Effective online advertising

Advertising in the Internet is a wide, attractive and growing market. In this market, revenue of websites that host ads depends on the number of user clicks received on displayed ads. Thus, in order to increase the revenue, websites try to select top ads and rank them based on their quality. Ad quality depends on different factors, such as relevance of ads to the users who are surfing a web page, relevance of ads to web page contents, and previous performance of ads. In this thesis, we address research problems related to improving online advertising. More specifically, we investigate the problems of choosing the most relevant ads to users. We divide this problem into two related sub-problems. The first problem is predicting the quality of new ads in search engines. The second problem is matching ads with online videos. We propose novel approaches related to each problem, and show that they outperform them against other existing approaches in the literature.