Discovering entities' behavior through mining Twitter

摘要: The unprecedented amount of user generated content from emerging social media platforms like Facebook and Twitter make them invaluable sources information for research. in particular has about 500 million registered accounts globally who are generating approximately 340 messages daily containing personal updates, general life observations, opinions, moods, etc. Twitter's vast data, which is generally available, offers an ideal source mining entities' behaviors. This thesis explores two research streams involving data. In the first work, we seek to understand Twitter-based stakeholder communication strategies firms. We analyze tweets posted by firms build a system that can automatically predict target groups given tweet. also examine incorporate firm characteristics into performance improvement. result will potentially provide valuable business intelligence market analysts would discover behaviors second investigate how readers different parts world react news headlines through their messages. design framework data collection, statistical analysis, sentiment language model comparison interests reactions users towards headlines. results this work possibly help organizations have better understanding audience services. Though directions may seem distinct, there points connection. both cases, interested impact companies (firms organizations). Moreover methods used similar. Our illustrate just gathering stream developing them, able many interesting insights