Suicidal tendencies: the automatic classification of suicidal and non-suicidal lyricists using NLP

From Brede Wiki

Jump to: navigation, search
Conference paper (help)
Suicidal tendencies: the automatic classification of suicidal and non-suicidal lyricists using NLP
Authors: Matthew Mulholland, Joanne Quinn
Citation: International Joint Conference on Natural Language Processing  : 680-684. 2013 October
Editors:
Publisher: Define publisher
Meeting: International Joint Conference on Natural Language Processing
Database(s):
DOI: Define doi.
Link(s): http://www.aclweb.org/anthology/I/I13/I13-1079.pdf
Search
Web: DuckDuckGo Bing Google Yahoo!Google PDF
Article: Google Scholar PubMed
Restricted: DTU Digital Library
Services
Format: BibTeX

Suicidal tendencies: the automatic classification of suicidal and non-suicidal lyricists using NLP is a study on whether it is possible to use text mining on song lyrics to determine whether a song was written by a lyricist who later committed suicide.

[edit] Data

  • Training set: 533 songs
  • Test set: 63 songs
  • Development set: 168 songs

The lyricists was, e.g., Bob Dylan, Bob Marley, Mike Ness, Paul Simon (no suicide) and Kurt Cobain and Stuart Adamson (suicide)

[edit] Methods

Features:

  • Syntactic features withthe UAM corpus tool:
    • Passive voice
    • Personal pronouns
    • mental-state verbs
  • Semantic class feiatures
    • sensuality, action, concreteness, death, love, depression, drugs
    • MPQA
    • AFINN sentiment analysis word list as one of the features.
  • N-gram features

Machine learning classifier:

[edit] Results

Personal tools