http://www.bibsonomy.org/burst/user/cschenk/algorithmsBibSonomy BuRST Feed for /user/cschenk/algorithms2008-08-21T05:17:26+02:00http://www.bibsonomy.org/bibtex/2c66fa858b58398f469558c5d85cf8a7a/cschenkcschenk2008-06-21T18:52:26+02:00information paper algorithms benchmark bayes ir winnow categorization classification automatic enron svm retrieval email read:2008 folders sri <span style="color:#555555;">R. <a href="http://www.bibsonomy.org/author/Bekkerman">Bekkerman</a> and A. <a href="http://www.bibsonomy.org/author/McCallum">McCallum</a> and G. <a href="http://www.bibsonomy.org/author/Huang">Huang</a> </span><em>Center for Intelligent Information Retrieval, Technical Report IR</em>(<em>2004</em>)Sat Jun 21 18:52:26 CEST 2008Center for Intelligent Information Retrieval, Technical Report IRAutomatic Categorization of Email into Folders: Benchmark Experiments on Enron and SRI Corpora4182004information paper algorithms benchmark bayes ir winnow categorization classification automatic enron svm retrieval email read:2008 folders sri Office workers everywhere are drowning in email—not only spam, but also large quantities of legitimate email to be read and organized for browsing. Although there have been extensive investigations of automatic document categorization, email gives rise to a number of unique challenges, and there has been relatively little study of classifying email into folders. This paper presents an extensive benchmark study of email foldering using two large corpora of real-world email messages and foldering schemes: one from former Enron employees, another from participants in an SRI research pro ject. We discuss the challenges that arise from differences between email foldering and traditional document classification. We show experimental results from an array of automated classification methods and evaluation methodologies, including a new evaluation method of foldering results based on the email timeline, and including enhancements to the exponential gradient method Winnow, providing top-tier accuracy with a fraction the training time of alternative methods. We also establish that classification accuracy in many cases is relatively low, confirming the challenges of email data, and pointing toward email foldering as an important area for further research.