跨语言实体关系抽取分析-计算机应用技术专业论文.docxVIP

跨语言实体关系抽取研究中文摘要 跨语言实体关系抽取研究 中文摘要 I I 跨语言实体关系抽取研究 中文摘要 语料库的数量和质量对基于机器学习的命名实体间语义关系抽取的性能具有重 要的影响，而语料库的人工标注是一项十分耗时又费力的工作。另一方面，多语言语 料库的出现以及机器翻译技术的发展，为研究多种语言之间的一致性和互补性提供了 良好的契机。本文提出了三种跨语言关系抽取的方法，旨在提高多语言关系抽取性能 的同时降低对语料库标注量的需求。研究内容包括以下三个方面： 1）基于机器翻译的跨语言关系抽取。首先利用机器翻译产生翻译语料，然后进 行实体对齐，最后直接将映射后的关系实例添加到目标语言训练集中，帮助目标语言 进行关系抽取； 2）基于双语协同训练的关系分类。给定小规模标注语料和大规模未标注语料， 以自举的方式将一种语言中可靠分类的关系实例所对应的翻译实例添加到另一种语 言的训练语料库中。两种语言上的关系分类互相帮助，共同提高； 3）基于双语主动学习的关系分类。将主动学习方法应用到中英文双语关系分类 中，在挑选信息量最丰富的未标注实例时，使用联合置信度来综合考虑一个实例在两 种语言中的分类置信度。 在 ACE 2005 中/英文语料库上的关系抽取实验表明，无论以何种方式使用翻译语 料，一种语言中人工标注和自动标注的关系实例对另一种语言的关系抽取都具有一致 且稳定的帮助作用，并且当训练语料规模较小时，这种帮助就尤为显著。 关键词：实体关系抽取，机器翻译，实体对齐，协同训练，主动学习  作者：胡亚楠 指导老师：钱龙华 Abs Abstract Research on Cross-lingual Relation Extraction between Named Entities Research on Cross-lingual Relation Extraction between Named Entities Abstract The quantity and quality of training corpora have an important impact on the performance of machine learning-based semantic relation extraction between named entities, however, the annotation of corpora is time-consuming and labor-intensive. Meanwhile, the emergence of multi-language corpora and the development of machine translation technology provide a good opportunity for the research on redundancy and complementariness between languages. This paper proposes three approaches for cross-lingual relation extraction, aiming at improving multilingual relation extraction performance as well as reducing the amount of needed labeled corpora. The study includes three aspects as follows: Cross-lingual relation extraction via machine translation. First, we obtain translated corpora via machine translation, then perform entity alignment, and finally add these mapped translated instances into the training corpus of the target language to help its relation extraction. Bilingual co-training for relation classification. Given a small number of labeled instances and a large number of unlabeled instances in two languages, translated instances which correspond to reliably