Exploring Social Relationships in Text Streams

Abstract

Mining social relationships offers us an opportunity to gain insights from non-obvious relationships between individuals. Its applications can be seen in various scenarios ranging from market planning, fraud detection to the protection of national security. Most raw information related to social relationships are continuously generated by social networks in a form of text, for the reason that it has the lowest storage consumption while still possesses powerful expression abilities. However, when these continuous texts are aggregated together forming enormous text streams, applying existing data mining approaches will encounter efficiency or usability issues, either due to their overlook of the dynamic property of streams or the inapplicability of traditional store-then-process paradigm. In this paper, we specify the research gap and present a review report for dynamic text streams.