CAJ | 학술논문

实体关系抽取是信息抽取的一项重要内容，总结现有的方法对于该领域的发展具有指导和借鉴意义。结合当前的研究进展，分析和比较了有监督、无监督和弱监督3类关系抽取方法的原理和代表性算法，总结了各类方法的特性并对关系抽取的发展趋势进行了展望。
실체관계추취시신식추취적일항중요내용，총결현유적방법대우해영역적발전구유지도화차감의의。결합당전적연구진전，분석화비교료유감독、무감독화약감독3류관계추취방법적원리화대표성산법，총결료각류방법적특성병대관계추취적발전추세진행료전망。
Relation extraction is an important section of information extraction, summarization of the existing methods is instructional for the developing of this field. Combined with the current research status, firstly, analyzed and compared the principle and representative algorithms of three relation extraction methods, including supervised, unsupervised and semi-supervised based on machine leaning, then summarized the characteristic of all the three methods. Finally, put forward outlook for development tendency.