Machine-learning for password guess risk determination

A method comprising:

• accessing information about a computing resource;
• accessing information about a user attempting to access the computing resource;
• using the information about the computing resource to retrieve a computing domain-specific text corpus for a domain encompassing the computing resource;
• using the information about the user to retrieve a user information-specific text corpus for a piece of user information in the accessed information about the user;
• passing the domain-specific text corpus to a first natural language processing (NLP) machine-learned model trained specifically for the domain, the first NLP machine-learned model outputting domain-specific word embeddings;
• passing the user information-specific text corpus to a second NLP machine-learned model trained specifically for the piece of user information, the second NLP machine-learned model outputting user information-specific word embeddings;
• and
• inputting the domain-specific word embeddings and the user information-specific word embeddings into a probabilistic context free grammar (PCFG) machine-learned model, the PCFG machine-learned model updating itself based on the input and outputting a set of password guesses.

A non-transitory machine-readable medium storing instructions which, when executed by one or more processors, cause the one or more processors to perform operations comprising:

• accessing information about a computing resource;
• accessing information about a user attempting to access the computing resource;
• using the information about the computing resource to retrieve a computing domain-specific text corpus for a domain encompassing the computing resource;
• using the information about the user to retrieve a user information-specific text corpus for a piece of user information in the accessed information about the user;
• passing the domain-specific text corpus to a first natural language processing (NLP) machine-learned model trained specifically for the domain, the first NLP machine-learned model outputting domain-specific word embeddings;
• passing the user information-specific text corpus to a second NLP machine-learned model trained specifically for the piece of user information, the second NLP machine-learned model outputting user information-specific word embeddings;
• and
• inputting the domain-specific word embeddings and the user information-specific word embeddings into a probabilistic context free grammar (PCFG) machine-learned model, the PCFG machine-learned model updating itself based on the input and outputting a set of password guesses.