User-Specific XAI Explanation Generation via Variable Contribution Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing explainable artificial intelligence (XAI) methods struggle to provide user-specific explanatory information that is convincing to all users, as the importance of factors in machine-learning model predictions varies based on individual expertise and knowledge, leading to a risk of non-explainable information presentation.
Innovation Solution
A non-transitory computer-readable recording medium stores a program that calculates contribution degrees of data variables for machine-learning model predictions, allowing for user-specific variable selection and customized explanatory information generation based on user-oriented data, using an interpretable model to present relevant factors to each user.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single explanatory information is provided for all users, then the system complexity is reduced, but the explainability and user conviction are insufficient because factor importance varies by user expertise
Solution Approach 1:
The patent segments the user base into different user types based on their expertise levels. Instead of providing a single uniform explanation, the system divides explanations into multiple versions tailored to each user type, thereby improving explainability without significantly increasing overall system complexity through structured segmentation.
Solution Approach 2:
The patent implements dynamic explanation generation that adapts to each user's expertise level. The system dynamically selects and adjusts explanatory content based on user characteristics, making the explanation process flexible and adaptive rather than static and one-size-fits-all.
2Loss of information
If contribution degrees are calculated for all variables, then the comprehensiveness of explanation is improved, but the information overload increases and key insights become diluted
Solution Approach 1:
The patent extracts and highlights only the most important factors for each user type rather than presenting all variables. By taking out and focusing on key contributing factors, the system maintains completeness of important information while avoiding information overload and improving usability.
Solution Approach 2:
The patent applies local quality by providing different levels of detail and focus for different user types. Each user receives explanations with appropriate depth and emphasis on factors most relevant to their expertise level, rather than uniform treatment of all variables.
3Productivity
If generic explanatory information is used, then the processing speed is maintained, but the user conviction and understanding are reduced due to lack of personalization
Solution Approach 1:
The patent performs preliminary classification of users into different types based on their expertise levels before generating explanations. This preliminary action allows the system to prepare and select appropriate explanation templates in advance, maintaining processing speed while enabling personalized explanations that improve user conviction.
Data Source
AI summary
A non-transitory computer-readable recording medium stores a program for causing a computer to execute a process, the process includes calculating a contribution degree of each of a plurality of pieces of data each including a plurality of variables, with respect to a prediction result that is output by a machine-learning model in response to input of the plurality of pieces of data, by using an explanatory model generated based on the prediction result and the plurality of pieces of data, selecting a specific variable from among the plurality of variables, determining specific data among the plurality of pieces of data based on a value of the specific variable of each of the plurality of pieces of data and the contribution degree of each of the plurality of pieces of data, and outputting the specific data as explanatory information of the prediction result.