Institute

Studying

Research


 

IMS at EmoInt-2017, Code and Resources

Type ExperimentData
Title IMS at EmoInt-2017, Code and Resources
Author Maximilian Köper, Evgeny Kim, Roman Klinger

Description

This page contains the code and resources used by our system submission for the  WASSA Emotion Intensity Shared Task (EmoInt).  Our system (IMS) scored 2nd out of 22. The system makes use of affective norms, automatically extended norms and deep learning.

Full code and  walkthrough example can be found here:

https://github.com/koepermn/IMS-EmoInt

Automatically Extended Norms of Affective Ratings and Emotion Lexicons for ~1.6m English Words

can be downloaded here (65MB) or by contacting one of the authors. 

The file contains ratings for:  

  • Abstractness/Concreteness
  • Arousal
  • Valency
  • Dominance
  • Anger
  • Anticipation
  • Disgust
  • Fear
  • Happiness
  • Joy
  • Sadness
  • Surpris
  • Trust

The file is tab-seperated and the first column contains the word. Ratings range from [0,10].

Here are some (random) examples

Word AbsConc Arousal Valency Dominance Anger
#hate 1.4 6.697 0.693 2.514 5.742
babygiggles 6.128 3.618 9.544 6.772 3.613
pizza 8.261 4.205 7.382 6.203 1.602
#anxiety 1.615 4.849 1.823 1.607 3.796
           
           
Word Anticipation Disgust Fear Happiness Joy
#hate 2.666 5.83 2.909 0.999 1.197
babygiggles 2.814 2.503 1.946 10 7.067
pizza 1.584 2.375 1.462 7.473 0.979
#anxiety 2.402 3.308 9.484 2.295 2.945
           
           
Word Sadness Surprise Trust    
#hate 1.293 3.273 2.793    
babygiggles 3.443 6.379 2.045    
pizza 0.58 1.391 0.904    
#anxiety 3.858 5.048 3.723    

Reference

Please cite the official paper [BIB]

Maximilian Köper, Evgeny Kim, and Roman Klinger. IMS at EmoInt-2017: Emotion intensity prediction with affective norms, automatically extended resources and deep learning. In Proceedings of the 8th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis, Copenhagen, Denmark, 2017. Workshop at Conference on Empirical Methods in Natural Language Processing, Association for Computational Linguistics.

Please consider to cite in addition the publication reference for the underyling training data (see our paper for the data used per rating type). 


Download

Download of the ratings : Download (if link doesnt work, contact on of the authors)

Code and implementation: Github page