摘要
在大数据环境下,多源异构知识的融合为研究者从众多分散、异构的数据源和知识源中挖掘出隐含的、有价值的和尚未被发现的信息和知识提供了非常有效的手段和方法。针对目前知识融合方法的不足,在对大数据环境下的异构知识融合方法进行深入研究的基础上,将已有的数据融合算法合理地移植到知识融合中,设计并构造了大数据环境下的多源异构知识融合算法。为进一步提高获取知识的质量,依据知识源粒度的动态选择,提出了一种改进的知识源分解-合并算法,以获得合适粒度大小的知识源集合和尽可能真实可靠的知识。基于Hadoop和MapReduce框架所构建的实验平台对所提算法进行了实验验证。实验结果表明,所提出的多源异构知识融合算法有效可行,并能够有效显著地提高多源异构知识融合算法的性能。
In environment of big data, the integration of multi-source heterogeneous knowledge fusion has provided one of the most effec- tive means and methods for researchers to discover the implicit,valuable and undetected knowledge from a lot of knowledge sources that are dispersed arid heterogeneous. Aimed at the shortcomings of the current knowledge fusion methods, based on investigations on them un- der the big data environment, the existing data fusion methods have been employed, which are transplanted to the knowledge fusion rea- sonably. A kind of algorithm for multi-source heterogeneous knowledge fusion is proposed. In order to filrther improve the quality of the acquiring knowledge, an improved algorithm based on the dynamic selection of knowledge source granularity is proposed to obtain the ap- propriate size of the collection of knowledge sources and the true and reliable knowledge as possible. Its experimental verification is con- ducted based on the experimental plattorm constructed by Hadoop and MapReduce framework. Experimental results show that it is effec- tive and feasible and effectively improves the performance of multi-source heterogeneous knowledge fusion algorithms.
出处
《计算机技术与发展》
2017年第9期12-16,共5页
Computer Technology and Development
基金
国家自然科学基金资助项目(41271387)
关键词
大数据
多源异构知识
知识融合
融合算法
big data
multi-source heterogeneous knowledge
knowledge fusion
fusion algorithm
