基于CCA子空间和GMM的自动图像标注

为了避免图像分割,并提高图像标注精度,提出一种基于典型相关分析(CCA)和高斯混合模型(GMM)的自动图像标注方法.利用CCA对图像的全局颜色特征与全局局部二值模式(LBP)纹理特征进行特征融合.使用融合后的语义特征,对每一个关键词建立GMM模型来估计单词类密度,从而在特征子空间中得到每个单词的概率分布.采用贝叶斯分类器确定每个标注词和测试图像的联合概率,运用词间语义关系优化标注结果.实验结果表明,使用该方法后的图像标注性能有了较大程度的改善.     

基于语义学习的图像多模态检索

针对语义鸿沟问题,在语义学习的基础上设计图像的多模态检索系统。该系统结合3种查询方式进行图像检索。基于视觉特征的查询通过特征提取与相似度匹配进行排位。基于标签的查询建立在图像自动标注的基础上,但在语义空间之外的泛化能力较差。基于语义图例的查询能够在很大程度上克服这个缺陷,通过在显式或隐式的语义空间上进行查询,使检索结果更符合人类感知。实验结果表明,与基于纹理特征的图像检索相比,基于语义图例的检索具有更高的精度及召回率。     

从Web获取部分整体关系语料的方法

部分整体关系获取是知识获取中的重要组成部分。Web逐步成为知识获取的重要资源之一。搜索引擎是从Web中获取部分整体关系知识的有效手段之一,我们将Web中包含部分整体关系的检索结果集合称为部分整体关系语料。由于目前主流搜索引擎尚不支持语义搜索,如何构造有效的查询以得到富含部分整体关系的语料,从而进一步获取部分整体关系,就成为一个重要的问题。该文提出了一种新的查询构造方法,目的在于从Web中获取部分整体关系语料。该方法能够构造基于语境词的查询,进而利用现有的搜索引擎从Web中获取部分整体关系语料。该方法在两个方面与人工构造查询方法和基于语料库查询构造查询方法所获取的语料进行对比,其一是语料中含有部分整体关系的语句数量;二是从语料中进一步获取部分整体关系的难易程度。实验结果表明,该方法远远优于后两者。     

The Construction of Beauty: A Cross-Cultural Analysis of Women's Magazine Advertising

As a media genre, advertising offers a unique opportunity to study how the beauty ideal is constructed across cultures. This research analyzes the content of advertisements from women's fashion and beauty magazines in Singapore, Taiwan, and the U.S. to compare how beauty is encoded and found a noticeable difference between the portrayals of women from the U.S. and from the two East Asian societies in terms of sexual portrayal. In addition, Asian ads contained a large proportion of cosmetics and facial beauty products whereas the U.S. ads were dominated by clothing. These findings suggest that beauty in the U.S. may be constructed more in terms of "the body," whereas in Singapore and Taiwan the defining factor is more related to a pretty face. The article also discusses how feminist critiques of the sexual objectification of women in advertising may need to be considered within their historical, Western context of origin.     

面向国防科技领域的技术和术语语料库构建方法

互联网存在海量的文献和科技信息,隐含着大量高价值情报。识别国防科技领域中的技术和术语可以为构建国防科技知识图谱奠定基础。该文基于此领域的海量军事文本,以维基百科中军事领域的新技术为基点采集语料,涵盖了新闻、文献和维基百科三种体裁。在分析军事技术文本特点的基础上制定了一系列标注规范,开展了大规模语料的标注工作,构建了一个面向国防科技领域的技术和术语语料库。该语料库共标注了479篇文章,包含24 487个句子和33 756个技术和术语。同时,该文探讨了模型预标注策略的可行性,并对技术和术语类别在不同体裁上的分布以及语料标注的一致性进行了统计分析。基于该语料库的实验表明,技术和术语识别性能F₁值达到70.40%,为进一步的技术和术语识别研究提供了基础。     

Representation of Women in News and Photos: Comparing Content to Perceptions

This study uses a feminist framework of masculine cultural hegemony to examine the representation of women in two newspapers—a medium-sized newspaper (Study 1) and a larger newspaper (Study 2). Surveys gauged news staff and news reader perceptions of female representation in news content to determine if perceptions matched content-analysis findings. Content-analysis results revealed a greater proportion of males than females in both content and photos for both studies. Female staffers were more likely than their male counterparts to perceive this disparity, and news readers perceived an even greater disparity than the news staff. News staffs were less likely to perceive disparities in the editorial, travel, and entertainment sections. We discuss the findings in relation to cultural hegemony and the representation of women in the news.     

Language supports for journal abstract writing across disciplines

Various writing assistance tools have been developed through efforts in the areas of natural language processing with different degrees of success of curriculum integration depending on their functional rigor and pedagogical designs. In this paper, we developed a system, WriteAhead, that provides six types of suggestions when non‐native graduate students of English from different disciplines are composing journal abstracts, and assessed its effectiveness. The method involved automatically building domain‐specific corpora of abstracts from the Web via domain names and related keywords as query expansions, and automatically extracting vocabulary and n‐grams from the corpora in order to offer writing suggestions. At runtime, learners' input in the writing area of the system actively triggered a set of corresponding writing suggestions. This abstract writing assistant system facilitates interactions between learners and the system for writing abstracts in an effective and contextualized way, by providing suggestions such as collocations or transitional words. For assessment of WriteAhead, we compared the writing performance of two groups of students with or without using the system, and adopted student perception data. Findings show that the experiment group wrote better, and most students were satisfied with the system concerning most suggestion types, as they can effectively compose quality abstracts through provided language supports from WriteAhead.     

基于SRNN神经网络的汉语文本词类标注方法

词类标注是语料库加工流程一个关键环节，是句法，语义标注分析的前提，本文提出了一种基于ＳＲＮＮ神经网络的词类标记方法，ＳＲＮＮ在三层前向神经网络的结构基础上，增加了隐层节点与输入层状态节点之间的反馈联接，这种结构使用网络具有利用上下文词类信息的处理能力，本文还讨论了网络的训练算法，以人工标注的句子作训练集，经过训练收敛后的网络对新语料的词类标注正确率达到了９４％。