Rediscovering Social Science and Business Studies Using Web Data and the Text Mining Approach

Over the past few years, the growth of Internet and web services has produced a large amount of web data. A majority of them is in the form of text such as social media messages and online news articles. Web text data presents tremendous opportunity for business and social science research. It can provide critical intelligence to company stakeholders in strategic decision making process for containing the opinion of web users. It can often be used as a supplement to survey instrument when the cost of conducting survey is high because of its public availability. However, the high volume, high data generation speed, high level of noise, and diversified formats of web text data brought serious challenges for researchers who want to utilize it.;Text mining refers to the process of discovering hidden knowledge from unstructured text data through the use of text analytics and data mining. It is an essential technique in analyzing text data in Web 2.0's context. The general purpose of my dissertation is to demonstrate how to effectively use web text data and text mining to study the topics of interests in business and social science areas that include competitive intelligence (CI) and cultural difference. In essay 1 (chapter 2), I designed a text mining and social media analytics-based model called SoM-AGA to study culture and cultural difference from social data collected in different cultural backgrounds. In essay 2 (chapter 3), I designed an automated CI model, which can produce actionable intelligence about competitors and competitions for company executives by conducing text mining analysis on press release and third party news. In essay 3 (chapter 4), I designed a systematic approach to study CI from social media data. It makes use of text mining techniques such as sentiment analysis and IE to identify competitive threats both inside and outside of the focal firm.;The evaluation results in my three essays have shown that text mining-based models have consistent performance in studying cultural difference and CI. They consume less resources comparing with traditional models. In addition, they also help us acquire additional knowledge about the problem domain. In a larger sense, my dissertation has shown how web data and data analytics techniques such as text mining have fundamentally changed the way in which we process technology in Information Systems (IS) field.