Aaltodoc publication archive (Aalto University institutional repository)
School of Business | Department of Business Technology | Information Systems Science | 2010
Thesis number: 12430
Social media as a medium of word-of-mouth - Investigating movie eWOM dynamics using Tweets
|Title:||Social media as a medium of word-of-mouth - Investigating movie eWOM dynamics using Tweets|
|Year:||2010 Language: eng|
|Department:||Department of Business Technology|
|Academic subject:||Information Systems Science|
|Index terms:||tietojärjestelmät; information systems; sosiaalinen media; social media; viestintä; communication; markkinointi; marketing|
|Key terms:||Word-of-mouth, social media, Twitter, text mining, sentiment analysis|
RESEARCH OBJECTIVES The objective of this thesis is to explore the field of electronic word-of-mouth in social media and to provide insight how companies can aggregate, analyze and utilize WOM data collected from social media services. The topic was approached with an introduction to past WOM research and an examination of how it is transmitted in different social media services. Finally, eWOM dynamics were studied in the context of Hollywood movies using real word-of-mouth data collected from microblogging service Twitter. DATA AND METHODOLOGY A program was written to collect status updates mentioning a specified Hollywood movie title from Twitter. Data collection started two days before premiere and lasted until the end of second weekend. Two eWOM attributes were measured: volume – the amount of WOM that comes about, and valence – the measure of opinion expressed in the message. To take into account the unique characteristics of Twitter updates, the data was preprocessed to improve the data quality. RESULTS The main findings show that movie eWOM volume and box office sales figures are highly correlated. They form a feedback system, both impacting each other. Both volume and sales are highest in the first two weekends highlighting the importance of pre-release marketing. Valence seems to rather static during the measurement period, but more positive than negative. The uniqueness of Twitter data requires customized text mining application to get more accurate valence analysis.
Master's theses are stored at Learning Centre in Otaniemi.