BibTeX
@MISC{Bull10automaticparody,
author = {Sarah Bull},
title = {Automatic Parody Detection in Sentiment Analysis},
year = {2010}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Sentiment analysis is here defined as a machine learning problem to analyse human documents and extract the human opinion they convey. Kanayma et al describe the field as: ‘a task to obtain writers ’ feelings as expressed in positive or negative comments, questions, and requests, by analysing large numbers of documents ’ [H Kanayama and Watanabe, 2004]. Much of the work on automated sentiment analysis is relatively recent and has focused upon explicit sentiment such as ‘I like ’ or ‘I hate’. This form of sentiment can be analysed using simple lexicons of positive or negative words and phrases. Little work in sentiment analysis focuses on more complex domains of sentiment such as parody and sarcasm which require the use of more machine learning techniques. This thesis contributes to the literature on sentiment analysis firstly by providing an overview of the field, and secondly by practical experiments and feature engineering in the specific sentiment domain of parody versus non parody. The experiments
Keyphrases
sentiment analysis automatic parody detection machine learning problem human document feature engineering explicit sentiment simple lexicon negative comment complex domain parody versus non parody little work large number document kanayama specific sentiment domain automated sentiment analysis negative word machine learning technique human opinion practical experiment writer feeling
