介绍Markov逻辑网的理论模型,阐述Markov逻辑网的结构和参数学习算法及2种基本类型的推理,从命名实体识剐、实体关系抽取和实体解析3个方面总结Markov逻辑网在信息抽取中的应用现状.分析结果表明,Markov逻辑网模型能较好地将一阶谓词逻辑和概率图模型相结合,灵活地在Markov网中融入模块化知识,描述复杂的特征.%This paper addresses the theoretical model, structure and parameter learning algorithms, two kinds of inference problems of Markov Logic Network(MLN) and its applications in Information Extraction(IE), including named entity recognition, entity relation extraction and entity resolution. MLN model can combine one-order predicate logic and probability graph model, fuse modular knowledge into Markov network, and describe complicate characteristics.
展开▼
