Computer Science Seminar: The search for emotions in language

Date Tuesday 10 July 2018
Time 11am - 12pm
Where G.1.15
Presenter Saif M. Mohammad, Senior Research Scientist, National Research Council Canada
Contact Felipe Bravo Márquez
Contact email
Admission Cost Free

Emotions are central to human experience and behavior. They are crucial for organizing meaning and reasoning about the world we live in. They are ubiquitous and everyday, yet complex and nuanced. In this talk, Saif will describe work on the search for emotions in language -- by humans and by machines.

Saif will describe large crowdsourced studies asking people to detect emotions associated with words, phrases, sentences, and tweets. He will flesh out the various ways in which emotions can be represented, challenges in obtaining reliable annotations, and approaches that address these problems.  The lexicons thus created, with entries for tens of thousands of English terms, provide scores along the dimensions of basic emotions as well as valence, arousal, and dominance (argued by some to be the core dimensions of meaning). The lexicons have wide-ranging applications in natural language processing, psychology, social sciences, digital humanities, data sonification, and art. Saif will highlight some of the applications they have explored in literary analysis and automatic text-based music generation.

In the second part of the talk, Saif will show how word-emotion lexicons are crucial in detecting emotions in sentences and tweets. This includes the use of the lexicons in our NRC-Canada system that stood first in three SemEval-2013 and SemEval-2014 sentiment analysis shared task competitions, as well as in neural approaches in emotion detection that came first in recent shared tasks they organized. These new tasks include inferring stance from tweets and detecting fine-grained emotion intensity. Finally, Saif will conclude with work on assessing the extent of inappropriate biases in automatic emotion systems.

Sanitized HTML from output session variable