Facilities Management Knowledge Mapping from Text Documents: A Case Study for Using NLP for Facilities Management Knowledge Mapping

Abstract

Facility data are essential for any efficient facilities management and operations practices. However, due to the complexity and massive amount of information, interoperability is an important issue in facilities management. A knowledge map is a visual aid that can provide semantic clarity among different knowledge items and can lead directly to where the knowledge is stored. It can help facilities management staff save time when dealing with large amounts of complex data. This thesis presents a way to create a knowledge map from unstructured text documents. It first uses a natural language processing technical method to recognize knowledge items and the relations among them from the facilities management documents; it then uses Protégé, an ontology-modeling tool, to create an Ontology Web Language based knowledge map to improve the interoperability, data management, and decision making. The knowledge map was evaluated in terms of its accuracy, integrity, and readability. It was found that the relations in the map are accurate, and suggest two instances, equipment and manager, where the label content could be expanded to better reflect what a user would be looking for. The knowledge map represents facts that can be used by the inference engine and can be further developed as an expert system that can emulate the decision-making ability of a human expert.