基于MapReduce计算框架的文档领域本体模型构建

Construction of Document Domain Model Based on MapReduce Computing

针对传统文档表示模型中语义关系缺失、特征词权重计算单一及海量数据的实时处理等问题,基于领域本体概念间的语义关系,结合MapReduce框架,提出一种特征权重自适应增强的文档领域本体模型(EAS-VSM)构建算法。该算法通过构造概念语义关系矩阵,将领域本体中概念之间的语义关系增强至每一个概念特征词中,从而实现概念特征词权重的自适应增强。实验结果显示,算法的加速比和可扩展性两项指标与数据规模呈明显的线性关系,证实模型的并行算法性能良好,且相较于传统的VSM和LSA模型,EAS-VSM模型计算的结果与专家经验更为接近,更能反映文档间的相似程度。

Aiming at the problems of missing semantic relations,single feature word weight calculation and real-time processing of massive data in traditional document representation models,based on the semantic relations between domain ontology concepts,combined with MapReduce framework,this paper proposes an eigenvector adaptive strengthening VSM(EAS-VSM)based on the conceptual semantic relationship and weight adaptation mechanism of domain ontology.By constructing the concept semantic relation matrix,the algorithm enhances the semantic relation between concepts in domain ontology to each concept feature word,so as to realize the adaptive enhancement of the weight of concept feature words.The experimental results show that the acceleration ratio and scalability of the algorithm have an obvious linear relationship with the data size,which proves that the parallel algorithm of the model has good performance.Compared with the traditional VSM model and the LSA model,the calculation results of EAS-VSM model are closer to the expert experience and can better reflect the similarity between documents.