SentiWordNet is a lexical resource for opinion mining. SentiWordNet assigns to each synset of WordNet three sentiment scores: positivity, negativity, objectivity. SentiWordNet is described in details in the papers:

 SentiWordNet: A Publicly Available Lexical Resource for Opinion Mining
 SentiWordNet 3.0: An Enhanced Lexical Resource for Sentiment Analysis and Opinion Mining

---

How to obtain SentiWordNet

The current "official" version of SentiWordNet is 3.0, which is based on WordNet 3.0.

SentiWordNet is distributed under the Attribution-ShareAlike 3.0 Unported (CC BY-SA 3.0) license.
Among the other possibilities, this license allows the use of SentiWordNet in commercial applications, provided that the application mentions the use of SentiWordNet and SentiWordNet is attributed to its authors.

Click here to download SentiWordnet 3.0

---

micro-WordNet-Opinion 3.0

micro-WordNet-Opinion 3.0 is the automatic mapping of the micro-WordNet-Opinion corpus to WordNet 3.0.

---

SentiWordNet has been used in...

Check Google for a list of the papers that use SentiWordNet 3.0

Check Google for a list of the papers that use SentiWordNet 1.0

User feedback

The SentiWordNet interface now provides the possibily to accept user feedback on the values assigned to synsets.

This feature is a first step towards building a community of SentiWordNet users that collaboratively improve SentiWordNet.

Feedback log packets

Any feedback collected from the community will be shared to the public domain through this site, i.e., what comes from the community goes back to the community.

Feedback log packets will be released periodically on this site.

The feedback process is currently anonymous, the IP and time of the feedback action is recorded only for the purpose of spam filtering, the feedback log packets will not contain the IP address.

Improving SentiWordNet with feedback

The feedback currently does not directly affect SentiWordNet values, we (Baccianella, Esuli, Sebastiani) manually check it in order to correct SentiWordNet values.

Sample terms:
 good  bad  best  worst  short  long  ill  healthy

Sample code

Here you can find a simple java class that uses SentiWordNet to approximate the sentiment value of a word with a label.

P.S. Note that in this piece of code there is no word sense disambiguation and the scores are approximated with labels, so it's very basic.

News

28/05/2010
The new SentiWordNet site has been launched!
Check out the new Feedback button in the search page!