Computational & Technology Resources
an online resource for computational,
engineering & technology publications
PROCEEDINGS OF THE TENTH INTERNATIONAL CONFERENCE ON CIVIL, STRUCTURAL AND ENVIRONMENTAL ENGINEERING COMPUTING
Edited by: B.H.V. Topping
A Hybrid Soft Computing Approach for Knowledge Discovery in Construction Engineering
W.D. Yu and G.W. Fan
Institute of Construction Management, Chung Hua University, Hsinchu, Taiwan
W.D. Yu, G.W. Fan, "A Hybrid Soft Computing Approach for Knowledge Discovery in Construction Engineering", in B.H.V. Topping, (Editor), "Proceedings of the Tenth International Conference on Civil, Structural and Environmental Engineering Computing", Civil-Comp Press, Stirlingshire, UK, Paper 18, 2005. doi:10.4203/ccp.81.18
Keywords: data mining, knowledge discovery in databases, soft computing, neuro-fuzzy systems.
Construction has been conceived as an experience-based discipline ; therefore, knowledge acquired from previous works plays a key role for the successful performance of new projects. Not only the construction know-how's of the contractors, but also the design capabilities of the design firms and the management skills of construction management consultants rely heavily on such knowledge. This has made construction an ideal industry for the knowledge-based economy. In the past two decades, tremendous efforts have been contributed to the formation and application of construction knowledge provided by experienced engineers and managers to new construction projects. However, modern KDD (knowledge discovery in databases) or DM (data mining) technologies have not yet been widely exploited and adopted in the field of construction engineering and management to acquire valuable knowledge from previous projects. This results in the leaking of knowledge from construction firms. This is due to two main causes: (1) the construction industry is not familiar with KDD and DM technologies [2,3]; (2) the existing KDD and DM technologies do not fit the special characteristics of data in the field of construction engineering and management .
For the construction industry to pursue a knowledge-based economy, obstacles caused by the above two reasons must be removed and the reusable domain knowledge must be generated from historical data. To this end, this paper tackles problems encountered in knowledge discovery in real world construction databases. The focuses are: (1) development of DM algorithms for the knowledge discovery of unique construction data characteristics; (2) generation of human understandable knowledge, so that domain experts can visualize and verify it. At first, the existing KDD  and DM [5,6] methods are reviewed. Problems faced in applications of KDD and DM for construction engineering and management are broadly surveyed to identify the special characteristics of construction data, which hinder the implementation of KDD and DM in the construction industry. The existing soft computing techniques, including fuzzy sets , artificial neural networks , genetic algorithms , and case-base reasoning , are reviewed to propose the most appropriate hybridization for handling unique domain data characteristics. The data mining algorithms are developed to discover knowledge from construction data, which are usually uncertain, incomplete, partially true, and scarce in their nature. A Hybrid Soft Computing System is developed for implementation of data mining and knowledge discovery in the construction industry.
Various real world databases provided by the industrial partners are used for validation and verification of the proposed system. The proposed hybrid soft computing approach provides an effective tool to various disciplines for KDD implementation and business intelligence building in the field of construction and civil engineering.
purchase the full-text of this paper (price £20)