Arama Motorlarında Yeni Konu Tanılamada Karakter N-Gram ve Yapay Sinir Ağları Uygulaması

Abstract

Nowadays, the estimate of web users' behaviors has been important due to the web search engine usage increase. To date, many content-ignorant studies have been performed for automatic new topic identifica- tion. Although, some studies performed well, it was observed that they often made mistakes when queries had spelling differences. In this study the character n-gram methodology, which is content ignorant, was used for new topic identification. In addition, it was aimed to improve previous content-ignorant studies. Consideration of previous studies it was observed that the neural network applications gave better results than the other studies. Thus, the neural network method's estimations were used in this study and character n-gram methodology was used in order to eliminate wrong estimations, because of spelling errors.