Viewed by too many or viewed too little: Using information dissemination for audience segmentation. Issue 1 (24th October 2017)

• Record Type: Journal Article
• Authors: Jansen, Bernard J.; Jung, Soon‐Gyo; Salminen, Joni; An, Jisun; Kwak, Haewoon
• Is Part Of: Proceedings of the Association for Information Science and Technology. Volume 54:Issue 1(2017)
• Page Start: 189
• Page End: 196
• Publication Date: 2017-10-24
• Subjects: Social Media Analytics -- User Experience Research -- Data‐driven design -- Data Science -- User Analytics -- Market Segmentation; Information science -- Congresses; Information technology -- Congresses; Information science; Information technology; Conference papers and proceedings; 020
• DOI: 10.1002/pra2.2017.14505401021
• Languages: English
• ISSNs: 2373-9231

Abstract

ABSTRACT: The identification of meaningful audience segments, such as groups of users, consumers, readers, audience, etc., has important applicability in a variety of domains, including for content publishing. In this research, we seek to develop a technique for determining both information dissemination and information discrimination of online content in order to isolate audience segments. The benefits of the technique include identification of the most impactful content for analysis. With 4, 320 online videos from a major news organization, a set of audience attributes, and more than 58 million interactions from hundreds of thousands of users, we isolate the key pieces of content in terms of identifying audience segments that are both (a) least and most discriminating in terms of audience segments and (b) the least and most impactful. By empirical methods, we show that 25.3 percent of the videos are so widely disseminated (i.e., viewed by so many different segments) that they are non‐discriminatory, while 29.7 percent of the videos are very discriminatory (i.e., can clearly identify one or more audience segments) but their impact is marginal, as the user base is small. Implications are that there are critical values that can be identified to isolate the set of both distinct and impactful content in a given data set of online content. We demonstrate the utility of this line of analysis by using the approach to identify critical cut‐off values for dynamic persona generation.