Transaction policy audit

A system comprising:

• one or more computers;
• and
• a computer-readable medium coupled to the one or more computers having instructions stored thereon which, when executed by the one or more computers, cause the one or more computers to perform operations comprising:
• identifying policy questions that each correspond to a conformance or a violation of a policy selected by at least one policy-enforcer entity that enforces the policy, wherein policy questions for a first policy-enforcer entity include a first set of policy questions specific to the first policy-enforcer entity and a second set of policy questions common to multiple policy-enforcer entities that include the first policy-enforcer entity and at least a second policy-enforcer entity different from the first policy-enforcer entity;
• training, for each respective policy question, a machine learning policy model for the respective policy question based on historical determinations of policy question answers for the respective policy question, and wherein each machine learning policy model is trained to determine whether a given request corresponds to a policy conformance or a policy violation;
• receiving data associated with a request associated with the first policy-enforcer entity;
• identifying, for each respective policy question associated with the first policy-enforcer entity, the trained machine learning policy model for the respective policy question;
• identifying, for each identified machine learning policy model, tuning parameters for the first policy-enforcer entity, wherein first tuning parameters for a first machine learning policy model for the first policy-enforcer entity are different than second tuning parameters for the first machine learning policy model for the second policy-enforcer entity;
• tuning each identified machine learning model for the first policy-enforcer entity using tuning parameters specific to the first policy-enforcer entity;
• using, for each respective policy question associated with the first policy-enforcer entity, the identified machine learning policy model to automatically determine a respective policy question answer to the respective policy question for the request that indicates whether the request conforms to or violates the respective policy question;
• and
• in response to determining that a first policy question answer corresponds to a policy violation:
• generating an audit alert regarding the policy violation that identifies the request as a fraudulent request;
• and
• providing the audit alert to one or more systems, wherein any of the one or more systems that receives the audit alert regarding the policy violation automatically rejects the request based on the audit alert.

A computer program product encoded on a non-transitory storage medium, the product comprising non-transitory, computer readable instructions for causing one or more processors to perform operations comprising:

• identifying policy questions that each correspond to a conformance or a violation of a policy selected by at least one policy-enforcer entity that enforces the policy, wherein policy questions for a first policy-enforcer entity include a first set of policy questions specific to the first policy-enforcer entity and a second set of policy questions common to multiple policy-enforcer entities that include the first policy-enforcer entity and at least a second policy-enforcer entity different from the first policy-enforcer entity;
• training, for each respective policy question, a machine learning policy model for the respective policy question based on historical determinations of policy question answers for the respective policy question, and wherein each machine learning policy model is trained to determine whether a given request corresponds to a policy conformance or a policy violation;
• receiving data associated with a request associated with the first policy-enforcer entity;
• identifying, for each respective policy question associated with the first policy-enforcer entity, the trained machine learning policy model for the respective policy question;
• identifying, for each identified machine learning policy model, tuning parameters for the first policy-enforcer entity, wherein first tuning parameters for a first machine learning policy model for the first policy-enforcer entity are different than second tuning parameters for the first machine learning policy model for the second policy-enforcer entity;
• tuning each identified machine learning model for the first policy-enforcer entity using tuning parameters specific to the first policy-enforcer entity;
• using, for each respective policy question associated with the first policy-enforcer entity, the identified machine learning policy model to automatically determine a respective policy question answer to the respective policy question for the request that indicates whether the request conforms to or violates the respective policy question;
• and
• in response to determining that a first policy question answer corresponds to a policy violation:
• generating an audit alert regarding the policy violation that identifies the request as a fraudulent request;
• and
• providing the audit alert to one or more systems, wherein any of the one or more systems that receives the audit alert regarding the policy violation automatically rejects the request based on the audit alert.