Questionnaire Generation Using Tripartite Graph Rules
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional expert and decision support systems face difficulties in representing complex logic and ensuring completeness due to the limitations of decision trees, leading to issues with questionnaire authoring and the need for multiple trees and additional computations, as well as challenges in determining if all necessary questions and answers are covered.
Innovation Solution
The system employs a tripartite graph structure and a rules-based approach for questionnaire generation and completeness checking, using a rules engine to dynamically present questions based on user answers and automatically ensure that all necessary questions are asked, leveraging intermediate variables and monotonic expressions to guarantee deterministic and unambiguous output.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If decision trees are used to represent questionnaires, then the structure is simple and easy to understand, but the system cannot easily represent complex logic and requires duplication of information
Solution Approach 1:
The patent segments the questionnaire representation into separate components: decision nodes, choice nodes, and independent context definitions. Each context is defined independently rather than duplicating information throughout the tree structure. This allows complex logic to be represented without redundant information repetition.
Solution Approach 2:
The patent creates a universal context definition that can be referenced by multiple decision trees and questionnaires. A single context definition can serve multiple purposes across different evaluation scenarios, eliminating the need to redefine the same information in multiple places while maintaining simplicity in each individual view.
2Reliability
If decision trees are used for complex contexts, then the system can provide structured evaluation, but the size of the trees becomes huge and requires additional computation outside the tree
Solution Approach 1:
The patent extracts complex context definitions from within the decision tree structure and places them in separate, reusable context definitions. This removes the bloat from decision trees while preserving the structured evaluation capability. The trees become smaller and more manageable while still providing reliable evaluation through references to external context definitions.
3Adaptability or versatility
If multiple decision trees are created for independent contexts, then each context can be evaluated separately, but the authoring difficulty increases and completeness verification becomes challenging
Solution Approach 1:
The patent creates universal context definitions that can be shared across multiple decision trees and evaluation scenarios. Authors define each context once and reference it wherever needed, significantly reducing authoring difficulty while maintaining the ability to evaluate multiple independent contexts. The system automatically verifies completeness by checking that all required contexts are defined and properly referenced.
4Reliability
If conventional expert systems are made complete to cover all answer combinations, then all necessary questions are answered, but the system becomes too complex to allow provable completeness
Solution Approach 1:
The patent implements automatic completeness verification that checks whether all necessary contexts are defined and all required questions are present. The system self-verifies its own completeness by examining the relationship between context definitions, questionnaire structures, and required answers. This automated approach makes completeness provable without requiring manual verification of all possible answer combinations.
Data Source
AI summary
A questionnaire generation process presents a first subset from a set of questions of the questionnaire and receives first answers from a user. The first answers are used to determine whether the first answers are sufficient to give definite values to conditions of first rules, wherein the first rules have conditions for providing output. When the first answers are not sufficient, the conditions of the first rules can be used to identify a second subset of the questions, wherein the second subset of questions has second answers such that a combination of the first and second answers is sufficient to give definite values to the respective conditions of the first rules, and the second subset of questions can be presented to the user.