欢迎访问《图学学报》 分享到：
图学学报

图学学报

• 建筑与城市信息模型 • 上一篇    下一篇

基于属性相似性度量的 BIM 构件聚类

  

  1. (1. 中国铁道科学研究院集团有限公司电子计算技术研究所，北京 100081； 
    2. 清华大学软件学院，北京信息科学与技术国家研究中心，北京 100084)
  • 出版日期:2020-04-30 发布日期:2020-05-15
  • 基金资助:
    国家重点研发计划项目(2018YFB0505400)；国铁集团科技研究开发计划项目(K2018G055, 2017X003)

Clustering of BIM components based on similarity measurement of attributes

  1. (1. Institute of Computing Technology, China Academy of Railway Sciences Corporation Limited, Beijing 100081, China; 
    2. School of Software, BNRist, Tsinghua University, Beijing 100084, China)
  • Online:2020-04-30 Published:2020-05-15

PDF (PC)

278

摘要： 近年来，随着建筑信息模型(BIM)构件库资源在互联网上迅猛增长，对大量 BIM 构件资源的聚类和检索应用变得日益迫切。现有方法还缺乏对 BIM 构件所承载的领域信息提取， 基于 BIM 构件所承载的领域信息，对 BIM 构件库资源开展聚类研究：①针对 BIM 构件，提出 了一种基于属性信息量的 BIM 构件相似性度量算法，以充分利用 BIM 构件属性信息。通过与 传统的Tversky相似性度量算法以及几何形状相似匹配算法相比，其在相似性度量上效果更好。 ②基于 BIM 构件间的相似性度量算法，提出了一种 BIM 构件库聚类方法。并在 BIMSeek 检索 引擎中集成了 BIM 构件的关键字检索功能以及分类器查看功能，为用户提供更丰富的检索和查 看方式。通过与传统的 K-medoids 和 AP 聚类算法相比，其聚类方法效果更好。

关键词: 建筑信息模型, 工业基础类, 信息检索, 相似性度量, 聚类

Abstract: In recent years, resources in the Building Information Modeling (BIM) components library are expanding rapidly on the Internet. There is an increasing demand for ways to cluster and retrieve appropriate BIM components among countless resources. However, the way to extract domain information of BIM components still can not be found in existing methods. This paper studies the clustering of BIM components based on the domain information of BIM components: ①For BIM components, tan algorithm measuring similarity is proposed based on the attribute information. Compared with the traditional Tversky similarity measure algorithm and geometry similarity matching algorithm, the newly proposed one the present study has produced a better result. ②A clustering method of BIM component library is proposed based on the similarity measure algorithm of BIM components. Users are provided with diverse ways to retrieve and check information thanks to the search engine of BIMSeek integrated with functions of keyword-based retrieval and classifier view. Compared with the K-medoids algorithm and AP algorithm, the results of ours are more desirable.

Key words: building information modeling, industry foundation class, information retrieval, similarity measure, clustering