Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.678970
Title: Mining and analysing social network in the oil business : Twitter sentiment analysis and prediction approaches
Author: Aldahawi, Hanaa
ISNI:       0000 0004 5371 0225
Awarding Body: Cardiff University
Current Institution: Cardiff University
Date of Award: 2015
Availability of Full Text:
Access through EThOS:
Access through Institution:
Abstract:
Twitter is a rich source of data for opinion mining and sentiment analysis that companies can use to improve their strategy with the public and stakeholders. However, extracting and analysing information from unstructured text remains a hard task. The aim of this research is to investigate the use of Twitter by “controversial” companies and other users. In particular, it looks at the nature of positive and negative sentiment towards oil companies and shows how this relates to cultural effects and the network structure. This has required the evaluation of existing automated methods for sentiment analysis and the development of improved methods based on user classification. The research showed that tweets about oil companies were noisy enough to affect the accuracy. In this thesis, we analysed data collected from Twitter and investigated the variance that arises from using an automated sentiment analysis tool versus crowd sourced human classification. Our particular interest lay in understanding how users’ motivation to post messages affected the accuracy of sentiment polarity. The dataset used Tweets originating from two of the world’s leading oil companies, BP America and Saudi Aramco, and other users that follow and mention them, representing Western and Middle Eastern countries respectively. Our results show that the two methods yield significantly different positive, natural and negative classifications depending on culture and the relationship of the poster of the tweet to the two companies. This motivated the investigation of the relationship between sentiment and user groups extracted by applying machine learning classifiers. Finally, clustering based on similarities in the network structure was used to connect user groups, and a novel technique to improve the sentiment accuracy was proposed. The analytical technique used here provided structured and valuable information for oil companies and has applications to other controversial domains.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.678970  DOI: Not available
Keywords: QA75 Electronic computers. Computer science
Share: