作者: Andrew T. Fiore , Scott Lee Tiernan , Marc A. Smith
关键词:
摘要: In this paper we describe an evaluation of behavioral descriptors generated from analysis a large collection Usenet newsgroup messages. The metrics aspects authors' behavior over time; such information can aid in filtering, sorting, and recommending content public discussion spaces like newsgroups. To assess the value variety these descriptors, compared 22 participants' subjective evaluations authors whose messages they read to describing same authors. We found that many metrics, particularly longevity frequency participation, number newsgroups which contribute messages, amount each thread, correlate highly with readers'