Determining the Sentiment of Opinions

By Soo-Min Kim, Eduard H. Hovy. 2004

1,757 citations recorded by the source corpus.

Identifying sentiments (the affective parts of opinions) is a challenging problem. We present a system that, given a topic, automatically finds the people who hold opinions about that topic and the sentiment of each opinion. The system contains a module for determining word sentiment and another for combining sentiments within a sentence. We experiment with various models of classifying and combining sentiment at word and sentence levels, with promising results.

Subfields: Artificial Intelligence

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. B. Liu
2. Erik Cambria
3. Bing Liu
4. Soujanya Poria
5. Philip S. Yu
6. Lei Zhang
7. Saif M. Mohammad
8. Rada Mihalcea

Citations in the corpus

• Text Understanding from Scratch (2015, background): “Many researchers have shown that with some random text, the inter-rater consensus on polarity is only about 60% - 80%(Gamon & Aue, 2005)(Kim & Hovy, 2004)(Strappa-rava & Mihalcea, 2008)(Viera et al., 2005)(Wiebe et al., 2001)(Wilson et al., 2005).”
• SENTIWORDNET: A Publicly Available Lexical Resource for Opinion Mining (2006, methodology): “…of opinionated content has a Positive or a Negative connotation (Esuli and Sebastiani, 2005; Hatzivassiloglou and McKeown, 1997; Kamps et al., 2004; Kim and Hovy, 2004; Takamura et al., 2005; Turney and Littman, 2003), since it is by considering the combined contribution of these terms that one…”
• Sentiment Analysis and Subjectivity (2010, methodology): “Such holders need to be identified by the system [5, 14, 46].”
• Sentiment Analysis: An Overview from Linguistics (2016, positive): “Kim and Hovy (2004) further incorporate the source, and define opinion as a quadruple, [Topic, Holder, Claim, Statement], in which the Holder believes a Claim about the Topic, and associates a Sentiment with that belief.”
• Rated aspect summarization of short comments (2009, background): “Sentiment classification is usually defined as the problem of binary classification of a document or a sentence [18, 14, 2, 7].”
• A Survey of Opinion Mining and Sentiment Analysis (2012, background): “In [48-50], researchers also built models to identify some specific types of opinions.”
• Computational approaches for mining user's opinions on the Web 2.0 (2014, background): “…(1) corpus-based approaches (e.g. Hatzivassiloglou & Wiebe, 2000; Turney, 2002; Wiebe & Mihalcea, 2006) and dictionary-based/lexicon-based approaches (e.g. Ding, Liu, & Yu, 2008; Hu & Liu, 2004a; Kim & Hovy, 2004; Popescu & Etzioni, 2005; Steinberger et al., 2012), (2) machine learning approaches.”
• Quality evaluation of product reviews using an information quality framework (2011, methodology): “Kim and Hovy [15] also usedWordNet to expand a human-composed opinion lexicon.”
• Opinion Word Expansion and Target Extraction through Double Propagation (2011, background): “Extensive work has been done on sentiment analysis at word, expression (Breck, Choi, and Cardie 2007; Takamura, Inui, and Okumura 2007), sentence (Yu and Hatzivassiloglou 2003; Kim and Hovy 2004) and document (Pang, Lee, and Vaithyanathan 2002; Turney 2002) levels.”
• A review of natural language processing techniques for opinion mining systems (2017, methodology): “Some methods have been proposed to create lexicons from some seed words through synonyms and antonyms [31, 32, 33].”
• A survey of multimodal sentiment analysis (2017, background): “Kim and Hovy [34] defined an opinion with a quadruple including topic, holder, claim and sentiment.”
• Multimodal Sentiment Analysis: A Survey of Methods, Trends, and Challenges (2023, background): “[47] presented a dictionary-based approach to determine the sentiment of opinions.”
• Are They Different? Affect, Feeling, Emotion, Sentiment, and Opinion Detection in Text (2014, background): “Sentiments and opinions have also been used interchangeably , perhaps because most NLP research on opinions has focused on detecting their subjective part, which has been referred to as sentiment (see Kim and Hovy [24]).”