At present, there are some static code analyses and optimizations that can be applied to Concurrent C programs to improve their performance or verify their logical correctness. These analyses and optimizations are int...At present, there are some static code analyses and optimizations that can be applied to Concurrent C programs to improve their performance or verify their logical correctness. These analyses and optimizations are inter-process. In order to make their implementation easy, we propose a new method to construct an optimizing compiling system CCOC for Concurrent C. CCOC supports inter-process code analysis and optimization to Concurrent C programs and does not affect the system's portability and separate compilation of source programs. We also discuss some implementation details of CCOC briefly.展开更多
预测性过程监控(predictive process monitoring,PPM)技术通过分析已有的事件日志对正在运行的业务流程中的某些关键指标进行预测.目前的预测性过程监控方法在特征提取方面通常预设案例仅受自身属性的影响或仅通过提取资源案例间行为属...预测性过程监控(predictive process monitoring,PPM)技术通过分析已有的事件日志对正在运行的业务流程中的某些关键指标进行预测.目前的预测性过程监控方法在特征提取方面通常预设案例仅受自身属性的影响或仅通过提取资源案例间行为属性进行编码,通常未涉及活动视角下的案例间行为信息.本文提出了一种捕获活动案例间行为信息的方法IABC(inter-activity behaviour of case),该方法设计了一个三维特征构建框架涵盖时间窗口、活动粒度、行为状态这3大维度,共构造36种活动案例间行为特征.同时,本文提出了两项关键算法:影响力分布算法用于挖掘活动间的正/负影响传播;批次行为检测算法用于识别潜在的批量操作行为.最后在3个公开可用的事件日志上评估IABC方法的有效性,结果表明集成了IABC方法的时间预测模型性能优于未使用该方法的基线模型和采用了资源案例间特征的模型.展开更多
摘要At present, there are some static code analyses and optimizations that can be applied to Concurrent C programs to improve their performance or verify their logical correctness. These analyses and optimizations are inter-process. In order to make their implementation easy, we propose a new method to construct an optimizing compiling system CCOC for Concurrent C. CCOC supports inter-process code analysis and optimization to Concurrent C programs and does not affect the system's portability and separate compilation of source programs. We also discuss some implementation details of CCOC briefly.
摘要预测性过程监控(predictive process monitoring,PPM)技术通过分析已有的事件日志对正在运行的业务流程中的某些关键指标进行预测.目前的预测性过程监控方法在特征提取方面通常预设案例仅受自身属性的影响或仅通过提取资源案例间行为属性进行编码,通常未涉及活动视角下的案例间行为信息.本文提出了一种捕获活动案例间行为信息的方法IABC(inter-activity behaviour of case),该方法设计了一个三维特征构建框架涵盖时间窗口、活动粒度、行为状态这3大维度,共构造36种活动案例间行为特征.同时,本文提出了两项关键算法:影响力分布算法用于挖掘活动间的正/负影响传播;批次行为检测算法用于识别潜在的批量操作行为.最后在3个公开可用的事件日志上评估IABC方法的有效性,结果表明集成了IABC方法的时间预测模型性能优于未使用该方法的基线模型和采用了资源案例间特征的模型.