A Hybrid Model for Automatic Emotion Recognition in Suicide Notes

From Clinfowiki
Jump to: navigation, search

First Review


"We describe the Open University team’s submission to the 2011 i2b2/VA/Cincinnati Medical Natural Language Processing Challenge, Track 2 Shared Task for sentiment analysis in suicide notes. This Shared Task focused on the development of automatic systems that identify, at the sentence level, affective text of 15 specific emotions from suicide notes. We propose a hybrid model that incorporates a number of natural language processing techniques, including lexicon-based keyword spotting, CRF-based emotion cue identification, and machine learning-based emotion classification. The results generated by different techniques are integrated using different vote-based merging strategies. The automated system performed well against the manually-annotated gold standard, and achieved encouraging results with a micro-averaged F-measure score of 61.39% in textual emotion recognition, which was ranked 1st place out of 24 participant teams in this challenge. The results demonstrate that effective emotion recognition by an automated system is possible when a large annotated corpus is available."[1]


This article describes how the authors created a system that uses sentiment analysis to not only detect emotion in text, but to identify what type of emotion (i.e anger, sadness, joy,.etc.)[1]. They developed this system for the 2011 i2b2/VA/Cincinnati Medical Natural Language Processing Challenge[1]. The objective of the competition was to identify 15 specified emotion classes, which they grouped into 3 broad sentiment categories: Negative Emotions : abuse, anger, blame, fear, guilt, hopelessness, sorrow [1]. Positive Emotions : forgiveness, happiness_peacefulness, hopefulness, love, pride, thankfulness[1]. Neutral Contexts : information, instructions[1].

System Architecture

The system has five components;

  • Text pre-processing module: This module processes the notes by carrying out NLP tasks such as sentences tokenization, lemmatization, parts-of-speech tagging [1].
  • Negation detection module: This module identified instances of negation within notes. This system handled negation words (not, never, unable) as modifiers of objects,subjects or verbs[1].
  • Emotion instance identification module: This system identifies sentences that explicitly or implicitly express an emotion state that falls into any of the 15 mentioned emotion classes. This consists of two sub components; a token based identification system and a machine learning identification system[1].
  • Result Integration Module: This module merges the output results obtained from the different modules in the

emotion instance identification module[1].

  • Post-processing Module: This module identifies instances of neutral emotions, informations and instructions that may have been missed. It then applies a number of smoothing rules to identify ongoing affective contexts across a number of sentences[1].


Their system was ranked 1st place out of 24 participants and scored an F-measure score of 61.39% in textual emotion recognition[1].

Second review

Write something here!


  1. 1.00 1.01 1.02 1.03 1.04 1.05 1.06 1.07 1.08 1.09 1.10 1.11 Yang, H., Willis, A., De Roeck, A., & Nuseibeh, B. (2012). A hybrid model for automatic emotion recognition in suicide notes. Biomedical Informatics Insights, 5(Suppl 1), 17.http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3409477/