Common Criteria for Genre Classification: Annotation and Granularity

In this paper,we present two experiments that use machine learning for automatically classifying web pages by genre. These experiments highlight the influence that genre annotation and genre granularity can have on the accuracy of the classification. From a practical point of view these experiments show that a collection annotated with the criteria of ‘objective sources’ and consistent genre granularity ensures a very good classification accuracy (Experiment 1). Additionally, the classification model built out of such a collection can be exported more profitably for predictive tasks on an unclassified web page collection (Experiment 2). These experiments represent a starting point for a discussion about the need of common criteria for building a genre collection in the absence of an official genre-annotated benchmark.