基于层叠条件随机场的CFN自动标注研究  

Research of CFN Automatic Labeling based on Cascaded Conditional Random Fields

作　　者：陈雪艳[1] 吕国英[1] 李茹[1] 何鹏[2] 

机构地区：[1]山西大学计算机与信息技术学院,太原030006 [2]太原理工大学计算机与软件学院,太原030024

出　　处：《电脑开发与应用》2009年第3期1-3,共3页Computer Development & Applications

基　　金：国家863高技术研究发展计划(2006AA0lZ142)资助

摘　　要：提出了一种基于层叠条件随机场的CFN自动标注方法。该方法在低层条件随机场模型中解决了框架元素的识别,将识别结果传递到上层短语类型识别的条件随机场模型,再将识别结果传递到上层句法功能识别的条件随机场模型,其低层模型为上层模型提供决策支持。实验选用CFN中"陈述"框架下的句子库,实现了基于层叠条件随机场CFN自动标注的原型系统。This paper presents an approach of CFN automatic tagging based on cascaded conditional random fields model. The frame elements are first recognized by the lower random fields model. The result then is passed to the upper phrase type identification with the random fields mode, and then transfer the results to identify the upper syntactic function identification with the random fields mode. The lower model supports the decision for high model. The experiment selects the sentence of ＂ Statement＂ frame in CFN, has realized CFN automatic tagging based on cascaded conditional random fields model.

关 键 词：层叠条件随机场 CFN 框架元素 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]