Discovering the value of unstructured data in business settings

Yang, Nan (2023) Discovering the value of unstructured data in business settings. Doctoral thesis, University of East Anglia.

Preview

PDF Download (7MB) | Preview

Abstract

With the increasing amount of unstructured data in business settings, the analysis of unstructured data is reshaping business practices in many industries. The implementation of unstructured data analysis will eventually have dominant presence in all department of organisations thus contributing to the organisations. This dissertation focuses the most widely utilised unstructured data-textual data within the organisation. A variety of techniques has been applied in three studies to discover the information within the unstructured textual data. Study I proposed a dynamic model that incorporates values from topic membership, an outcome variable from Latent Dirichlet Allocation (a probabilistic topic model), with sentiment analysis for rating prediction. A variety of machine learning algorithms are employed to validate the model. Study II focused on the exploration of online reviews from customers in the OFD domain. In addition, this study examines the outcomes of franchising in the service sector from the customer’s perspective. This study identifies key issues during the processes of producing and delivering product/services from service providers to customers in service industries using a large-scale dataset. Study III extends the data scope to the firm-level data. Latent signals are discovered from companies’ self-descriptions. In addition, the association between the signals and the organisation context of the entrepreneurship is also examined, which could display the heterogeneity of various signals across different organisation context.

Item Type:	Thesis (Doctoral)
Faculty \ School:	Faculty of Social Sciences > Norwich Business School
Depositing User:	Chris White
Date Deposited:	11 Jul 2024 10:06
Last Modified:	11 Jul 2024 10:06
URI:	https://ueaeprints.uea.ac.uk/id/eprint/95868
DOI:

Downloads

Actions (login required)

View Item