Measuring Web Search Personalization

Personalization in online services is the practice of tailoring data contents for customers according to what is supposed to be their expectation. By design, personalization has been considered an important tool to help users to find the most interesting relevant data. By doing that, personalization also filters the web contents and potentially narrows the view of users. In this context, our work aims to measure personalization levels of web search results. First, we study personalization for web search engines with two case studies about Google Search. The results show a remarkable level of Google personalized search results based on sets of keywords, and we show that a specific website appears as prevalent in the results of web searches. Second, we measure the personalization degrees of an online news aggregator to provide a wider view of the problems. It also shows that compelling evidence such as “suggested for you" heavily depends on past users’ activities. Finally, we study personalization of search results on an online shopping platformby measuring price steering phenomenon. Particularly, we investigate the impacts of online behaviours, locations and economic performance factors, and we observe that price steering is based on user’s behaviours but it is also influenced by the geographic location of users.