2017acl---Learning Cognitive Features from Gaze Data for Sentiment and Sarcasm Classification笔记

文章标题：

Learning Cognitive Features from Gaze Data for Sentiment and Sarcasm
Classification using Convolutional Neural Network

摘要

认知NLP系统是利用行为数据抽取的从眼动模式，脑图认知特征来增强传统的基于文档特征。

实验证明了，用自动学习的眼动数据特征和文本特征的组合，分类性能优于（1）基于单独依赖文本特征输入的CNN和（2）现有的基于手工制作的眼动和文本特征的系统。

介绍

就是说之前的系统难以处理微妙的情感分类，比如说“我真是非常爱我的工作，每天泡在工作里40个小时”其实应该是消极的情绪，但在缺乏外部信息的情况下现有的系统很难分类正确，但2016年的两篇文章介绍了利用读者的快速眼动信息来提升这些微妙的情感分类处理的性能。现有工作很少有利用眼动信息的，并且眼动信息依赖于手工提取特征。

现有的工作主要局限：a.可能不能完全的充分的提取眼动信息特征。

B. 眼动信息不能像文本信息那样直接提取特征，也给现有工作造成了困难。

因此, 在这项工作中, 我们尝试用卷积神经网络 (CNN) 从眼动和文本数据中学习特征表示, 而不是用手工制作基于眼动和文本的特征。

数据集是在现有的公用数据集上加上眼动数据

数据集：

The datasets used in our experiments, resources and other relevant pointers are available at http://www.cfilt.iitb.ac.in/ cognitive-nlp

We conduct experiments for two binary classification tasks of sentiment and sarcasm using two publicly available datasets enriched with eye-movement information. Dataset 1 has been released by Mishra et al. (2016a). It contains 994 text snippets with 383 positive and 611 negative examples. Out of the 994 snippets, 350 are sarcastic. Dataset 2 has been used by Joshi et al. (2014) and it consists of 843 snippets comprising movie reviews and normalized tweets out of which 443 are positive, and 400 are negative.
Eye-movement data of 7 and 5 readers is available for each snippet for dataset 1 and 2 respectively

2017acl---Learning Cognitive Features from Gaze Data for Sentiment and Sarcasm Classification笔记相关推荐

UFLDL教程: Exercise:Learning color features with Sparse Autoencoders
Linear Decoders Deep Learning and Unsupervised Feature Learning Tutorial Solutions 以三层的稀疏编码神经网络而言,在s ...
【DeepLearning】Exercise:Learning color features with Sparse Autoencoders
Exercise:Learning color features with Sparse Autoencoders 习题链接:Exercise:Learning color features with ...
视频分类论文阅读笔记——Learning Spatiotemporal Features With 3D Convolutional Networks
论文:Learning Spatiotemporal Features With 3D Convolutional Networks 作者:FaceBook AI研究院来源:ICCV2015 代码: ...
论文阅读《Meta-FDMixup：Cross-Domain Few-Shot Learning Guided by Labeled Target Data》
Background & Motivation 之前看的小样本论文大部分是目标域和源域属于同一个域,比如 COCO 数据集里的小样本设定:60类为 Base,20类为 Novel.Base 和 ...
图像去噪--Noise2Noise: Learning Image Restoration without Clean Data
Noise2Noise: Learning Image Restoration without Clean Data ICML 2018 1 Introduction 基于 corrupted or ...
时空特征--Learning Spatiotemporal Features with 3D Convolutional Networks
Learning Spatiotemporal Features with 3D Convolutional Networks ICCV 2015 http://vlg.cs.dartmouth.ed ...
【论文笔记】Learning Enriched Features for Real Image Restoration and Enhancement
Learning Enriched Features for Real Image Restoration and Enhancement Abstract Contribution Method(M ...
《Learning Enriched Features for Real Image Restoration and Enhancement》
一.论文 <Learning Enriched Features for Real Image Restoration and Enhancement> 摘要:以从降级版本中恢复高质量图像 ...
读论文：Noise2Noise: Learning Image Restoration without Clean Data
很有意思的文章,我之前做过一些深度学习的降噪任务,针对不同光照条件下的图像进行降噪.这个时候就会涉及到标签样本的制作. 在监控领域中,相差6个db的照度,图像的噪点就会有明显的差异性.所以我们在训练模 ...

2017acl---Learning Cognitive Features from Gaze Data for Sentiment and Sarcasm Classification笔记

2017acl---Learning Cognitive Features from Gaze Data for Sentiment and Sarcasm Classification笔记相关推荐

最新文章

热门文章