly on the World Wide Web. However, numerous BIM resources are usually from heterogeneous systems or various manufacturers with ambiguous expressions and uncertain categories for product descriptions, which cannot provide effective support for information retrieval and categorization applications. Therefore, there is an increasing need for semantic annotation to reduce the ambiguity and unclearness of natural language in BIM documents. Based on Industry Foundation Classes (IFC) which is a major standard for BIM, this paper presents a concept-based automatic semantic annotation method for the documents of online BIM products. The method mainly consists of the following two stages. Firstly, with reference to the concepts and relationships explicitly defined in IFC, a word-level annotation algorithm is applied to the word-sense disambiguation. Secondly, based on latent semantic analysis technique, a document-level annotation algorithm is proposed to discover the relationships which are not explicitly defined in IFC. Finally, a prototype annotation system, named BIMTag, is developed and combined with a search engine for demonstrating the utility and effectiveness of our method. The BIMTag system is available at http://cgcad.thss.tsinghua.edu.cn/liuyushen/bimtag/.