CTR Prediction: A Bibliometric Review of Scientific Literature

Abstract

Internet giants like Google, Facebook, and Amazon have relied heavily on revenue from online advertising sales in recent years. The Click-through rate of an advertisement is the percentage of people who clicked it out of those who viewed it. The CTR as a metric represents the performance of their online advertisements. For over two decades now, researchers in academia and industry have paid great attention to getting good CTR prediction accuracy because of the demand it has generated in the digital world. This paper reviews the scholarly literature on CTR prediction during the previous decade by a bibliometric analysis of 1051 publications from journals indexed by Scopus. The goals of this research are to (1) conduct a structured quantitative analysis of the bibliometric data, (2) chart the development of CTR Prediction research, and (3) identify the most recent and relevant research literature and viewpoints in the field. A handful of studies have been conducted on this subject, and they have presented an in-depth analysis of specific methods and learning models being applied for CTR prediction. In addition to the previously submitted studies, this literature review aims to provide an overall bibliometric analysis showing the evolution of techniques employed for CTR prediction in the articles published over the last ten years, using a combination of bibliographic coupling, citation, and co-citation. The outcome of this literature evaluation will aid future researchers in gaining a deeper comprehension of the scientific studies around CTR prediction.