Mining Experiences from Large-scale Blog Entries

Abstract

An important characteristic of Weblogs(blogs) is that they contain many descriptions of people's expe- riences in the real world. This paper proposes a method for extracting people's experiences from large-scale blog entries and also a method for mining association rules between location, time, activity, and emotion. An activity consists of action and its object. We also categorize people's emotions into nine types, and classify each experience into success and failure based on the emotion categories. We constructed a system that mines association rules from about 48 million blog entries, and analyzes data from many directions using several interesting measures for data