Umigon-core

The classification engine for sentiment analysis. The basic operations are:

• decompose the text into n-grams
• create a version of the n-gram which is stripped from accents etc.
• for each of these two versions of the ngrams: if the ngram belongs to a pre-established lexicon, apply a heuristic for this term.
• the lexicons are listed in a repo umigon-heuristics also to be found on github.

Key features

• Multilingual. French and English are covered and new languages will be added.

• Enables state of the art sentiment analysis in French and English.

To do

The denominations (what is a heuristic? a conditional expression?) are not super clear and consistent. It should get straightened up.

License

Apache v2 license. These data and code can be used for non commercial and commercial purposes. Just make sure to include the Apache v2 license and the copyright notice.

Credit

For academic use, cite this reference:

Levallois, Clement. “Umigon: Sentiment analysis on Tweets based on terms lists and heuristics”. Proceedings of the 7th International Workshop on Semantic Evaluation (SemEval), 2013, Atlanta, Georgia

Contact

Clement Levallois, @seinecle on Twitter or https://clementlevallois.net