Workflow Generation Using Weighted KPI Fluctuation Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for policy planning face challenges in generating new policies when predicted KPIs do not reach the target, leading to increased computational load due to exponential combinations of synthesized policies, especially when many parameters are varied.
Innovation Solution
A computer-implemented method generates candidate workflows by modifying existing workflows, obtaining KPI fluctuation vectors, synthesizing these vectors, and calculating predicted values with weight parameters that favor upper modifications over lower ones to achieve target KPI values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all synthesized policies are simulated to find policies that reach the goal, then a policy candidate that can reach the goal can be found, but the computational load increases to an unrealistic level
Solution Approach 1:
The patent segments the policy synthesis process by dividing synthesized policies into groups based on the number of modifications. Instead of evaluating all 2^N synthesized policies, the system evaluates policies with 1 modification, then 2 modifications, and so on, until a goal-achieving policy is found. This segmentation dramatically reduces computational load while maintaining the ability to find effective policies.
Solution Approach 2:
The patent applies partial action by evaluating only a subset of synthesized policies rather than all possible combinations. The system performs KPI prediction on synthesized policies with increasing numbers of modifications (1, 2, 3, ...) and stops when a policy meeting the goal is found, avoiding the excessive computational effort of evaluating all 2^N combinations.
2Adaptability or versatility
If many parameters that can be varied in the policy are present, then policy flexibility and adaptability increase, but combination of synthesized policy becomes enormous and computational load increases
Solution Approach 1:
The patent segments the vast combination space by organizing synthesized policies according to the number of modifications applied. This segmentation transforms the overwhelming 2^N combination problem into a manageable sequence of smaller evaluation tasks, where policies with 1 modification are evaluated first, then 2 modifications, and so on.
Solution Approach 2:
The patent introduces dynamics by adaptively adjusting the evaluation strategy based on results. The system dynamically determines how many modifications to evaluate next based on whether previous policies met the goal, allowing flexible exploration of the parameter space without being constrained by fixed combinatorial complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for generating a workflow includes: generating multiple candidate workflows by making a modification to part of an existing workflow that defines multiple condition branches and following contents at respective destinations of the condition branches; obtaining KPI fluctuation vectors one for each candidate workflow from the existing workflow; generating multiple synthesized policies by synthesizing two or more of the KPI fluctuation vectors; calculating, when one of selected routes of multiple candidate workflows contains the modifications, a KPI predicted value of each synthesized policy by reflecting a weight parameter on the KPI fluctuation vector, the weight parameter being set such that a weight for a modification to an upper point of the selected route is higher than that to a lower point of the selected route among the modifications; and outputting a workflow of one of the synthesized policies the KPI predicted value of which satisfies a target value.