面向交互式建模的科学工作流活动推荐方法

Scientific workflow activity recommendation for interactive modeling

为提高科学工作流建模的效率和智能性,针对现有科学工作流管理系统在支持流程建模方面的特点以及用户的交互式建模需求,提出一种面向交互式建模的科学工作流活动推荐方法。该方法以科学工作流库为起点,将科学工作流片段作为推荐参考模型,主要包括聚类预处理、匹配度计算等步骤,可根据构建中的科学工作流模型及用户建模需求,推荐合适的科学工作流活动。基于真实数据集的评估实验验证了该算法的有效性。

How to effectively recommend scientific workflow is a burning problem of importance. However, scientific workflow activity recommendation is ignored by most of existing researches. To facilitate scientific workflow design and improve its efficiency, by considering the characteristic of scientific workflow management system environment and user demand for interactive modeling, a method of scientific workflow activity recommendation for interactive modeling was proposed, which was named Scientific Workflow Activity Recommendation for interactive modeling(SWAR). It was started from scientific workflow repository and utilized scientific workflow fragments as reference models to provide recommendation in a context-sensitive manner. After clustering preprocessing scientific workflow fragments and computing the similarities between reference models and incomplete scientific workflow model, candidate scientific workflow activity set could be determined for recommendation. The result of experiments based on real datasets showed its effectiveness.