System, method and computer readable medium for determining an event generator type
Inventors
Assignees
Lipari Paul • Oracle International Corp • Oracle America Inc • Suboti LLC
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11582139-B2
Publication Date
2023-02-14
Expiration Date
Abstract
Human interaction with a webpage may be determined by processing an event stream generated by the client device during the webpage interaction. A classification server receives the event stream and compares components of the event stream, including components of an event header message, with prerecorded datasets. The datasets include prerecorded event streams having a known interaction type. Training clients may be provided for generating the prerecorded datasets.
Core Innovation
The disclosed invention addresses a method and systems for determining an interaction status and/or categorizing an event generator type for webpage interactions between a client and a webpage server. Client-generated event streams during a client/webpage interaction include an event header message and one or more event messages, where the event header message includes one or more parameters corresponding to hardware components and each event message is associated with an event type.
An analysis server defines event generator types for event types and generates event instances that include the event header message and at least one of the event messages, together with an identifier of the event generator type. The invention stores event instances in a repository of event instances and trains a machine-learning model using the one or more first event messages and the one or more second event messages.
At runtime, the system receives a third event stream and an additional fourth event stream that include an event header message and one or more event messages generated based on webpage interaction. The invention generates event instances including event header parameters and one or more event messages, compares the event instances to individual event instances stored in the repository using the machine-learning model, determines whether an instance matches a particular event instance with a degree of confidence exceeding a threshold value, and categorizes the event stream by assigning the event generator type of the particular event instance and storing the event generator type.
The invention further specifies event header parameters and simulation/device parameters that correspond to hardware components facilitating the webpage interaction, including an identification of a type of operating system, a type of browser, and an indication of a type of client device and/or an indication of a screen size. Dependent claim refinements also include categories for the first event type, including human interaction, computer-based interaction, robot interaction, natural action interaction, simulated input device interaction, and fluctuating input device interaction.
Claims Coverage
The provided excerpt includes four independent claims covering a method, a system, a non-transitory computer readable storage medium, and a system/machine-learning-enabled method with browser-parameter emphasis. Each independent claim centers on training a machine-learning model using labeled event instances generated from event streams and then comparing new event instances against a stored repository to determine matching or non-matching via a confidence threshold, while categorizing by assigning an event generator type.
Training and repository-based event instance generation for event generator type mapping
receiving first training data from a simulation device including a first event stream with a first event header message and one or more first event messages associated with a first event type; defining an event generator type corresponding to the first event type; generating a first event instance including the first event header message, at least one of the one or more first event messages, and an identifier of the event generator type; storing the first event instance in a repository of event instances.
Differentiated labeled event generator types from multiple event types
receiving second training data from the simulation device including a second event stream with a second event header message and one or more second event messages; defining a new event generator type corresponding to a second event type different from the event generator type; generating a second event instance including the second event header message, at least one of the one or more second event messages, and an identifier of the new event generator type; storing the second event instance in the repository of event instances; training a machine-learning model using the one or more first event messages and the one or more second event messages.
Confidence-threshold matching of new event instances to repository instances and assignment of event generator type
receiving a third event stream including a third event header message and one or more third event messages; generating a third event instance including the third event header message comprising one or more third event parameters and at least one of the one or more third event messages; comparing the third event instance to individual event instances stored in the repository using the machine-learning model; determining that the third event instance matches a particular event instance with a degree of confidence exceeding a threshold value; categorizing the third event stream by assigning an event generator type of the particular event instance to the third event stream; storing the event generator type of the third event stream.
Non-matching determination for an additional event stream via confidence threshold
receiving a fourth event stream including a fourth event header message and one or more fourth event messages; generating a fourth event instance including the fourth event header message comprising one or more fourth event parameters and at least one of the one or more fourth event messages; comparing the fourth event instance to the individual event instances stored in the repository of event instances; determining that the fourth event instance does not match an additional particular event instance with the degree of confidence exceeding the threshold value.
Simulation device hardware parameters in event header messages for machine-learning comparison
including in the event header message one or more parameters corresponding to hardware components of the simulation device, where the one or more parameters comprise an identification of a type of operating system, and/or an indication of a type of browser, and where event instances generated from event header parameters are compared using the machine-learning model to repository instances.
Across the independent claims, the shared inventive concept is the training and use of a machine-learning model that compares event instances built from event header parameters and associated webpage-interaction event messages against a stored repository, then determines matches or non-matches using a confidence threshold and categorizes streams by assigning an event generator type associated with the matched repository instance.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?