Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Assignees
Member
SRI InternationalSRI International is an independent nonprofit research institute with a rich history of supporting government and industry. For almost 80 years, SRI has collaborated across technical and scientific disciplines to discover and develop groundbreaking products and technologies, delivering world-changing solutions for a safer, healthier, and more sustainable future. The Nomura-SRI Innovation Center (NSIC) further enhances SRI's mission by bridging the gap between Japan and Silicon Valley, fostering corporate innovation and talent development.
SRI International is an independent nonprofit research institute with a rich history of supporting government and industry. For almost 80 years, SRI has collaborated across technical and scientific disciplines to discover and develop groundbreaking products and technologies, delivering world-changing solutions for a safer, healthier, and more sustainable future. The Nomura-SRI Innovation Center (NSIC) further enhances SRI's mission by bridging the gap between Japan and Silicon Valley, fostering corporate innovation and talent development.
Publication Number
US-11941012-B2
Publication Date
2024-03-26
Expiration Date
2039-06-17
Abstract
In general, the disclosure describes techniques for identifying sequences of user actions from event data and logs of user actions for at least one user of a computing system. In one example, a system includes a sequence mining unit that processes event data and logs of user actions for at least one user of a computing system to obtain a set of one or more candidate action sequences each comprising a sequence of one or more user actions. A sequence filtering unit of the system applies, to the set of one or more candidate action sequences, one or more filters informed by a model of user actions for an application domain to obtain a set of one or more filtered action sequences to improve a quality of action sequences identified by the system. An output device of the system outputs an indication of the set of one or more filtered action sequences usable for generating at least one automated workflow or information usable for improving a workflow.
Core Innovation
The invention describes a system and method for identifying sequences of user actions from event data and logs corresponding to at least one user in a computing system. The process involves a sequence mining unit that processes event data and user action logs to obtain candidate action sequences, each containing one or more user actions. These candidate sequences undergo further processing by a sequence filtering unit, which applies filters based on a user action model specific to the application domain, yielding a set of filtered action sequences that are intended to accurately reflect real user workflows.
The core problem addressed is that conventional statistical sequence mining techniques, which analyze user action logs to find repeated patterns, often yield many sequences that are statistically frequent but not semantically meaningful or causally coherent. Statistically driven approaches require large datasets and are not effective when user action sequences are noisy, contain extraneous actions, or are composed of actions without meaningful parameters, preconditions, or effects. These limitations make prior-art systems impractical for broader adoption or workflow automation applications.
By integrating domain-specific user action models, which specify the parameters, preconditions, and effects of user actions within an application domain, the described system filters out invalid or irrelevant action sequences and actions, ensuring higher quality in the mined sequences. The output can be used to generate automated workflows, provide explanations for workflow automation, and suggest workflow improvements. The approach enables the learning and automation of meaningful user workflows from smaller data sets than conventional methods and offers new functionality such as automated procedure learning and workflow explanation.
Claims Coverage
There are three primary inventive features covered by the independent claims.
System for identifying and filtering user action sequences based on a user action model
A system comprising: - A sequence mining unit that identifies repeating patterns of one or more user actions from logs and event data for at least one user, generating candidate action sequences. - A sequence filtering unit that filters these candidate sequences using filters (candidate filters and action filters) derived from a model of user actions for the specific application domain. The unit can discard either invalid entire sequences or invalid individual actions within a sequence, depending on the filter type. - A workflow automation unit that generates automated workflows from the filtered sequences and produces information based on the workflow and user action model. - An output device that outputs the automated workflow for performance or provides information for workflow modification.
Method for identifying, filtering and automating user action sequences
A method including: 1. Identifying, via a sequence mining unit, patterns of one or more repeating user actions from logs and event data, creating candidate action sequences. 2. Obtaining filtered action sequences by applying at least one candidate filter or action filter (as derived from the user action model) to discard invalid sequences or actions. 3. Generating, by a workflow automation unit, an automated workflow from the filtered action sequences and producing information based on the workflow and user action model. 4. Outputting, by an output device, either the automated workflow for automation or information for user workflow modification.
Non-transitory computer-readable medium for executing the action sequence mining and filtering process
A non-transitory computer-readable medium with instructions that, when executed, cause processing circuitry to: - Identify repeated patterns of user actions from event data and logs to generate candidate action sequences. - Apply, to these sequences, one or more candidate or action filters derived from an action model for the application domain, discarding invalid sequences or actions and producing filtered action sequences. - Generate an automated workflow and relevant information using the filtered sequences and action model. - Output either the automated workflow for system performance or information suitable for workflow modification.
These inventive features define a system, a method, and a computer-readable medium for mining, filtering, and automating user action sequences by leveraging filters based on user action models tailored to a given application domain.
Stated Advantages
The techniques enable improvement in the quality of extracted action sequences compared to conventional statistical sequence mining, resulting in higher quality and more useful data for workflow automation and analysis.
The system can learn useful action sequences from a much smaller amount of data than purely statistical techniques, allowing its use where large datasets are not easily obtainable.
Filtered action sequences better reflect higher level human workflows, making them more useful and relevant for generating automated workflows and recommendations to improve workflows.
The use of user action models enables generation of explanations for filtered action sequences and the user workflows they represent, supporting transparency in workflow automation.
Improved action sequence quality allows for more meaningful analysis, enabling not only automation of workflows but also the autonomous determination of when to proactively initiate such workflows.
Documented Applications
Generating automated workflows from filtered user action sequences for automation of user tasks in computing systems.
Providing explanations for workflow automations or performance models generated from filtered user action sequences.
Generating suggestions for improving the workflows of users based on detailed analysis of user action logs.
Learning and generalizing user procedures from filtered action sequences to create parameterized workflows for a broad range of settings, applications, or environments.
Interested in licensing this patent?