Visual Affect Around the World: A Large-scale Multilingual Visual Sentiment Ontology

TL;DR

This paper introduces a large-scale multilingual visual sentiment ontology derived from social multimedia, capturing cultural and language-specific affective concepts, and demonstrates its application and generalizability across 12 languages.

Contribution

It presents a novel multilingual, hierarchical visual sentiment ontology and a language-dependent method for automatic concept discovery in social multimedia.

Findings

Created a dataset with 15.6K sentiment-biased concepts across 12 languages

Developed language-specific models for sentiment prediction in images

Demonstrated the ontology's applicability in multilingual social multimedia analysis

Abstract

Every culture and language is unique. Our work expressly focuses on the uniqueness of culture and language in relation to human affect, specifically sentiment and emotion semantics, and how they manifest in social multimedia. We develop sets of sentiment- and emotion-polarized visual concepts by adapting semantic structures called adjective-noun pairs, originally introduced by Borth et al. (2013), but in a multilingual context. We propose a new language-dependent method for automatic discovery of these adjective-noun constructs. We show how this pipeline can be applied on a social multimedia platform for the creation of a large-scale multilingual visual sentiment concept ontology (MVSO). Unlike the flat structure in Borth et al. (2013), our unified ontology is organized hierarchically by multilingual clusters of visually detectable nouns and subclusters of emotionally biased versions of these nouns. In addition, we present an image-based prediction task to show how generalizable language-specific models are in a multilingual context. A new, publicly available dataset of >15.6K sentiment-biased visual concepts across 12 languages with language-specific detector banks, >7.36M images and their metadata is also released.