As a whole, the World Wide Web displays a striking “rich getricher” behavior, with a relatively small number of sites receivinga disproportionately large share of hyperlink references andtraffic. However, hidden in this skewed global distribution, wediscover a qualitatively different and considerably less biased linkdistribution among subcategories of pages—for example, among alluniversity homepages or all newspaper homepages. Although theconnectivity distribution over the entire web is close to a pure powerlaw, we find that the distribution within specific categories istypically unimodal on a log scale, with the location of the mode, andthus the extent of the rich get richer phenomenon, varying acrossdifferent categories. Similar distributions occur in many othernaturally occurring networks, including research paper citations, movieactor collaborations, and United States power grid connections. Asimple generative model, incorporating a mixture of preferential anduniform attachment, quantifies the degree to which the rich nodes growricher, and how new (and poorly connected) nodes can compete. The modelaccurately accounts for the true connectivity distributions ofcategory-specific web pages, the web as a whole, and other socialnetworks.
展开▼