Skip to Main Content
This paper describes an algorithm for semi-automatically creating an emotion dictionary using WordNet and evaluating the created dictionary. The algorithm takes as input a set of seed words that have been manually assigned with WordNet senses and emotion information. An initial dictionary is automatically created using the seed words and WordNet. Then, various correction stages are performed where parts of the dictionary are shown to the user for verification. An almost 6,000 word sense emotion dictionary was created from only 130 seed words using the proposed algorithm. Evaluation of the created dictionary shows that it has a good vocabulary coverage for an independently tagged emotion corpus. The created dictionary was also used to classify the opinion of news articles. A simple algorithm that relied completely on the strength of the created dictionary achieved an accuracy of 84%.