Talks from the workshop can now be found here


Traditional research on different forms of media analysis has been primarily focused on the low and mid-level tasks, such as indexing and summarization. With the availability of big data and powerful learning techniques, the human factors in media have gained traction, such as the affective content, insight generation, societal impact and trends. This workshop on Media Analytics for Societal Trends (MAST) comes at a critical time when interest in human-centric media analytics is on the rise. To this end, we solicit original papers related (but not limited) to the topics listed below.

  • Impact of media
    • Quantifying and analyzing media impact at individual, group and societal level
  • Affect in media
    • Affect prediction analysis
    • Expressed vs. perceived affect in multimedia
    • Therapeutic uses of media forms
  • Computational narratology
    • Understanding narratives, tropes and character portrayals
    • Computational analysis of storytelling
    • Interaction among characters in multimedia content
  • Methodologies
    • Data mining for labeling media data
    • Semi-supervised and self-supervised learning


  • [1st] ICME 2018, San Diego
  • [2nd] ICMI 2019, Suzhou, China

Workshop organizers

Workshop Program

  • Workshop date : Oct. 12, 2020
  • Time : 2 pm - 5 pm PST
  • Talks : link

Time Title Speaker
2:05 pm Introduction
2:20 pm Energy from 70mm – Using films as a resource for quantifying energy behaviour Ronita Bardhan, University of Cambridge
2:40 pm Exploring Speech Cues in Web-mined COVID-19 Conversational Vlogs Kexin Feng, Texas A&M
3:00 pm Break
3:10 pm Decoding and Advertising Emotional Ads-- The Science & The Engineering Ram Subramanian, IIT Ropar
3:30 pm Understanding Gender Stereotypes and Electoral Success from Visual Self-presentations of Politicians in Social Media Kunwoo Park, UCLA
3:50 pm Machine Understanding of Social Situations Makarand Tapaswii, INRIA
4:15 pm Panel discussion Prof. Theodora Chaspari and Dr. Makarand Tapaswi: Media understanding and the role of vision and speech technologies

Invited Speakers

Dr. Ram Subramanian, IIT Ropar

Decoding and Advertising Emotional Ads-- The Science & The Engineering
Advertising is a pivotal industry in today’s digital world, and emotions play a crucial role in advertisements (ads) conveying an effective message to viewers. Therefore, the ability to objectively characterize video ads in terms of their emotions has multiple applications– e.g., inserting appropriate ads at optimal temporal points within a video stream can benefit both advertisers and consumers of video streaming websites. My talk will focus on employing content and user-centric methods to decode ad emotions, and aspects of ad design that contribute to conveying these emotions. Finally, I will outline work on placing emotional ads within an affective program, and whether certain placements are better than others.

Dr. Ronita Bardhan, University of Cambridge

Energy from 70mm – Using films as a resource for quantifying energy behaviour
The key to predicting the future energy demand lies in understanding the how the user interacts with energy while carrying out the daily activities. This is especially stochastic for domestic energy consumption. Yet approaches to effectively capture the everyday life in domestic environment is resource intensive and requires extensive technical knowhow on energy sensors, which often limit the knowledge expansion. This talk presents an innovative and experimental approach of using fictional films as a data resource to generate data about energy culture and energy use schedules, which are essential in energy prediction. Taking the case of chawls (the typical low-income housing) from Mumbai, Bollywood films that depict/document the everyday life of such resource constraints communities were chosen for the study. A mixed-mode research method that combines qualitative knowledge with data-driven techniques were used to infer on domestic energy use, device ownership and energy culture in poverty.

Dr. Makarand Tapaswi, INRIA

Machine Understanding of Social Situations
There is a growing interest in AI to build socially intelligent robots. This requires machines to have the ability to read peoples' emotions, motivations, and other factors that affect behavior. Towards this goal, I will discuss some of our recent work. I will introduce MovieGraphs, a dataset that provides detailed, graph-based annotations of social situations depicted in movie clips. I will show how common-sense can emerge by analyzing various social aspects of movies, and also describe applications ranging from querying videos with graphs, interaction understanding via order, and reason prediction. I will also briefly present our recent work on analyzing whether machines can recognize and benefit from the joint interplay of interactions and relationships between movie characters.

Program Committee Members

  • Abhinav Dhall, Monash University
  • James Kennedy, Disney Research
  • Jangwon Kim, Alexa Speech
  • Makarand Tapaswi, INRIA Paris
  • Samuel Kim, Canary Speech
  • Subarna Tripathi, Intel AI