Test Assembly Algorithm Using Tabu Search for Non-Overlapping Pools
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated test assembly methods struggle to efficiently assemble multiple non-overlapping tests while ensuring optimal use of items and passages, often leading to suboptimal solutions due to sequential assembly techniques that remove used items and passages, which can block further assembly of non-overlapping tests.
Innovation Solution
The implementation of a random search method that utilizes a hierarchical representation of test constraints and a 'divide and conquer' approach to shrink the search domain, allowing for the assembly of multiple non-overlapping sections and tests by prioritizing constraint checking and using tabu search to reduce infeasible combinations, thereby ensuring optimal assembly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If sequential assembly technique is used to assemble tests by removing used items and passages from the pool, then the assembly process is simplified, but the optimal solution is blocked because removal of items can prevent further assembly of non-overlapping tests
Solution Approach 1:
The invention divides the test assembly problem into two distinct phases: (1) generating multiple candidate tests that satisfy all constraints simultaneously, and (2) selecting the maximum number of non-overlapping tests from these candidates. This segmentation allows the system to avoid the pitfalls of sequential assembly while maintaining computational tractability.
Solution Approach 2:
The invention performs preliminary generation of multiple candidate tests before making final selections. By pre-generating a pool of valid candidate tests that all satisfy constraints, the system enables subsequent optimization steps to find the maximum non-overlapping set without having to make irreversible removal decisions during assembly.
2Reliability
If the entire search space of item combinations is explored to ensure optimal assembly, then the optimal solution is found, but the computational time and resources increase significantly
Solution Approach 1:
The invention extracts and utilizes structural properties and constraints of the test assembly problem to define a reduced search space. By identifying and enforcing constraints early in the candidate generation process, the system eliminates large portions of the search space that would not yield valid solutions, thereby reducing computational effort while maintaining optimality.
Solution Approach 2:
The invention employs dynamic programming or iterative optimization approaches where the search strategy adapts based on feedback from previous iterations. The system learns from constraint violations and adjusts the candidate generation process to focus on promising regions of the search space, improving efficiency over time.
Data Source
AI summary
Creation of a standardized test, that includes a sub-pool of questions, involves assembling the sub-pool from a pool of questions. The sub-pool satisfies one or more constraints. Multiple mutually disjoint sub-pools of questions are assembled from the pool of questions.