논문 검색 엔진을 위한 랭킹 방법

초록

Researchers use a search engine for finding what they want to read from an article database. Because the search engine normally returns a large number of articles as a search result, they could not find the articles that they want. A ranking method solves this problem by giving proper ranks to articles. In this paper, we define the articles that researchers want to read as high quality articles, and propose a ranking method, ArtRank, to give high ranks to these articles. For determining the rank of an article, ArtRank simultaneously considers the authority of the articles that cite it and the reputation of a journal or a conference where it is published. We point out the problem of the existing impact factor that is employed for determining the reputation of a journal or a conference, and propose its solution. Also, for solving the problem of underestimation of the recent articles, ArtRank analyzes the publication years of articles and applies the result of analysis to the ranking computation process. We show the effectiveness of ArtRank by comparing its search results with those of prior article ranking methods.

키워드