Kyungsik Han,Sanghack Lee,Jin Yea Jang,Yong Ju Jung,Dongwon Lee
标识
DOI:10.1145/2908131.2908160
摘要
We present behavioral characteristics of teens and adults in Instagram and prediction of them from their behaviors. Based on two independently created datasets from user profiles and tags, we identify teens and adults, and carry out comparative analyses on their online behaviors. Our study reveals: (1) significant behavioral differences between two age groups; (2) the empirical evidence of classifying teens and adults with up to 82% accuracy, using traditional predictive models, while two baseline methods achieve 68% at best; and (3) the robustness of our models by achieving 76%---81% when tested against an independent dataset obtained without using user profiles or tags. Our datasets are available at: https://goo.gl/LqTYNv