Semantic Hubs and Authorities for Citation Analysis

摘要

We describe a method for analyzing the impact of papers on their field. We classify the paper's citations by semantic meaning with numeric encoding and use them as weights on our extension of the Hubs and Authorities algorithm to determine the most authoritative papers. We compare our authorities to those determined by straight citation counts. While current common analysis methods only take into account the local citation counts, in part due to the difficulty of classifying large numbers of citations, we rely on graph analysis. We not only use the structure of the links among papers but also the semantics of the citations in order to get a step closer to the actual meanings underlying the citations.