一种软件缺陷预测规则的筛选方法及系统
By filtering frequent itemsets using multiple support and lift thresholds, and employing dual-confidence redundancy pruning rules, the problems of redundant generation and class imbalance in association rule algorithms are solved, achieving more efficient software defect prediction.
CN116185817BActive Publication Date: 2026-07-17BEIHANG UNIV
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIHANG UNIV
- Filing Date
- 2022-11-30
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
In existing software defect prediction technologies, association rule algorithms suffer from problems such as redundant rule generation, low efficiency, and failure to effectively handle class-imbalanced data.
Method used
Frequent itemsets are filtered using multiple support thresholds and lift thresholds, and redundant rules are pruned using dual confidence. The prediction accuracy and efficiency are improved by using a class association rule algorithm.
Benefits of technology
It improves the accuracy and efficiency of software defect prediction, reduces redundant rules, and enhances the performance and generalization ability of association rule algorithms.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116185817B_ABST
Abstract
本发明涉及一种软件缺陷预测规则的筛选方法及系统，属于关联规则筛选技术领域，解决了现有软件缺陷特征选择复杂且预测规则存在冗余的问题。包括获取历史软件缺陷数据，构建样本集；基于样本集执行如下步骤，进行迭代训练和测试：将样本集划分为训练集和测试集；基于关联规则算法，根据三个支持度阈值从训练集中生成频繁项集，根据不同长度的频繁项集的提升度阈值，筛选出频繁项集并转化为关联规则，得到关联规则集合；从关联规则集合中提取类关联规则，根据选择的预测指标对测试集进行预测，根据预测结果计算分类性能指标；迭代训练和测试结束后，取分类性能指标最优时的类关联规则，作为软件缺陷预测规则。提高了筛选预测规则的准确性和高效性。
Need to check novelty before this filing date? Find Prior Art