Semantic relatedness based on searching engines

Semantic relatedness plays an important role in Natural Language Processing area since it has the ability to imitate human judgments. Recently the most popular methods are depend on the hierarchy of WordNet, which result in restricted calculation objects and language, moreover, the accuracy is not satisfied. This paper presents a new semantic relatedness calculation method by considering number of returned web pages which resulted from searching combination of measured words in a searching engine. It is assumption that words appear in a same web page have some relatedness. The searching Engine semantic relatedness method doesn´t have the limitation on part of speech and language. Experiment results show the method is better than the current important methods, and the accuracy can be over eighty percent.