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

From Brede Wiki
Jump to: navigation, search
Conference paper (help)
A new ANEW: evaluation of a word list for sentiment analysis in microblogs
Authors: Finn Årup Nielsen
Citation: Proceedings of the ESWC2011 Workshop on 'Making Sense of Microposts': Big things come in small packages 718 in CEUR Workshop Proceedings : 93-98. 2011 May
Editors: Matthew Rowe, Milan Stankovic, Aba-Sah Dadzie, Mariann Hardey
Publisher: Define publisher
Meeting: Making Sense of Microposts
Database(s): arXiv (arxiv/1103.2903) Citeulike Google Scholar cites Microsoft Academic Search
DOI: Define doi.
Link(s): http://ceur-ws.org/Vol-718/paper_16.pdf
Web: DuckDuckGo Bing Google Yahoo!Google PDF
Article: Google Scholar PubMed
Restricted: DTU Digital Library
Format: BibTeX

A new ANEW: evaluation of a word list for sentiment analysis in microblogs describes an evaluation for a word list ("AFINN") of sentiment analysis of Twitter messages.

The word list is available from:




It is also part of the "afinn" Python module:


Amazon Mechanical Turk-labeled Twitter data provided by Alan Mislove was used.

The software used for the paper programmed in Python is available from:


Slides from the Making Sense of Microposts:



[edit] Abstract

Sentiment analysis of microblogs such as Twitter has recently gained a fair amount of attention. One of the simplest sentiment analysis approaches compares the words of a posting against a labeled word list, where each word has been scored for valence, -- a "sentiment lexicon" or "affective word lists". There exist several affective word lists, e.g., ANEW (Affective Norms for English Words) developed before the advent of microblogging and sentiment analysis. I wanted to examine how well ANEW and other word lists performs for the detection of sentiment strength in microblog posts in comparison with a new word list specifically constructed for microblogs. I used manually labeled postings from Twitter scored for sentiment. Using a simple word matching I show that the new word list may perform better than ANEW, though not as good as the more elaborate approach found in SentiStrength.

[edit] Related papers

  1. Analyzing customer sentiments in microblogs-a topic-model-based approach for Twitter datasets
  2. Building lexicon for sentiment analysis from massive collection of HTML documents
  3. Classifying sentiment in microblogs: is brevity an advantage?
  4. Feature sentiment diversification of user generated reviews: the FREuD approach
  5. Good friends, bad news - affect and virality in Twitter
  6. Sentence level sentiment analysis in the presence of conjuncts using linguistic analysis
  7. Sentiful: generating a reliable lexicon for sentiment analysis
  8. Sentiment analysis of Twitter data
  9. Sentiment in short strength detection informal text
  10. Sentiment-based text segmentation
  11. SentiSense: an easily scalable concept-based affective lexicon for sentiment analysis

[edit] Use of word list

For further papers see AFINN. That list is more complete

  1. Aesthetic considerations for automated platformer design (2012)
  2. Crowd sentiment detection during disasters and crises
  3. Good friends, bad news - affect and virality in Twitter (2011)
  4. Networks and language in the 2010 election
  5. Retweets--but not just retweets: quantifying and predicting influence on Twitter (2012)
  6. Semi-automated argumentative analysis of online product reviews (2012)
  7. Summarization of yes/no questions using a feature function model (2011)
  8. The QWERTY effect: how typing shapes the meanings of words (2012)

Blog posts:

  1. Tracking US Sentiments Over Time In Wikileaks also posted: Tracking US Sentiments Over Time In Wikileaks
  2. Painting a Novel

[edit] Other mentioning

  1. Increasing the willingness to collaborate online: an analysis of sentiment-driven interactions in peer content production

[edit] Critique

  1. Performance are only given as correlation, - not more standard performance metrics: accuracy, F1, precision, recall.
  2. There are other and larger word lists than those tested. And these might be better.
  3. It is not compared with state-of-art which would probably entail some machine learning, better features, e.g., emoticons, negativity detection, ...
  4. There is no in-depth examination of why the sentiment analyzer fails on specific posts.
Personal tools