Ranking of new sponsored online ads

Online advertising in search engines is a wide, attractive and growing market. In this market, revenue of search engines depends on the number of user clicks received on displayed ads. Thus, in order to increase the revenue, search engines try to select top ads and rank them based on their quality. For ads which were in the system for a period of time, this quality could be measured empirically. But for new ads, or those ads without enough historical data, search engines should predict the quality. The purpose of this work is to find a method for estimating quality of new ads in attracting user´s clicks. We use semantic and feature based similarity algorithms to predict the quality score of new ads using historical similar ads. Our trace-based evaluations show that the proposed method outperforms other approaches in the literature in terms of the accuracy of prediction. In addition, the proposed is less computationally expensive than previous methods and it can run in real time.