학술 데이터베이스에서 논문 랭킹 알고리즘의 성능 평가

초록

Due to the increasing usage of search engines for scientific literature data, various ranking methods for scientific literature data have been proposed. It is quite important to compare and to analyze the performances of those ranking methods. However, there is no such a research result that covers representative ranking methods. In this paper, we perform comparative performance study on six representative methods for ranking scientific literature data. We first introduce the six representative ranking methods that are based on the well-known framework of ‘random walk with restart’ and point out their characteristics. Then, we perform extensive experiments with a real-life literature data for comparing their performances. We analyze the results and also discuss the critical factors in each method that affect the accuracy of the ranking results.

키워드