报告人：Fan Hongqin（樊宏钦）博士，Department of Building and Real Estate, the Hong Kong Polytechnic University (香港理工大学建筑与房地产学系).
题目：PROJECT-BASED AS-NEEDED INFORMATION RETRIEVAL FROM UNSTRUCTURED CONSTRUCTION DOCUMENTS
内容摘要：The management and exchange of construction information has increasingly become more efficient due to the rapid and innovative developments in the field of information and communication technologies. Many large contractors have partially or completely turned to integrated construction management systems to manage their construction projects. As a result, construction documents are mainly created, exchanged, and archived in electronic format. Decision making on various management tasks, such as claims, quality control and post-construction study, is primarily based on information gathered from project documents. However, retrieving the relevant and only the relevant information from the large collection of project documents is a challenging task owing to the inherent complexity of handling unstructured natural language text. Hence, based on the theoretical foundations of information retrieval, a framework is proposed for retrieving as-needed information from project documents with due consideration to the specific features of the construction documents and the project works.
Considering traditional term frequency concept and vector space model as the baseline, a number of techniques are used in this research to improve the retrieval and ranking of documents in response to user query on project documentation. Certain typical machine learning algorithms, including Naive Bayes (probabilistic model), K-nearest neighbor (case-based learning) and decision tree (rule based learning) classifiers, are also employed to optimize the ranking of retrieved documents. Then, their effects are evaluated through a case study on a real life redevelopment building project in Hong Kong. The proposed framework is expected to contribute to information retrieval (IR) and re-use through the incorporation of semantic parsing and automatic inference of construction textual documents on the retrieval process.
报告人简介：Fan Hongqin (樊宏钦), PhD., assistant professor, award coordinator of Building Engineering and Management (Honors)programme in the Department of Building and Real Estate, the Hong Kong Polytechnic University. Dr. Fan received his Bachelor’s degree from Tsinghua University, his Master’s degree and Doctorate in construction engineering and management from the University of British Columbia, and the University of Alberta respectively. He worked with China Harbor Engineering Company for 9 years before changing to academia; His research interests cover a wide range of areas including construction equipment management, construction document management, data mining, and construction information management and knowledge discovery, etc. His research findings have been published in leading international journals such as ASCE Journal of Computing in Civil Engineering, CSCE Canadian Journal of Civil Engineering, Automation in Construction. He is a member of Hong Kong Institute of Construction Managers.