IMAVIS

Image and Vision Computing

Multimodal Sentiment Analysis and Mining in the Wild

There is a rapidly growing interest in understanding users’ intention, affect, and sentiment while generating and consuming multimedia. Text-based Sentiment analysis has shown its potentials in opinion mining in different domains. Recently, multimedia and human-computer interaction, and computer-mediated human-human conversation researchers started working on the automatic detection of sentiment expressed in visual and multimodal content. What differentiates sentiment from affect or mood is its dispositional nature. It means it exists in a person, like an opinion, whether expressed or not.

Here, we follow the current trend to face sentiment “in the wild”, i.e., out of the lab such as users in their homes, on the street, or in public spaces in all sort of varying conditions. Given the scope of the journal, submissions must include visual analysis.

Topics of interest includes but not limited to:

• Sentiment analysis from facial, vocal, and bodily expressions recorded in the wild
• Databases for training and testing
• Intelligent methods for active and efficient learning for sentiment analysis
• Efficient and reliable crowdsourcing of large sentiment and behavior data and labels
• Sentiment analysis in multimedia and interaction
• Multimedia mid-level attributes for sentiment analysis
• Sentiment analysis, empathic and socially-aware computing applications
• User affective comment prediction