作者: A. Cohen , R. N. Mantegna , S. Havlin
DOI: 10.1142/S0218348X97000103
关键词:
摘要: We perform a numerical study of the statistical properties natural texts written in English and two types artificial texts. As tools we use conventional Zipf analysis distribution words inverse frequencies words, vocabulary growth, Shannon entropy quantity which is nonlinear function frequency "entropy". Our results, obtained by investigation eight complete books sixteen related texts, suggest that, among these analyses, growth shows most striking difference between results also those who give greater weight to low succeed better distinguishing The seems than "entropy" usual word ...