您的位置:首页  > 论文页面

基于堆叠降噪自动编码器的评价对象抽取

发表时间:2016-08-31  浏览量:2260  下载量:910
全部作者: 李娟,高志强
作者单位: 东南大学计算机科学与工程学院
摘 要: 提出基于堆叠降噪自动编码器(stacked denoising auto-encoders,SDAE)的模型来抽取评价对象,仅利用词向量作为输入特征。原始的SDAE 是用于分类任务的模型,本研究将评价对象抽取任务看作序列标记任务。利用贪婪算法对SDAE 模型进行改进,使得原本用于分类问题的模型适用于序列标记任务。实验表明,本研究提出的模型的抽取效果接近SemEval-2014 受限模型(模型训练仅使用提供的数据集)中性能最好的模型。
关 键 词: 自然语言处理;评价对象抽取;堆叠降噪自动编码器;观点挖掘
Title: Opinion target extraction using stacked denoising auto-encoders
Author: LI Juan, GAO Zhiqiang
Organization: School of Computer Science and Engineering, Southeast University, Nanjing 211189, China
Abstract: We propose a new model based on stacked denoising auto-encoders (SDAE), which only uses word embeddings as input features. The original SDAE model is a classification model, however, we regard the opinion target extraction task as a sequential labeling task in this paper. Thus, in order to adapt the original classification model to a sequential labeling model, we propose to use a greedy algorithm to extend the SDAE model. Experimental results demonstrate that the proposed model rival the top performing constrained model which only uses the provided training data in SemEval-2014.
Key words: natural language processing; opinion target extraction; stacked denoising auto-encoders; opinion mining
发表期数: 2016年8月第16期
引用格式: 李娟,高志强. 基于堆叠降噪自动编码器的评价对象抽取[J]. 中国科技论文在线精品论文,2016,9(16):1617-1626.
 
0 评论数 0
暂无评论
友情链接