A multi-faceted and automatic knowledge elicitation system for managing unstructured information

Abstract

Nowadays, knowledge is becoming a new competitive factor in the knowledge economy. Knowledge work deals with a huge of information and its manipulation. However, many researchers point that most of all potentially usable business information originates in an unstructured form. Unstructured Information Management (UIM) is becoming the current state-of-the-art of technology. In this thesis, a Multi-faceted and Automatic Knowledge Elicitation System (MAKES) is proposed to manage the mass of unstructured information and support knowledge work. There are four phases in MAKES. The first phase is collecting data automatically and text mining. Multiple patterns of dynamic taxonomies are developed to classify the unstructured information. In the second phase, some knowledge models are adopted to represent knowledge elicited from large amounts of unstructured information. Concept Relationship Model (CRM) illustrates the relationships of concepts elicited from unstructured information. The algorithm named Concept Relationship Exploring Technique (CRET) which is developed to measure the relationship of two concepts. A pattern of knowledge flow can be captured as a Dynamic Knowledge Flow Model (DKFM). Knowledge Capability Model (KCM) evaluates the knowledge capability of a knowledge worker from the traffic of knowledge flow. Thirdly, a multi-faceted navigation platform supports the analysis of knowledge models. Finally, some reports about managing unstructured information and knowledge assets of organizations are produced derived from the knowledge models. The capability and advantages of the MAKES are demonstrated through the two cases of verification testing. The first case is an application of MAKES in an emergency management system in the Committee of Guang Zhou City Management of the Guang Zhou Municipal Government in China. The efficiency of decision-making when responding to emergency incidents is evaluated based on an evaluation architecture. Another case is applying MAKES to a knowledge audit in an electronic trading firm in Hong Kong. MAKES has significantly improved the efficiency of the knowledge audit process. This research points to a new direction of automatic text mining and to the elicitation of useful organizational knowledge from the very large amount of dynamic and unstructured information that is often neglected in an organization.