System and Method for Automatic Learning and Selecting of a Candidate Workflow Representative of Machine Learning Flow Set
AssigneeAtlantic Technical Organization
InventorArturo Geigel
Life of this patent
Estimating the term…
Abstract
A system and method for determining a candidate workflow from a cluster of similar workflows is presented. The process uses the differences classified as insertions of operators, deletions of operators, transpositions of operators and operator shifting in a parallel workflow to determine similarities in the workflow cluster and extract a candidate similar to the workflow in the comparison query. The extracted candidate workflow can then be used to suggest modifications to the workflow in the comparison query.
Claims18
A method for selecting a workflow candidate representative of a workflow set wherein each workflow of said workflow set comprises a plurality of operators, comprising the steps of:
- calculating a number of instances for each different operator in said plurality of operators in each workflow of said workflow set;
- calculating a frequency of each different operator in said plurality of operators in all the workflows of said workflow set;
- establishing an operator cutoff threshold and an operator addition threshold based on a maximum frequency from said calculation of frequency of each different operator;
- eliminating all instances of each operator from said plurality of operators that has a frequency below said operator cutoff threshold;
- creating a provisional operator set that includes each operator from said plurality of operators that has a frequency above said operator cutoff threshold and below said operator addition variable;
- creating a critical operator set that includes each operator from said plurality of operators that has a frequency above said operator addition threshold;
- assigning an addition score to each workflow from said workflow set that contains at least one operator from said critical operator set, wherein said addition score is based on the total amount of operators in each workflow;
- assigning a provisional score to each workflow from said workflow set that contains at least one operator from said provisional operator set, wherein said provisional score is based on the total amount of operators in each workflow;
- assigning an addition weight to each operator in said critical operator set;
- assigning a provisional weight to each operator in said provisional operator set;
- calculating a workflow score for each workflow from said workflow set that contains at least one operator from said critical operator set and contains at least one operator from said provisional operator set, wherein said workflow score is based on said addition weight, said provisional weight and the amount of operators in each workflow;
- selecting a candidate workflow from said workflow set, wherein said candidate workflow contains the highest workflow score from said workflow set.
A system for selecting a workflow candidate representative of a workflow set wherein each workflow of said workflow set comprises a plurality of operators, comprising:
- one or more computer processors;
- one or more computer readable storage devices;
- program instructions stored on said one or more computer readable storage devices for execution by at least one of said one or more computer processors, said stored program instructions comprising:
- program instructions for calculating a number of instances for each different operator in said plurality of operators in each workflow of said workflow set;
- program instructions for calculating a frequency of each different operator in said plurality of operators in all the workflows of said workflow set;
- program instructions for establishing an operator cutoff threshold and an operator addition threshold based on a maximum frequency from said calculation of frequency of each different operator;
- program instructions for eliminating all instances of each operator from said plurality of operators that has a frequency below said operator cutoff threshold;
- program instructions for creating a provisional operator set that includes each operator from said plurality of operators that has a frequency above said operator cutoff threshold and below said operator addition variable;
- program instructions for creating a critical operator set that includes each operator from said plurality of operators that has a frequency above said operator addition threshold;
- program instructions for assigning an addition score to each workflow from said workflow set that contains at least one operator from said critical operator set, wherein said addition score is based on the total amount of operators in each workflow;
- program instructions for assigning a provisional score to each workflow from said workflow set that contains at least one operator from said provisional operator set, wherein said provisional score is based on the total amount of operators in each workflow;
- program instructions for assigning an addition weight to each operator in said critical operator set;
- program instructions for assigning a provisional weight to each operator in said provisional operator set;
- program instructions for calculating a workflow score for each workflow from said workflow set that contains at least one operator from said critical operator set and contains at least one operator from said provisional operator set, wherein said workflow score is based on said addition weight, said provisional weight and the amount of operators in each workflow;
- program instructions for selecting a candidate workflow from said workflow set, wherein said candidate workflow contains the highest workflow score from said workflow set.
Dependencies are inferred from the claim text.