arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1405.7295cs.AI
收藏

多上下文系统的代价复杂性

On the cost-complexity of multi-context systems

Delft University of Technology(代尔夫特理工大学)

Peter Novák, Cees Witteveen

摘要:

多上下文系统提供了一个强大的框架,建模信息聚合系统具有异构推理组件。然而,执行这些措施可能会产生不可忽视的成本。在这里,我们专注于这种系统的成本复杂性。为此,我们引入了成本感知的多上下文系统,考虑到执行单个上下文的语义运算符所产生的成本的非单调多上下文系统框架的扩展。我们制定了成本复杂性的概念,在MCS的一致性和推理问题。随后,我们提供了一系列的结果,逐步越来越多的约束类的MCS,最后介绍了一个增量成本降低算法解决的推理问题,确定MCS。

英文摘要:

Multi-context systems provide a powerful framework for modelling information-aggregation systems featuring heterogeneous reasoning components. Their execution can, however, incur non-negligible cost. Here, we focus on cost-complexity of such systems. To that end, we introduce cost-aware multi-context systems, an extension of non-monotonic multi-context systems framework taking into account costs incurred by execution of semantic operators of the individual contexts. We formulate the notion of cost-complexity for consistency and reasoning problems in MCSs. Subsequently, we provide a series of results related to gradually more and more constrained classes of MCSs and finally introduce an incremental cost-reducing algorithm solving the reasoning problem for definite MCSs.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑