An Architecture of an Academic Search Engine with Personalized Search Result Ranking Mechanism

A rapid increasing of information on the Internet and World Wide Web causes information overloaded problem. Thus, search engines become important tools to help WWW users to discover the information they need. With an exponentially increasing of published research paper, community-based research paper sharing systems and academic search engines become indispensable tools for researchers to search for any research papers in their fields of expertise and related fields according to their interests. To improve a quality of research paper searching, an academic search engines' capability should be enhanced. This paper proposed an architecture of an academic search engine with personalized search result ranking mechanism. To evaluate the performance of personalized search result ranking mechanism, twenty-five graduate students were invited to be participants in this research study. As a criterion, the participants were asked to use a prototype of academic search engine to find and bookmark any research papers according to their interests. This would guarantee that each participants' list of interesting research paper could be recorded. The Normalized Discounted Cumulative Gain (NDCG) was used as a metric to determine that performance of personalized search result ranking mechanism. During the experiment, each participant was asked to search for research papers according to their interests. The result of the experiment suggested that the personalized search result ranking mechanism outperformed the original search result ranking. Hence, the proposed architecture of the academic search engine with personalized search engine mechanism does benefit a tasks of research paper discovery. It improves the quality of research paper searching.