Methods and systems for detecting social engineering attacks comprise: extracting one or more non-semantic data items from an incoming email; determining whether the one or more non-semantic data items match information stored in a data store of previously collected information; performing behavioral analysis on the one or more non-semantic data items; analyzing semantic data associated with the email to determine whether the non-semantic data matches one or more patterns associated with malicious emails; and based on the determining, performing, and analyzing, identifying the email as potentially malicious or non-malicious. The system also includes processes for collecting relevant information for storage within the data store and processes for harvesting information from detected social engineering attacks for entry into the data store and seeding of the collection processes.