HUMOR: A Crowd-Annotated Spanish Corpus for Humor Analysis
share
Computational Humor, as the name implies, studies humor from a computational perspective, and it fosters several tasks, such as humor recognition, humor generation and humor scoring. The area has been little explored, making it attractive to tackle by novel Natural Language Processing and Machine Learning techniques. However, human-curated data is necessary. In this work we present a corpus of almost 40,000 tweets written in Spanish and crowd-annotated by their humor and funniness value with respect to several people on the Internet. It is equally divided between tweets coming from humorous accounts and from non-humorous accounts. There is certain humor value agreement between the raters, with a Krippendorff's alpha value of 0.3654, that allows building a humor classifier upon it. However, it shows an absence of agreement in the funniness value. The dataset is available for general usage and has already been used successfully for humor recognition. Additionally, more aspects of the dataset are analyzed in this paper, such as the distribution by the number of annotations and by categories.
READ FULL TEXT