Humor in the Blogosphere: First Clues for a Verbal Humor Taxonomy

摘要

The importance of analyzing processes related to cognitive phenomena through Natural Language Processing techniques is acquiring a greater relevance every day. Areas such as Opinion Mining, Sentiment Analysis or Automatic Humor Recognition are samples about how this kind of research gradually grows. In this paper, we focused on studying the features that define a corpus of humorous data (one-liners) to assess whether they may be used as elements for building a verbal Humor taxonomy. We analyzed, through several experiments, a set of well-known features described in the literature, besides a set of new ones, to determine the importance of each. An evaluation of all features was performed by means of an automatic classification task over a collection of humorous blogs. Our results suggest that some features may represent basic information for creating the taxonomy.