Causal Probe for Explainable Decision Service Logic
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current decision services lack the ability to provide transparent and explainable outcomes, as traditional methods fail to capture the reasoning behind complex decisions, making it difficult for users and business analysts to understand and improve decision logic.
Innovation Solution
A computer-implemented method and system that includes a decision service with an interface and decision logic, where a causal probe records the causal history of the decision-making process, allowing for the identification of only the necessary components required to determine outputs, and generates an explainable causal model for post-execution explanations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional decision services are used, then decision logic executes efficiently, but the decision-making process lacks transparency and explainability
Solution Approach 1:
The patent applies preliminary action by pre-defining a causal model structure and inserting causal probes into the decision logic before execution. The causal model is prepared in advance with defined causal relationships between inputs, decision logic components, and outputs, allowing the system to capture relevant causal information without ad-hoc processing during decision execution.
Solution Approach 2:
The patent uses an intermediary approach by introducing a causal probe as a mediator between the decision logic and the output. The causal probe captures causal information and passes it to the causal model, which then generates explanations. This intermediary layer enables transparent decision-making without directly modifying the core decision logic.
2Loss of information
If complete decision logic is captured for explanations, then transparency is improved, but runtime performance deteriorates
Solution Approach 1:
The patent applies the extraction principle by isolating only the necessary causal components from the complete decision logic. The causal model extracts and captures only the relevant causal relationships that contribute to the final output, excluding unnecessary components. This selective extraction maintains explanation completeness while reducing the overhead during runtime execution.
3Ease of operation
If detailed causal history is recorded, then explainability is improved, but system complexity increases
Solution Approach 1:
The patent applies universality by designing a causal model that serves multiple functions: it captures causal relationships, generates explanations, and maintains transparency. The unified causal model framework handles various decision logic types and output formats, reducing the need for separate mechanisms for different explanation scenarios and thereby managing complexity.
4Loss of information
If causal probes are inserted into decision logic, then transparency is improved, but execution overhead increases
Solution Approach 1:
The patent applies extraction by removing unnecessary causal capture operations from the decision logic execution path. The causal probe is designed to capture only essential causal information that contributes to the final output, extracting and recording only relevant data points. This minimizes the overhead introduced by the causal probe while maintaining adequate explanation capability.
Data Source
AI summary
A computer implemented method comprising accessing a decision service, determining a subset of the decision logic, inserting a causal probe into the decision service, receiving a query at the interface of the decision service, executing the decision logic to determine the one or more outputs for the decision service for the received query, and outputting the one or more outputs and the causal history from the decision service for the received query. The causal history can be accessed at a later date and used to generate a causal model that can be used to determine an explanation for the original decision.