SENTIWORDNET: A Publicly Available Lexical Resource for Opinion Mining

By Andrea Esuli, Fabrizio Sebastiani. International Conference on Language Resources and Evaluation, 2006

2,934 citations recorded by the source corpus.

Opinion mining (OM) is a recent subdiscipline at the crossroads of information retrieval and computational linguistics which is concerned not with the topic a document is about, but with the opinion it expresses. OM has a rich set of applications, ranging from tracking users’ opinions about products or about political candidates as expressed in online forums, to customer relationship management. In order to aid the extraction of opinions from text, recent research has tried to automatically determine the “PNpolarity” of subjective terms, i.e. identify whether a term that is a marker of opinionated content has a positive or a negative connotation. Research on determining whether a term is indeed a marker of opinionated content (a subjective term) or not (an objective term) has been instead much scarcer. In this work we describe SENTIWORDNET, a lexical resource in which each WORDNET synset sis associated to three numerical scores Obj(s), Pos(s) and Neg(s), describing how objective, positive, and negative the terms contained in the synset are. The method used to develop SENTIWORDNET is based on the quantitative analysis of the glosses associated to synsets, and on the use of the resulting vectorial term representations for semi-supervised synset classi.cation. The three scores are derived by combining the results produced by a committee of eight ternary classi.ers, all characterized by similar accuracy levels but different classification behaviour. SENTIWORDNET is freely available for research purposes, and is endowed with a Web-based graphical user interface.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/96761dc3d8adaeff44eb9c07501eb3109802ee4b

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. Erik Cambria
2. Soujanya Poria
3. Yejin Choi
4. Saif M. Mohammad
5. Percy Liang
6. Mohit Bansal
7. Rada Mihalcea
8. B. Liu

Citations in the corpus

• Learning Sentiment Lexicons in Spanish (2012, methodology): “Since subjectivity and polarity are qualities that were shown to be most robustly expressed in English at the sense level (Wiebe and Mihalcea, 2006), we attempt to transfer the manual annotations onto the English WordNet by enforcing SentiWordNet (Esuli and Sebastiani, 2006) based constraints.”
• Towards Cross-Lingual Generalization of Translation Gender Bias (2021, methodology): “We consulted the Senti-WordNet [9] for English sentiment words, and sorted out the nouns of which the gloss starts with either ‘ A person who ’ or ‘ Someone who ’.”
• R3: Reverse, Retrieve, and Rank for Sarcasm Generation with Commonsense Knowledge (2020, methodology): “As we expect the evaluative words to be negative words, we rely on the word level negative scores obtained from SentiWordNet (Esuli and Sebastiani, 2006).”
• Enhanced Sentiment Learning Using Twitter Hashtags and Smileys (2010, background): “It was suggested that sentiment words may have different senses (Esuli and Sebastiani, 2006; An-dreevskaia and Bergler, 2006; Wiebe and Mihal-cea, 2006), thus word sense disambiguation can improve sentiment analysis systems (Akkaya et al., 2009).”
• Inducing Domain-Specific Sentiment Lexicons from Unlabeled Corpora (2016, methodology): “Dictionary-based approaches use hand-curated lexical resources—usually WordNet (Fellbaum, 1998)—in order to propagate sentiment from seed labels (Esuli and Sebastiani, 2006; Hu and Liu, 2004; Kamps et al., 2004; Rao and 2009; San Vicente et al., 2014; Takamura et al., 2005; Tai and Kao, 2013).”
• Sentiment Analysis: Detecting Valence, Emotions, and Other Affectual States from Text (2016, methodology): “Commonly used automatically generated sentiment lexicons include SentiWordNet (SWN) (Esuli & Sebastiani, 2006), Sentiment 140 lexicon (Mohammad, Kiritchenko, & Zhu, 2013), and NRC Hashtag Sentiment Lexicon (Mohammad & Kiritchenko, 2014).”
• Reducing Non-Normative Text Generation from Language Models (2020, methodology): “To address the relationship between sentiment and normativity, we sample 300 sentences from the sci-fi corpus and 300 sentences from the generation results of the trained GPT-2 model and classify them using the normative text classifier and Sen-tiWordNet (Esuli and Sebastiani, 2006).”
• A Survey of Opinion Mining and Sentiment Analysis (2012, background): “Several opinion word lists have been produced [17, 21, 31, 90, 104].”
• Opinion Mining and Sentiment Analysis (2008, background): “it/ SentiWordnet [91] is a lexical resource for opinion mining.”
• Lexicon-Based Methods for Sentiment Analysis (2011, methodology): “We cannot therefore conclude definitively that any of the SentiWordNet dictionaries is superior for our task; in fact, SentiWordNet performs better than either the Google or Maryland dictionaries, but it is still somewhat low; again, we believe it suffers from the same problem of too much coverage: P”
• SentiLARE: Linguistic Knowledge Enhanced Language Representation for Sentiment Analysis (2019, background): “Guerini et al. (2013) obtains the prior sentiment polarity by weighting the sentiment scores over all the senses of a word in SentiWord-Net (Esuli and Sebastiani, 2006; Baccianella et al., 2010).”
• Sentiment Analysis: An Overview from Linguistics (2016, positive): “Esuli and Sebastiani (2006) define the problem as having three different aspects: (i) determining the text’s subjectivity (i.e., whether the text is factual in nature or whether it expresses an opinion on its subject matter); (ii) determining the text’s polarity, or deciding if a given subjective…”
• Sentiment Analysis and Subjectivity (2010, background): “So far, several opinion word lists have been generated [19, 23, 36, 87, 98].”
• Quantifying Adaptability in Pre-trained Language Models with 500 Tasks (2021, methodology): “We then construct outputs for each input word using either WordNet or SentiWordNet.”
• Like It or Not (2016, background): “In TSA, researchers have leveraged lexicons developed for other domains such as SentiWordNet [Esuli and Sebastiani 2006] and Multi-perspective Question Answering (MPQA) Opinion Corpus [Wilson et al. 2005] to either produce scores in an unsupervised way or as additional features for training…”