We cite as we communicate

Building on ideas from linguistics, psychology, and social sciences about the possible mechanisms of human decision-making, we propose a novel theoretical framework for the citation analysis. Given the existing trend to investigate citation statistics in the context of various forms of power and Zipfian laws, we show that the popular models of citation have poor predictive ability and can hardly provide for an adequate explanation of the observed behavior of the empirical data. An alternative model is then derived, using the apparatus of statistical mechanics. The model is applied to approximate the citation frequencies of scientific articles from two large collections, and it demonstrates a predictive potential much superior to the one of any of the citation models known to the authors from the literature. Some analytical properties of the developed model are discussed, and conclusions are drawn. Directions for future work are also given at the paper’s end. PACS codes: 89.65.-s, 89.70.+c, 89.75.-k