Systems and methods for identifying illegitimate activities based on historical data
AssigneeFacebook, Inc.
InventorDaniel Adam Jenson
Life of this patent
Estimating the term…
Abstract
Systems, methods, and non-transitory computer-readable media can acquire historical data including a plurality of features associated with known legitimate activities and with known illegitimate activities. A machine learning technique can be applied to the historical data to gain information about the plurality of features associated with the known legitimate activities and with the known illegitimate activities. A decision tree can be generated based on at least a portion of the information about the plurality of features. A node in the decision tree that satisfies specified precision criteria can be identified. A rule can be created based on the node. One or more illegitimate activities can be identified based on the rule.
Claims17
A computer-implemented method comprising:
- acquiring, by a computing system, historical data including a plurality of features associated with known legitimate activities and with known illegitimate activities, wherein at least some of the known legitimate activities and the known illegitimate activities include one or more financial transactions;
- applying, by the computing system, a machine learning technique to the historical data to gain information about the plurality of features associated with the known legitimate activities and with the known illegitimate activities, the applying comprising:
- determining a respective information gain for each feature in the plurality of features associated with the known legitimate activities and with the known illegitimate activities;
- generating, by the computing system, a decision tree based on at least a portion of the information about the plurality of features;
- identifying, by the computing system, a node in the decision tree that satisfies specified precision criteria;
- creating, by the computing system, a rule based on the node identified in the decision tree, wherein the rule corresponds to a conditional rule which indicates that a particular activity is illegitimate when one or more features associated with the particular activity respectively meet one or more feature values specified by the rule;
- and
- identifying, by the computing system, one or more illegitimate activities based on the rule.
A system comprising:
- at least one processor;
- and
- a memory storing instructions that, when executed by the at least one processor, cause the system to perform:
- acquiring historical data including a plurality of features associated with known legitimate activities and with known illegitimate activities, wherein at least some of the known legitimate activities and the known illegitimate activities include one or more financial transactions;
- applying a machine learning technique to the historical data to gain information about the plurality of features associated with the known legitimate activities and with the known illegitimate activities, the applying comprising:
- determining a respective information gain for each feature in the plurality of features associated with the known legitimate activities and with the known illegitimate activities;
- generating a decision tree based on at least a portion of the information about the plurality of features;
- identifying a node in the decision tree that satisfies specified precision criteria;
- creating a rule based on the node identified in the decision tree, wherein the rule corresponds to a conditional rule which indicates that a particular activity is illegitimate when one or more features associated with the particular activity respectively meet one or more feature values specified by the rule;
- and
- identifying one or more illegitimate activities based on the rule.
A non-transitory computer-readable storage medium including instructions that, when executed by at least one processor of a computing system, cause the computing system to perform:
- acquiring historical data including a plurality of features associated with known legitimate activities and with known illegitimate activities, wherein at least some of the known legitimate activities and the known illegitimate activities include one or more financial transactions;
- applying a machine learning technique to the historical data to gain information about the plurality of features associated with the known legitimate activities and with the known illegitimate activities, the applying comprising:
- determining a respective information gain for each feature in the plurality of features associated with the known legitimate activities and with the known illegitimate activities;
- generating a decision tree based on at least a portion of the information about the plurality of features;
- identifying a node in the decision tree that satisfies specified precision criteria;
- creating a rule based on the node identified in the decision tree, wherein the rule corresponds to a conditional rule which indicates that a particular activity is illegitimate when one or more features associated with the particular activity respectively meet one or more feature values specified by the rule;
- and
- identifying one or more illegitimate activities based on the rule.
Dependencies are inferred from the claim text.