NRC-Canada: Building the State-of-the-Art in Sentiment Analysis of Tweets

By Saif M. Mohammad, Svetlana Kiritchenko, Xiao-Dan Zhu. 2013

1,103 citations recorded by the source corpus.

In this paper, we describe how we created two state-of-the-art SVM classifiers, one to detect the sentiment of messages such as tweets and SMS (message-level task) and one to detect the sentiment of a term within a message (term-level task). Among submissions from 44 teams in a competition, our submissions stood first in both tasks on tweets, obtaining an F-score of 69.02 in the message-level task and 88.93 in the term-level task. We implemented a variety of surface-form, semantic, and sentiment features. We also generated two large word‐sentiment association lexicons, one from tweets with sentiment-word hashtags, and one from tweets with emoticons. In the message-level task, the lexicon-based features provided a gain of 5 F-score points over all others. Both of our systems can be replicated using freely available resources. 1

Subfields: Artificial Intelligence, Information Systems

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

In Information Systems

Citations in the corpus

Interactive Attention Networks for Aspect-Level Sentiment Classification (2017, background): “features to train a sentiment classifier [Rao and Ravichandran, 2009; Kaji and Kitsuregawa, 2007; Jiang et al., 2011; Perez-Rosas et al., 2012; Mohammad et al., 2013].”
Transfer Capsule Network for Aspect Level Sentiment Classification (2019, methodology): “Traditional methods mostly use machine learning models with handcrafted features to build sentiment classiﬁers for ASC tasks (Jiang et al., 2011; Mohammad et al., 2013).”
Twitter Sentiment Analysis (2015, background): “Examples of the existing lexicons include: Opinion Lexicon (Hu and Liu, 2004), SentiWordNet (Esuli and Sebastiani, 2006), AFINN Lexicon (Nielsen, 2011), LoughranMcDonald Lexicon, NRC-Hashtag (Mohammad et al., 2013), General Inquirer Lexicon 3 (Stone and Hunt, 1963).”
Attentional Encoder Network for Targeted Sentiment Classification (2019, background): “, 2011), mainly focus on extracting a set of features like sentiment lexicons features and bag-of-words features to train a sentiment classifier (Rao and Ravichandran, 2009; Kaji and Kitsuregawa, 2007; PerezRosas et al., 2012; Mohammad et al., 2013).”
Large Scale Sentiment Learning with Limited Labels (2017, methodology): “We also employed a dictionary which contained sentimental hashtags [15] and counted the occurrences of positive and negative hashtags in our tweets, if any (the extraction was done over the original text of tweets, not the preprocessed ones).”
Challenges in Sentiment Analysis (2017, methodology): “Mohammad et al. (2013b) employed the Turney method to generate a lexicon (Hashtag Sentiment Lexicon) from tweets with certain sentiment-bearing seed-word hashtags such as (#excellent, #good, #terrible, and so on) and another lexicon (Hashtag Sentiment Lexicon) from tweets with emoticons.4 Since the…”
A Multilayer Perceptron based Ensemble Technique for Fine-grained Financial Sentiment Analysis (2017, background): “Sentiment, Sentiment140, Sentiment140 Context) lexicons (Kiritchenko et al., 2014; Mohammad et al., 2013) which associate a positive or negative score to a token.”
How Translation Alters Sentiment (2016, methodology): “The ablation experiments in the study by Mohammad et al. (2013) showed that the NRCCanada sentiment analysis system benefited most from the use of the NRC Hashtag Sentiment Lexicon and the NRC Emoticon Lexicon.”
Positive, Negative, or Neutral: Learning an Expanded Opinion Lexicon from Emoticon-Annotated Tweets (2015, methodology): “The same idea was used for Twitter lexicon expansion in [Becker et al., 2013], [Mohammad et al., 2013], [Zhou et al., 2014], which all model the joint probabilities from collections of tweets labelled in automatic ways.”
Gated Neural Networks for Targeted Sentiment Analysis (2016, methodology): “Traditional sentiment analysis systems rely on manual features (Pang, Lee, and Vaithyanathan 2002; Go, Bhayani, and Huang 2009; Mohammad, Kiritchenko, and Zhu 2013).”
SemEval-2014 Task 9: Sentiment Analysis in Twitter (2014, methodology): “Popular this year was also the NRC lexicon (Mohammad et al., 2013), created by the best-performing team in 2013, which is top-performing this year as well.”