"Malicious software attacks on the Internet are on the rise in both frequency and sophistication. The use of Naïve Bayes technique has been shown to be capable to detect malware. This paper proposes the usage of domain knowledge in the algorithm to improve detection accuracy. The domain knowledge consists of all supporting information that can be used to guide the learning process. In this paper, SNORT signatures are used as the domain knowledge and thus, only descriptive features in the learning corpora are trained to generate the classifier model."