Preliminary Study on the Automatic Lessons-Learned File Generator

Lessons-learned file (LLF) is commonly adopted to retain previous knowledge and experiences for future use in many construction organizations. Current practice in capturing LLF is mainly through the costly and time-consuming manual processes conducted by the construction engineers or managers. Moreover, many construction knowledge accumulated from previous projects is berried in the construction documents such as construction journals, proposals, as-built drawings, SPECs, plans, etc. It is impossible to develop LLFs from these documents manually. This paper presents the work of a preliminary attempt to develop an Automatic Lessons-Learned File Generator (ALLFG) based on text mining techniques. A prototype system is programmed. Case study is conducted to extract meaningful LLF from sample Chinese construction document automatically. Although the results are still experimental, promising potentials can be envisioned for practical applications.