Spam Corpus Creation for TREC

By Gordon V. Cormack, T. Lynam. 2005

110 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

The security of machine learning (2010, methodology): “We use the Text Retrieval Conference (TREC) 2005 spam corpus (Cormack and Lynam 2005), which is based on the Enron email corpus (Klimt and Yang 2004) and contains 92,189 emails (52,790 spam and 39,399 ham).”
Hybrid Deterministic-Stochastic Methods for Data Fitting (2011, methodology): “Our experiments for binary logistic regression are based on the TREC 2005 data set, which contains 823,470 binary variables describing the presence of word tokens in 92,189 email messages involved in the legal investigations of the Enron corporation [9].”
Minimizing finite sums with the stochastic average gradient (2013, unknown): “rotein 145 751 74 [Caruana et al., 2004] covertype 581 012 54 [Frank and Asuncion, 2010] rcv1 20 242 47 236 [Lewis et al., 2004] news 19 996 1 355 191 [Keerthi and DeCoste, 2005] spam 92 189 823 470 [Cormack and Lynam, 2005, Carbonetto, 2009] rcv1Full 697 641 47 236 [Lewis et al., 2004] sido 12 678 ”
TREC 2007 Spam Track Overview (2007, background): “[6] This corpus is the first TREC public corpus that contains exclusively ham and spam sent to the same server within the same time period.”
Evaluating sentence-level relevance feedback for high-recall information retrieval (2018, methodology): “The Track coordinators used an iterative process (Cormack and Lynam, 2005a) in which a number of spam classiﬁers were applied to the corpus, and disagreements between the classiﬁers and a provisional labeling were adjudicated by the coordinators.”
Exploiting Machine Learning to Subvert Your Spam Filter (2008, methodology): “In our experiments we use the Text Retrieval Conference (TREC) 2005 spam corpus [4], which is based on the Enron email corpus [11] and contains 92,189 emails (52,790 spam and 39,399 ham).”
StopWasting My Gradients: Practical SVRG (2015, unknown): “Data set Data Points Variables Reference quantum 50 000 78 [24] protein 145 751 74 [24] sido 12 678 4 932 [25] rcv1 20 242 47 236 [26] covertype 581 012 54 [27] news 19 996 1 355 191 [28] spam 92 189 823 470 [29, 30] rcv1Full 697 641 47 236 [26] alpha 500 000 500 Synthetic Table 1: Binary data sets ”
A survey of learning-based techniques of email spam filtering (2008, background): “The TREC Spam Track relies on a similar definition: spam is " unsolicited, unwanted email that was sent indiscriminately, directly or indirectly , by a sender having no current relationship with the user " [19].”
On the hardness of evading combinations of linear classifiers (2013, methodology): “We trained Logistic Regression, Linear Support Vector Machine (SVM) and Näıve Bayes classifiers on the 2005 TREC public spam corpus [4], which is comprised of 92,189 labeled email messages.”
Machine Learning in the Presence of an Adversary: Attacking and Defending the SpamBayes Spam Filter (2008, methodology): “For my experiments, I use the Text Retrieval Conference (TREC) 2005 spam conference corpus [3], which is based on the Enron dataset [11].”
Behavior of Machine Learning Algorithms in Adversarial Environments (2010, methodology): “In these experiments, I use the Text Retrieval Conference (TREC) 2005 spam corpus as described by Cormack and Lynam [2005], which is based on the Enron email corpus [Klimt and Yang, 2004] and contains 92, 189 emails (52, 790 spam and 39, 399 ham).”
Misleading Learners: Co-opting Your Spam Filter (2009, methodology): “In our experiments we use the Text Retrieval Conference (TREC) 2005 spam corpus [5], which is based on the Enron email corpus [12] and contains 92,189 emails (52,790 spam and 39,399 ham).”