Inferring Emotions from Social Images Leveraging Influence Analysis

摘要: Nowadays thriving image-based social networks such as Flickr and Instagram are attracting more people’s attention. When it comes to inferring emotions from images, previous researches mainly focus on the extraction of effective image features. However, in context networks, user’s emotional state is no longer isolated, but influenced by her friends. In this paper, we aim infer images leveraging influence analysis. We first explore several interesting psychological phenomena world’s largest image-sharing website Flickr. Then summarize these pattern into formalized factor functions. Introducing factors modeling, propose a partially-labeled graph model images. The experimental results shows 23.71% promotion compared with Naive Bayesian method 21.83% Support Vector Machine (SVM) under evaluation F1-Measure, which validates effectiveness our method.