Abstract

As the use of the Internet increased in recent years, UCC has become an important resource on the Web. At the same time, spam UCC has also become problematic. One type of spam UCC is word salad, which is created by injecting certain words or paragraphs into normal documents. This paper proposes a word-salad filtering algorithm for detecting word-salad documents using normal distribution of term’s features and utilizing ODP , the largest Web term directory in the world