Identifying phishing communications using templates

1. A computer-implemented method, comprising:

comparing, by one or more processors, content of a communication to indexed content of a plurality of templates, each template associated with a trustworthy entity and representing indexed content shared among a cluster of communications sent by the trustworthy entity;

identifying, by the one or more processors based on the comparing, one or more matched templates;

determining, by the one or more processors, that a reply-to address associated with the communication fails to match one or more patterns of reply-to addresses used by one or more trustworthy entities associated with the one or more matched templates;

classifying, by the one or more processors based on the determining, the communication as a phishing attempt; and

discarding or re-routing, by the one or more processors, the communication in response to the classifying.

2. The computer-implemented method of claim 1 , wherein comparing the content of the communication to the indexed content of the plurality of one or more templates comprises determining respective measures of similarity of the plurality of templates to the communication.

3. The computer-implemented method of claim 2 , further comprising:

ranking the plurality of templates based on their respective measures of similarity; and

selecting, as the one or more matched templates, a predetermined number of highest ranking templates.

4. The computer-implemented method of claim 2 , further comprising selecting, as the one or more matched templates, one or more templates with respective measures of similarity that satisfy a threshold.

5. The computer-implemented method of claim 1 , wherein the comparing comprises comparing one or more n-grams in the communication to one or more n-grams used to index the plurality of templates.

6. The computer-implemented method of claim 5 , wherein the one or more n-grams used to index the plurality of templates are extracted from content of the plurality of templates.

7. The computer-implemented method of claim 1 , wherein the comparing comprises comparing one or more overlapping n-grams in the communication to one or more overlapping n-grams used to index the plurality of templates.

8. The computer-implemented method of claim 1 , wherein the determining further comprises, for each of the one or more trustworthy entities associated with the one or more matched templates, comparing:

a combination of the reply-to address associated with the communication and a subject of the communication; to

a combination of a pattern of reply-to addresses used by the trustworthy entity and a pattern found among subjects of communications sent by the trustworthy entity.

9. A system including memory and one or more processors that when operated execute instructions stored in the memory, comprising the instructions to:

compare content of a communication to indexed content of a plurality of templates, each template associated with a trustworthy entity and representing indexed content shared among a cluster of communications sent by the trustworthy entity;

identify, based on the comparing, one or more matched templates;

determine that a reply-to address associated with the communication fails to match one or more patterns of reply-to addresses used by one or more trustworthy entities associated with the one or more matched templates;

classify, based on the determining, the communication as a phishing attempt; and

discard or re-route the communication in response to the classifying.

10. The system of claim 9 , wherein the system further comprises instructions to:

rank the plurality of templates based on respective measures of similarity between the plurality of templates and the communication; and

select, as the one or more matched templates, a predetermined number of highest ranking templates.

11. The system of claim 9 , wherein the system further comprises instructions to select, as the one or more matched templates, one or more templates with a threshold amount of respective indexed content that matches the content of the communication.

12. The system of claim 9 , wherein the system further comprises instructions to compare one or more n-grams in the communication to one or more n-grams used to index the plurality of templates.

13. The system of claim 9 , wherein the system further comprises instructions to compare one or more overlapping n-grams in the communication to one or more overlapping n-grams used to index the plurality of templates.

14. The system of claim 9 , wherein the system further comprises instructions to, for each of the one or more trustworthy entities associated with the one or more matched templates, compare:

a combination of the reply-to address associated with the communication and a subject of the communication; to

a combination of a pattern of reply-to addresses used by the trustworthy entity and a pattern found among subjects of communications sent by the trustworthy entity.

15. At least one non-transitory computer-readable medium comprising instructions that, in response to execution of the instructions by a computing system, cause the computing system to perform the following operations:

comparing content of a communication to indexed content of a plurality of templates, each template associated with a trustworthy entity and representing indexed content shared among a cluster of communications sent by the trustworthy entity;

identifying, based on the comparing, one or more matched templates;

determining that a reply-to address associated with the communication fails to match one or more patterns of reply-to addresses used by one or more trustworthy entities associated with the one or more matched templates;

classifying, based on the determining, the communication as a phishing attempt; and

discarding or re-routing the communication in response to the classifying.