Data Analytics for Enhancing Decision Making and Situation Awareness in Business and Crisis Management

Date

2021

Authors

Vemprala, Naga Sainarayana

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

The increasing internet usage worldwide has resulted in increased online user engagement, such as expressing their opinions on topics ranging from e-commerce to election to vaccines, and so on. Such information is often textual (e.g., Facebook wall posts, and messages on platforms such as Twitter and blogs), and can be mined to facilitate decision making. However, the significantly large and constantly increasing volume and veracity of text data compound the challenge of analyzing such data. This dissertation presents two separate studies to demonstrate how organizations can process such text data to raise situation awareness and inform decision making. Specifically, the first study aims to extract meaningful insights that leverage platform-embedded characteristics to generate short, crisp and much needed summaries of social discussions and display the information in a highly interactive visualization tool. The study uses Twitter social media data as the testbed and proposes a streamlined approach to integrate Hierarchical Clustering for topic extraction with automated generation of short summaries of thousands of tweets. The proposed topic extraction approach uses social media-specific features such as retweet counts, number of user likes, website links, and important topic-specific keywords to score tweets based on their relevance. Then, an automated summary is generated using the scored sentences. A comparison of the performance of the proposed method and several competing approaches such as recall and precision demonstrates that the proposed approach outperforms these competing approaches. The second study aims to construct a product network through product comparisons embedded in customer reviews to predict the demand for a product on an e-commerce platform. Using a longitudinal dataset with extensive customer reviews on Amazon, the study analyzes how the various network properties can influence the competitive advantage of a given product and how structural holes and cohesion of the network can be used to predict the potential demand in a given product category. The results show that a product network derived from comparative reviews can be more effective in predicting the market performance of the products than traditional online review metrics, and that the properties of such network yield new insights on understanding the competitive dynamics in a specific market segment.

Description

This item is available only to currently enrolled UTSA students, faculty or staff. To download, navigate to Log In in the top right-hand corner of this screen, then select Log in with my UTSA ID.

Keywords

Data Analytics, eWOM, Situation Awareness, Social Network Analysis, Structural Holes, Text Mining

Citation

Department

Information Systems and Cyber Security