Uniform resource locator classifier and visual comparison platform for malicious site detection preliminary

1. A computing platform, comprising:

at least one processor;

a communication interface communicatively coupled to the at least one processor; and

memory storing computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:

identify, by parsing a uniform resource locator (URL), one or more human-engineered features of a URL, wherein the one or more human-engineered features comprise one or more of: a protocol, a top level domain (TLD), a domain, a subdomain, a port, a port type, a path, or path components;

identify one or more deep learned features of the URL;

concatenate the one or more human-engineered features of the URL to the one or more deep learned features of the URL, resulting in a concatenated vector representation;

compute, by inputting the concatenated vector representation of the URL to a URL classifier, a first phish classification score;

in response to determining that the first phish classification score is between a first phish classification threshold and a second phish classification threshold, cause image data for the URL to be sent to a visual similarity classification platform, configured to produce a computer vision vector representation of the image data and a corresponding second phish classification score; and

in response to determining that the second phish classification score exceeds the first phish classification threshold, cause a cybersecurity server to perform a first action.

2. The computing platform of claim 1 , wherein identifying the one or more human-engineered features of the URL comprises identifying a popularity of the URL, popularity of a domain of the URL, popularity of a subdomain and a domain of the URL, or popularity of a subdomain, a domain and one or more path components of the URL.

3. The computing platform of claim 2 , wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:

generate a popularity score for each of a plurality of URLs, domains, subdomains, or path components, wherein the popularity scores correspond to one or more of:

a number of times during a predetermined period that corresponding URLs, domains, subdomains, or path components were accessed, or

a number of users that accessed the corresponding URLs, domains, subdomains, or path components; and

store, in a popularity index, the popularity scores, wherein the popularity scores are stored along with a correlation to their respective URLs, domains, subdomains, and path components and wherein the computing platform is configured to access the popularity scores by applying a lookup function.

4. The computing platform of claim 1 , wherein identifying the one or more human-engineered features of the URL comprises identifying that the URL contains an instance of brand mimicry.

5. The computing platform of claim 1 , wherein identifying the one or more human-engineered features of the URL comprises identifying that the URL contains a homoglyph, identifying a character length of the URL, identifying an encrypted protocol, identifying a page extension of the URL, identifying a signature associated with a domain generation algorithm, identifying a presence of a random substring, identifying a presence of one or more strings from a lexicon in the URL, or identifying a number of subdomains of the URL.

6. The computing platform of claim 1 , wherein computing the vector representation of the URL comprises computing the vector representation of the URL using a neural network.

7. The computing platform of claim 6 , wherein the neural network is pre-trained using one or more of: a language modeling task or another self-supervised task.

8. The computing platform of claim 1 , wherein identifying the one or more human-engineered features of the URL comprises:

identifying that the URL is a redirector;

initiating one or more requests to capture a redirection chain of one or more URLs associated with redirection actions taken by each URL in the redirection chain; and

identifying one or more features of URLs associated with the redirection chain of the one or more URLs associated with the redirection chain.

9. The computing platform of claim 1 , wherein the URL classifier is a neural network.

10. The computing platform of claim 1 , wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:

in response to determining that the second phish classification score exceeds the second phish classification threshold, cause the cybersecurity server to perform a second action different from the first action.

11. The computing platform of claim 10 , wherein:

causing the cybersecurity server to perform the first action comprises setting a first flag; and

causing the cybersecurity server to perform the second action comprises setting a second flag, wherein:

the first flag and the second flag are set in a cybersecurity database hosted by one of: the computing platform or a central repository, and

the cybersecurity database is accessible by the cybersecurity server.

12. The computing platform of claim 11 , wherein the cybersecurity server is configured to monitor the cybersecurity database at a predetermined interval to detect flags.

13. The computing platform of claim 12 , wherein performing the second action comprises, in response to detecting the second flag, one or more of:

generating one or more commands directing another computing device to display a graphical user interface indicating that the URL is malicious, or

adding the URL to a list of blocked URLs.

14. The computing platform of claim 12 , wherein performing the first action comprises, in response to detecting the first flag, generating one or more commands directing a holistic classification platform to analyze content of the URL; and

wherein performing the second action comprises, in response to detecting the second flag, one or more of:

generating one or more commands directing another computing device to display a graphical user interface indicating that the URL is malicious, or

adding the URL to a list of blocked URLs.

15. The computing platform of claim 12 , wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:

based on a failure to detect a flag, determine that the second phish classification score does not exceed the first phish classification threshold or the second phish classification threshold; and

determine, based on the determination that the second phish classification score does not exceed the first phish classification threshold or the second phish classification threshold, that the URL is legitimate.

16. The computing platform of claim 1 , wherein the visual similarity classification platform is further configured to produce the second phish classification score by comparing the image data to image data of one or more ancestor pages of the URL, wherein the one or more ancestor pages of the URL include one or more intermediate pages between a root page of the URL and a parent page of the URL.

17. The computing platform of claim 1 , wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:

receive the URL; and

tokenize the URL to reduce the URL into a plurality of components.

18. A method, comprising:

at a computing platform comprising at least one processor, a communication interface, and memory:

identifying, by parsing a uniform resource locator (URL), one or more human-engineered features of a URL, wherein the one or more human-engineered features comprise one or more of: a protocol, a top level domain (TLD), a domain, a subdomain, a port, a port type, a path, or path components;

identifying one or more deep learned features of the URL;

concatenating the one or more human-engineered features of the URL to the one or more deep learned features of the URL, resulting in a concatenated vector representation;

computing, by inputting the concatenated vector representation of the URL to a URL classifier, a first phish classification score;

in response to determining that the first phish classification score is between a first phish classification threshold and a second phish classification threshold, causing image data for the URL to be sent to a visual similarity classification platform, configured to produce a computer vision vector representation of the image data and a corresponding second phish classification score; and

in response to determining that the second phish classification score exceeds the first phish classification threshold, causing a cybersecurity server to perform a first action.

19. The method of claim 18 , wherein identifying the one or more human-engineered features of the URL comprises identifying a popularity of the URL, popularity of a domain of the URL, popularity of a subdomain and a domain of the URL, or popularity of a subdomain, a domain and one or more path components of the URL.

20. One or more non-transitory computer-readable media storing instructions that, when executed by a computing platform comprising at least one processor, a communication interface, and memory, cause the computing platform to:

identify, by parsing a uniform resource locator (URL), one or more human-engineered features of a URL, wherein the one or more human-engineered features comprise one or more of: a protocol, a top level domain (TLD), a domain, a subdomain, a port, a port type, a path, or path components;

identify one or more deep learned features of the URL;

concatenate the one or more human-engineered features of the URL to the one or more deep learned features of the URL, resulting in a concatenated vector representation;

compute, by inputting the concatenated vector representation of the URL to a URL classifier, a first phish classification score;

in response to determining that the first phish classification score is between a first phish classification threshold and a second phish classification threshold, cause image data for the URL to be sent to a visual similarity classification platform, configured to produce a computer vision vector representation of the image data and a corresponding second phish classification score; and

in response to determining that the second phish classification score exceeds the first phish classification threshold, cause a cybersecurity server to perform a first action.