Examining Personalization in Academic Web Search

Personalization promises to improve the accuracy of Web search and has been drawing much research attention recently. Some evidence indicates that for educational purposes, the disadvantages of personalized search are not justified by its benefits. The potential issues with search personalization, especially in an educational context, include loss of serendipity and capability, commercialization of education and the "Filter Bubble" effect where users are denied information if search engine algorithms decide it is irrelevant to them. The majority of students in higher education make use of general-purpose search engines to find academic information, however we have little knowledge about the effects of personalization on learners' experience and achievements. This observation motivates the research in this paper. First, we surveyed 120 university students to investigate which research sources, including search engines they predominately use and how much they depend on each for educational purposes. We learned that the majority of students prefer Google to other search engines; indeed sometimes it is their primary or only information-seeking tool. Additionally, about 80% of them use search engines for educational purposes on daily basis. Second, we measured the difference between personalized and non-personalized search results for 120 academic search queries divided equally into four categories: Education, IT, Health sciences and Business. Our results showed that on average only 53% of links appear, not necessarily in the same order, in both personalized and non-personalized search results. Interestingly, we observed only slight differences in the extent of personalization based on academic topics.