Using search engines and artificial neural networks for style checking

Choosing the best words for expressing the thoughts is never an easy task, especially in a foreign language. However, the huge amount of text available on the Web can provide a clue. In fact, the frequency of distinct expressions or phrases that have the same meaning can help to separate the wheat from the chaff. This paper proposes an approach based on artificial neural networks - ANN to generalize this idea, taking the frequency of the words in an expression as the input to the ANN and setting the output to one for the most common phrase and zero for the least, using the training data.