From Brede Wiki

Jump to: navigation, search
Data set (help)
Category: AFINN

Sentiment analysis

Papers: DOAJ Google Scholar PubMed (Open J-Gate)
Ontologies: MeSH NeuroLex Wikidata Wikipedia
Other: Google Twitter WolframAlpha

AFINN is an affective lexicon by Finn Årup Nielsen. Previous and present versions of the word list are available in a zip file:

The word list and its evaluation are described in:

A new ANEW: evaluation of a word list for sentiment analysis in microblogs


[edit] Relationship to ANEW

Although the title of the associated paper suggests that it is based on the ANEW labeled corpus, it is not. The title is simply a wordpun. It was developed independently of the wordlist, and it is not a revision of it. Compared to ANEW, the AFINN word list has more words and includes obscene words. ANEW on the other hand has (besides valence) arousal and dominance for each word and each word has been labeled by several persons and the mean and standard deviation are given. The AFINN was only labeled by Finn Årup Nielsen. Finn Årup Nielsen was in no way involved in development of ANEW. ANEW was developed by Margaret M. Bradley and Peter J. Lang.

The word list has been used for sentiment analysis and is developed in the Responsible Business in the Blogosphere project.

[edit] Evaluation

  1. Correlation with Alan Mislove's 1000 AMT-labeled tweet post, see A new ANEW: evaluation of a word list for sentiment analysis in microblogs
  2. Correlation with 50 positive and negative manually labeled tweets: 76%-78%, see Analyzing emotion on Twitter for using modeling
  1. Around 0.55 in three-class accuracy on RepLab Twitter data set. However, this is a combined list consisting of AFINN, SentiWordNet and Liu wordlist used as features in a decision tree machine learning classifier.[1]
  2. 0.72-0.77 accuracy in two-class polarity classification with Twitter corpora.[2]
  3. 0.65-0.79 accuracy in two-class subjectivity classification with Twitter corpora.[3]
  4. Performance on product reviews is reported in Data-driven integration of multiple sentiment dictionaries for lexicon-based sentiment classification of product reviews.
  5. 71% and 63% accuracy on two-class polarity classification on Twitter data.[4]
  6. 62.2% accuracy on two-class sentiment analysis of the movie review data set

[edit] See also

[edit] Papers

  1. A new ANEW: evaluation of a word list for sentiment analysis in microblogs. Evaluation of the word list with 2477 words.

[edit] Applications of the word list

  1. 2012 presidential elections on Twitter - an analysis of how the US and French election were reflected in tweets
  2. A clustering analysis of tweet length and its relation to sentiment
  3. A framework for knowledge derivation incorporating trust and quality of data (2013)
  4. A fine-grained sentiment analysis approach for detecting crisis related microposts (2013)
  5. A longitudinal study of follow predictors on Twitter (2013)
  6. A novel transit rider satisfaction metric: rider sentiments measured from online social media data
  7. Aesthetic considerations for automated platformer design (2012)
  8. Aspect based sentiment analysis (2014, PhD Thesis)
  9. Audience targeting by B-to-B advertisement classification: a neural network approach
  10. AUEB: two stage sentiment analysis of social network messages (2014, SemEval)
  11. BOUNCE: sentiment classification in Twitter using rich feature sets (2013)
  12. Building sentiment lexicons applying graph theory on information from three Norwegian thesauruses
  13. Capturing place semantics from users' interaction on the geosocial web (2014)
  14. Catching fish in the stream: real time analysis of audience behavior in social media (2013)
  15. Character-to-character sentiment analysis in Shakespeare's plays (2013)
  16. Citius: a naive-Bayes strategy for sentiment analysis on English tweets (2014)
  17. Combining strengths, emotions and polarities for boosting Twitter sentiment analysis
  18. Combining textual pre-game reports and statistical data for predicting success in the National Hockey League (2014)
  19. Contextual sentiment analysis using conditional random fields
  20. Contradiction detection between opinions (2013)
  21. Cooperative, dynamic Twitter parsing and visualization for dark network analysis (2013)
  22. Crawling JavaScript websites using WebKit – with application to analysis of hate speech in online discussions (2013)
  23. Crowd sentiment detection during disasters and crises
  24. Cues to deception in social media communications
  25. Data-driven integration of multiple sentiment dictionaries for lexicon-based sentiment classification of product reviews
  26. Discovering content-based behavioral roles in social networks
  27. Enhanced news-reading: interactive and visual integration of social media information
  28. Enhancing lexicon-based review classification by merging and revising sentiment dictionaries
  29. Exploratory search on Twitter utilizing user feedback and multi-perspective microblog analysis
  30. Extracting sentiment networks from Shakespeare's plays
  31. FBM-Yahoo! at RepLab 2012 (2012)
  32. Forex-Foreteller: currency trend modeling using news articles
  33. Full-FACE poetry generation
  34. Fuzzy subjective sentiment phrases: a context sensitive and self-maintaining sentiment lexicon (2014)
  35. Geo-spatial event detection in the Twitter stream
  36. Good friends, bad news - affect and virality in Twitter (2011)
  37. Identifying consumers' arguments in text (2012)
  38. Identifying disputed topics in the news (2014)
  39. In search of reputation assessment: experiences with polarity classification in RepLab 2013 (2013)
  40. Inferring topic-dependent influence roles of Twitter users
  41. KLUE: simple and robust methods for polarity classification
  42. language-related information for the Linguistic Linked Data Cloud
  43. Measuring the influence of mass media on opinion segregation through Twitter (2014)
  44. Meta-level sentiment models for big social data analysis
  45. Mining Facebook data for predictive personality modeling
  46. Multi-domain collaboration for Web-based literature browsing and curation
  47. Networks and language in the 2010 election
  48. New lessions from the War in Bosnia
  49. NTNU: domain semi-independent short message sentiment classification
  50. Predictingg tie strength with the Facebook API (2014)
  51. Privacy nudges for social media: an exploratory Facebook study (2013)
  52. Probing of geospatial stream data to report disorientation (2013)
  53. Real-time monitoring of sentiment in business related Wikipedia (2013)
  54. Rule-based visual mappings -- with a case study on poetry visualization (2013)
  55. Representing and resolving negation for sentiment analysis (2012)
  56. Sarcasm as contrast between a positive sentiment and negative situation (2013)
  57. Semi-automated argumentative analysis of online product reviews (2012)
  58. Sensing social media for corporate reputation management: a business agility perspective
  59. Sentiment classification of online political discussions: a comparison of a word-based and dependency-based method (2014)
  60. Some remarks on the internal consistency of online consumer reviews (2013)
  61. Someone to talk to in Advances in affective and pleasurable design.
  62. Suicidal tendencies: the automatic classification of suicidal and non-suicidal lyricists using NLP (2013)
  63. Summarization of yes/no questions using a feature function model (2011)
  64. The emergence of online community leadership
  65. The FloWr framework: automated flowchart construction, optimisation and alteration for creative systems (2014)
  66. The impact of affective verbal content on predicting personality impressions in YouTube videos
  67. The lexicon-based sentiment analysis for fan page ranking in Facebook (2014)
  68. The power of Twitter on predicting box office revenues (2012)
  69. The QWERTY effect: how typing shapes the meanings of words (2012)
  70. Towards automated personality identification using speech acts
  71. Topic and sentiment analysis on OSNs: a case study of advertising strategies on Twitter
  72. Trusting smartphone apps? To install or not to install, that is the question (2013)
  73. Tuned models of peer assessment in MOOCs (2013)
  74. Tweeting the meeting: an in-depth analysis of Twitter activity at Kidney Week 2011 (2012)
  75. Tweets are forever: a large-scale quantitative analysis of deleted tweets (2013)
  76. Twitter for public health: an open-source data solution
  77. US presidential election 2012 prediction using census corrected Twitter model (2012)
  78. Using hashtags as labels for supervised learning of emotions in Twitter messages (2014)
  79. Valence shifting: is it a valid task?
  80. Voices of victory: a computational focus group framework for tracking opinion shift in real time
  81. Whisper: tracing the spatiotemporal process of information diffusion in real time (2012)

[edit] Mentioning

  1. A machine learning attack against the civil rights CAPTCHA
  2. A novel sentiment analysis of social networks using supervised learning
  3. Adapting sentiment lexicons using contextual semantics for sentiment analysis of Twitter
  4. An integrated text analytic framework for product defect discovery (2014)
  5. Automatic mood classification of Indonesian tweets using linguistic approach
  6. Crawling JavaScript websites using WebKit - with application to analysis of hate speech in online discussions
  7. Demystifying MapReduce
  8. Detection and scoring of Internet slangs for sentiment analysis using SentiWordNet
  9. Engagement with health agencies on Twitter
  10. GU-MLT-LT: Sentiment Analysis of Short Messages using Linguistic Features and Stochastic Gradient Descent
  11. Increasing the willingness to collaborate online: an analysis of sentiment-driven interactions in peer content production (2011, closed access)
  12. NTNU: domain semi-independent short message sentiment classification
  13. Privacy nudges for OSNs: a review (2014)
  14. Reducing sparsity in sentiment analysis data using novel dimensionality reduction approaches
  15. SemEval-2013 task 2: sentiment analysis in Twitter
  16. SentiMeter-Br: Facebook and Twitter analysis tool to discover consumers' sentiment
  17. Social data integration and analytics for health intelligence
  18. Text mining of news-headlines for FOREX market prediction: a multi-layer dimension reduction algorithm with semantics and sentiment
  19. The muses of poetry - in search of the poetic experience
  20. Time-space varying visual analysis of micro-blog sentiment
  21. The next generation poetic experience
  22. The potential of microblogs for the study of public perceptions of climate change
  23. VADER: a parsimonious rule-based model for sentiment analysis of social media text
  24. Workshop on computational personality recognition: shared task

[edit] Student papers

  1. Análisis Estático y Dinámico de opiniones en Twitter (Felipe José Bravo Márquez, 2013)
  2. Analyzing emotion on Twitter for user modeling (Shaolong Li, Master Thesis, 2013)
  3. Classification and visualisation of Twitter sentiment data (mentioning, Mikael Brevik, Øyvind Selmer, 2013 Master Thesis)
  4. Distributional methods for sentiment analysis (Emmanuele Chersoni, 2013, Master Thesis)
  5. Evaluation of natural language processing techniques for sentiment analysis on tweets (2012, Bachelor Thesis)
  6. Exploitation of tweets to measure experienced utility (Allison Madigan, 2013)
  7. Forecasting success in the National Hockey League using in-game statistics and textual data (Master Thesis, 2014)
  8. Natural language processing methods for attitudinal near-synonymy
  9. Opinion mining and name entity detection from news comments
  10. Relationships between social media sentiment, customer satisfaction, and stock price performance (Master Thesis, 2013)
  11. Retweets--but not just retweets: quantifying and predicting influence on Twitter (2012, Bachelor thesis)
  12. Social media sentiment analysis (2014 Master Thesis)
  13. Using movie screenplays as a barometer for mood in history (2014, [1])
  14. Using social media content to inform agent-based models for humanitarian crisis response (Bachelor Thesis, 2014)
  15. Why not! Sequence labeling the scope of negation using dependency features (2012 Master Thesis)
  16. Sentiment analysis of microblogs (Tobias Günther, 2013)

Andrew Ng's CS 229 class:

  1. Robert Chang, Sam Pimentel, Alexandr Svistunov, Sentiment Analysis of Occupy Wall Street Tweets, 2011. [2]
  2. Derek Farren, Predicting retail website anomalies using Twitter data, 2012. [3]


  1. A framework to analyse and visualise public sentiment using Twitter data
  2. Inference of personality using social media profiles
  3. Mining social media: methods and approaches for content analysis
  4. Sentiment analysis for Bangla microblog posts

[edit] Blogs

  1. Finn Årup Nielsen's blog, posts tagged with 'afinn'.
  2. Tracking US Sentiments Over Time In Wikileaks
  3. Kaushal Agrawal – Data Visualization – Mood of the Artist
  4. First shot: Sentiment Analysis in R, Andy Bromberg
  5. All Your Tweets Are Belong To Us: the Twitterverse declares a winner

[edit] Mentioning

  1. Sometimes I think we don’t deserve good data Google Ngram.
  2. Introduction to Sentiment Analysis , Carl Anderson,

[edit] Tools

  1. Simplest sentiment analysis in Python with AFINN (note UNICODE issue for the word naïve, use "unicode(w, 'utf-8')" )
  2. Javascript
    1. AFINN-based sentiment analysis for Node.js
    2. application
    3. Retext sentiment
  3. lexicons, Python and Javascript libraries
  4. AFINN-based sentiment analysis in Perl
  5. Afinn-for-Norsk
  6. Django
  7. Common Lisp
  8. troll, Javascript, Andrew Sliwinski.
  9. C-sharp (C#) by Tomasz Cielecki
  10. Große Gefühle Heise c't magazine in German with source code and tools in Ruby and Java.
  11. SAS Text Mining ("what's new" for the commercial product)
  12. A Better Place, chrome extension. The extension was made in response to Experimental evidence of massive-scale emotional contagion through social networks.

[edit] Services

[edit] Coursera

Bill Howe's Coursera course Introduction to Data Science has a sentiment analysis task were - apparently - AFINN is used on Twitter posts

[edit] References

  1. In search of reputation assessment: experiences with polarity classification in RepLab 2013
  2. Meta-level sentiment models for big social data analysis
  3. Meta-level sentiment models for big social data analysis
  4. Fuzzy subjective sentiment phrases: a context sensitive and self-maintaining sentiment lexicon
Personal tools