Natural Language Processing-Driven Approach to Utilization of Unstructured Textual Data for Enhanced Project Planning

Abstract

Recent advances in computational techniques have contributed to effective project data analysis and provided invaluable insights into accurate project planning and management. The adoption of computational techniques allows project engineers to select and analyze historical project-related data stored in digital databases and conduct more accurate and reliable project planning for on-time, on-budget project execution. Many previous studies have conducted data-driven project planning based on well-organized or easy-to-analyze numeric data in structured databases. However, there have been few studies to leverage project textual data in project documents. Unstructured or free-formatted representations has posed a threat impeding the systematic usage of textual data. The recent advances in computation techniques have provided research opportunities to build models and algorithms that can extract information and knowledge represented by textual data to support project owners’ decision-makings. The purpose of this research is to develop alternative approaches for systematically analyzing unstructured textual data for advanced data-driven preconstruction project planning and management.

This research proposed systematic approaches based on three specific topics. First, the research aimed to a natural language processing (NLP)-driven model that allows for the automated extraction of critical information from unstructured documents for smart data archiving. Second, this research proposed a user-oriented approach of easily and straightforwardly retrieving desired data from a database without specialized knowledge. using advanced NLP techniques to transform human-familiar natural language queries into computer-readable queries. Third, the research developed a new process to assess project similarity based on the project scope to identify the most comparable historical projects.

The research outcomes have contributed to the efficient utilization of unstructured text data and exponentially growing databases for reliable data-driven project planning. Also, this research has provided the body of knowledge regarding the intelligent analysis of low-quality project data for smart project planning and management.