Learning to Filter Spam E-Mail: A Comparison of a Naive Bayesian and a Memory-Based Approach

By Ion Androutsopoulos, G. Paliouras, Vangelis Karkaletsis, Georgios Sakkis, Constantine D. Spyropoulos, Panagiotis Stamatopoulos. arXiv.org, 2000

410 citations recorded by the source corpus.

We investigate the performance of two machine learning algorithms in the context of antispam filtering. The increasing volume of unsolicited bulk e-mail (spam) has generated a need for reliable anti-spam filters. Filters of this type have so far been based mostly on keyword patterns that are constructed by hand and perform poorly. The Naive Bayesian classifier has recently been suggested as an effective method to construct automatically anti-spam filters with superior performance. We investigate thoroughly the performance of the Naive Bayesian filter on a publicly available corpus, contributing towards standard benchmarks. At the same time, we compare the performance of the Naive Bayesian filter to an alternative memorybased learning approach, after introducing suitable cost-sensitive evaluation measures. Both methods achieve very accurate spam filtering, outperforming clearly the keyword-based filter of a widely used e-mail reader.

Subfields: Artificial Intelligence, Information Systems

View Paper: https://www.semanticscholar.org/paper/d0aa1ae9d8df097524fda2b7ff0d4ee81ee8be73

How researchers see this paper

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

In Artificial Intelligence

1. Bo Li
2. Anton Bryl
3. Enrico Blanzieri
4. Yevgeniy Vorobeychik
5. Alina Oprea
6. Chang Liu
7. Wei You
8. Gang Yang

In Information Systems

1. Anton Bryl
2. Enrico Blanzieri
3. Bo Li
4. Wei You
5. Gang Yang
6. Sheng-yi Jiang
7. Bin Liang
8. Fernando Díaz

Citations in the corpus

• A survey of learning-based techniques of email spam filtering (2008, methodology): “Used in applied to to RIPPER B,H,W B [26] Stacking B,H,W B [79, 105] Na¨ıve Bayes B,H,W B,H,W [5, 4, 3, 6, 14, 41, 53, 62, 71, 77, 102, 105] Flexible Bayes B,H,W B [6] Boosting B,H,W B,H,W [6, 13, 26, 102, 105] Maximum Entropy Model B,H,W B,H,W [101, 102] Support Vector Machines B,H,W B,H,W [6, 9, 1”
• Cracking Classifiers for Evasion: A Case Study on the Google's Phishing Pages Filter (2016, background): “In fact, there are many classifiers equipped with different classification algorithms, e.g., [8][38].”
• An evaluation of statistical spam filtering techniques (2004, methodology): “WAcc was introduced by Androutsopoulos et al. [2000] and has been used in several spam .ltering benchmarks [Androutsopoulos et al. 2000; Carreras and M´ arquez 2001].”
• Applying lazy learning algorithms to tackle concept drift in spam filtering (2007, background): “As part of the study of anti-spam filtering techniques, there has also been intensive research into the use of memory-based classifiers (Androutsopoulos et al., 2000; And-routsopoulos et al., 2004; Sakkis et al., 2003).”
• Natural language processing: state of the art, current trends and challenges (2017, methodology): “Most text categorization approaches to anti-spam Email filtering have used multi variate Bernoulli model (Androutsopoulos et al., 2000) [5] [15]. d) Information Extraction Information extraction is concerned with identifying phrases of interest of textual data.”
• Robust Linear Regression Against Training Data Poisoning (2017, background): “An important class of machine learning applications enable scalable security defenses, such as spam filtering, traffic analysis, and fraud detection [2, 8, 27].”
• Using naive bayes to detect spammy names in social networks (2013, background): “This classifier has a long and successful history of dealing with the problem of email spam [25, 2, 16, 26, 27, 1].”
• An improved K-nearest-neighbor algorithm for text categorization (2012, methodology): “…filtering, weighted accuracy (WAcc), total cost ratio (TCR), spam recall (SR), spam precision (SP), SF 1 (the combination of SR and SP) are widely used to evaluate the spam filter performance (Androutsopoulos et al.,2000a; Androutsopoulos et al.,2000b; Sakkis et al.,2001; Koprinska et al.,2007).”
• Robust High-Dimensional Linear Regression (2016, background): “An important class of applications of machine learning uses it to enable scalable security solutions, such as spam filtering, traffic analysis, and fraud detection [1, 2, 3].”
• Spam Corpus Creation for TREC (2005, background): “Artificial collections (spamassassin.org, 2003; Androutsopoulos et al., 2000; Michelakis et al., 2004) may be created by using mailing list messages as opposed to personal email, by selecting non-sensitive messages from a real email collection, by mixing messages from diverse sources, or by obfuscat”
• Deep-learning inversion: a next generation seismic velocity-model building method (2019, background): “, 2013), spam filters (Androutsopoulos et al., 2000), fraud alerts (Ravisankar et al.”