Dynamic Utility Threshold Mining for High-Utility Itemsets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current high-utility itemset mining methods face challenges in accurately identifying high-utility itemsets due to the use of fixed minimum utility thresholds, which can lead to inaccurate results as different data items may correspond to different thresholds, affecting the accuracy of the mining process.
Innovation Solution
A method that determines a minimum itemset utility threshold for each itemset based on the lowest minimum utility threshold of its data items, allowing for a more accurate comparison and identification of high-utility itemsets by using a predefined minimum utility threshold table to set itemset utility thresholds dynamically.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a fixed minimum utility threshold is used for all itemsets, then the mining process is simple and fast, but the accuracy of identifying high-utility itemsets deteriorates
Solution Approach 1:
The patent applies local quality by assigning different minimum utility thresholds to different itemsets based on their specific characteristics. Instead of using a single fixed threshold for all itemsets, the system determines itemset-specific thresholds by considering the minimum utility thresholds of individual data items within each itemset. This localized approach improves the accuracy of high-utility itemset identification while maintaining reasonable computational complexity.
2Reliability
If a fixed minimum utility threshold is used for all itemsets, then the mining process is simple and fast, but the reliability of mining results deteriorates
Solution Approach 1:
The patent implements parameter changes by dynamically adjusting the minimum utility threshold parameter for each itemset based on the characteristics of its constituent data items. The system retrieves minimum utility thresholds for individual items from a predefined table and determines the itemset's threshold accordingly. This parameter adaptation improves result reliability while the efficient retrieval and comparison processes minimize the impact on mining productivity.
3Measurement precision
If itemset utility thresholds are set dynamically based on data item thresholds, then the accuracy of mining results is improved, but the computational complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-defining minimum utility thresholds for individual data items in a lookup table before the actual itemset mining process. This preliminary preparation allows the system to quickly retrieve and apply appropriate thresholds during mining without performing complex calculations in real-time. The pre-computed itemset thresholds are determined by finding the minimum among item thresholds, which simplifies the overall computational complexity while maintaining high accuracy.
Data Source
AI summary
A pattern mining method includes obtaining, according to each of transactions comprised in a transactional database, a candidate pattern set satisfying a condition, each of the transactions comprising at least one item, each of candidate patterns in the candidate pattern set comprising one or more items in an itemset, calculating a utility of a respective one of the candidate patterns in each of the transactions, determining, among the transactions, at least one target transaction in which the utility reaches a specified utility threshold, determining a target period value of a target candidate pattern corresponding to the at least one target transaction, according to a target time attribute of each of the at least one target transaction, and determining the target candidate pattern as a mining result, based on the target period value of the target candidate pattern being less than or equal to a period threshold.