Research Area:  Machine Learning

Abstract:

In the modern digital environment with the rise of adoption, diversification and specialization of social media,social media platforms have become a real-time,high velocity data bank of human opinions,emotions and experiences.Accordingly, organizations are increasingly depending on social media management techniques to analyse social media content to understand and serve stakeholders better and obtain a competitive advantage. However, social media data have not yet been fully investigated in leveraging for organizations to understand their stakeholders on a deeper level. This thesis explores novel opportunities and techniques presented through leveraging social media,to capture,analyse and understand human communication and interactions from an organizational perspective.The research is carried out using digital traces of human emotions,opinions and experiences available in social media. Machine learning techniques such as deep learning, natural language processing and word embedding are utilized together with a widely accepted social media framework.Established theories from psychology, marketing,business and communication such as human personality traits, brand personality and online social roles are utilized to build novel models and techniques. The contributions of the thesis include the introduction of a novel holistic model of a digital stakeholder,a framework to utilize this model to monitor stakeholder perception towards an organization across a range of organizational aspects,a framework to model, monitor and benchmark organizational brand (to understand the organization and competitors from a stakeholders perspective) and a framework to automate the ixextraction and monitoring of varied roles and levels played by digital stakeholders (to understand the stakeholder from an organizations perspective) in near real-time.The proposed models,techniques and frameworks have been trialled and the efficacyproven to generate strategical organizational insights in near real-time using a case study organization and a social media dataset of over 1.2 million textual conversations from public online discussion forums.

Name of the Researcher:  Piyumini Rasangika Wijenayake

Name of the Supervisor(s):   Prof.Damminda Alahakoon

Year of Completion:  2021

University:  La Trobe University

Thesis Link:   Home Page Url