基于原语的通用起源过滤框架  

Generic provenance sanitization framework based on primitives

作　　者：孙连山[1] 马胜天 陈秀婷 Sun Lianshan;Ma Shengtian;Chen Xiuting(School of Electronic Information&Artificial Intelligence,Shaanxi University of Science&Technology,Xi’an 710021,China)

机构地区：[1]陕西科技大学电子信息与人工智能学院,西安710021

出　　处：《计算机应用研究》2022年第3期874-878,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(61202019);陕西省自然科学基础研究计划资助项目(2019JM-354)。

摘　　要：现有起源过滤机制的通用性差,一个过滤机制仅能过滤某一特定类型的敏感元素,处理包含多种类型敏感元素的综合性起源过滤需求仍然非常困难,为此提出了一种基于原语的通用起源过滤框架。首先,介绍了起源过滤涉及的敏感元素类型以及过滤约束;其次,深入分析已有过滤机制改造起源图的基本操作和过程,形式地定义了一系列起源过滤原语,描述针对起源图的最小改造操作,将起源过滤过程划分为隐藏敏感元素、恢复有用依赖和验证过滤约束三个阶段,提出了一种基于原语组装的分阶段过滤策略空间构造方法;最后设计并实现了基于原语的通用过滤算法,并在公开数据集上验证了该算法的可行性。The genericity of existing data provenance sanitization mechanisms is very low. One mechanism is usually used to deal with one specific type of sensitive elements. It is still very difficult to deal with comprehensive sanitization requirements including multiple types of sensitive elements in a disciplined manner. To address this issue, this paper proposed a primitive-based generic framework of provenance sanitization. Firstly, this paper introduced the types of sensitive provenance elements and structural constraints that might be involved in data provenance sanitization. Secondly, it thoroughly analyzed existing provenance sanitization mechanisms and formally defined a set of provenance sanitization primitives. Each primitive was a minimal operation for editing a provenance graph. This paper divided the overall process of data provenance sanitization into three stages: hiding sensitive elements, recovering insensitive dependencies, and verifying constraints. Furthermore, it proposed a method for constructing the space of sanitization strategies by selecting and composing possible sanitization primitives stage by stage. Finally, this paper designed and implemented a primitive-based generic provenance sanitization algorithm. The experimental results in public provenance datasets verity the effectiveness of the proposed method.

关 键 词：数据起源 起源过滤 过滤原语 过滤框架 过滤策略空间 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]