Use of the vector space model in environmental scanning via the World Wide Web.

摘要

Environmental scanning is the process an organization uses to collect, analyze and use information. With the availability of vast quantities of information on the Internet, an organization has a great need for an automated methodology to scan and use this information. Additionally, the information available via the Internet is mostly text-based. Hence, the automated scanning methodology developed in this research uses the well-founded vector space model (VSM) to represent the documents available via the Internet and linear discriminant analysis to classify the documents. Chapter 1 of this dissertation provides an introduction to the environmental scanning problem in light of the current problem of gathering and using the vast quantity of information available via the Internet. Chapter 2 provides a review of the literature related to environmental scanning, the proposed methodology for solving the problem of developing an automated scanning process and the environment used to empirically test the methodology developed.; Chapter 3 describes the details of the methodology developed in this dissertation and the application environment used to empirically test the methodology. The methodology is tested by collecting news documents available via the Internet about publicly traded companies. Chapter 4 has additional details on the scanning process as well as a description of the experimental design used to empirically test the scanning process. The experimental design involves testing both a training set and a holdout sample for correct classification results. Chapter 5 presents the results. Finally, the Chapter 6 provides a summary, a conclusion and directions for future research.