基于支持向量机的文本隐写分析

通过对自然文本统计模型和特性的分析,指出隐藏消息后可能对文本统计特性带来的变化,提出基子支持向量机的通用检测算法。将文本的5个基本统计特征量作为分类特征向量,采用支持向量机对自然文本和载密文本进行有效分类检测。实验结果证明,该算法具有较好的适用性和可靠性。     

搜索引擎PageRank算法的改进

为了解决企业快速决策时信息检索的问题,提出一种改进的PageRank算法。在考虑网页产生时间因素的同时,通过锚文本与网页主题的相似度分析按权重分配网页各正向链接PageRank值,产生的PageRank值更贴合主题搜索引擎的要求,并保持算法的简洁性。实验结果证明该改进算法能有效减少主题漂移现象,恰当提升新网页PageRank值。     

基于汉字拼音声调的文本水印算法

提出一种基于汉字拼音声调的中文文本水印算法。该算法基于统计特征来动态确定嵌入标志代码。在由标志代码确定的水印插入区,通过改变汉字集合声调的特征值来嵌入文本水印。该方法的水印容量由标志代码的数量动态确定,可自主地提高水印容量。整篇文档可以分割成若干个嵌入部分,各部分可单独进行插入、提取计算,降低了计算的复杂性。     

文本分类中一种混合型特征降维方法

提出一种基于特征选择和特征抽取的混合型文本特征降维方法,分析基于选择和抽取的特征降维方法各自的特点,借助特征项的类别分布差异信息对特征集进行初步选择。使用一种新的基于PCA的特征抽取方法对剩余特征集进行二次抽取,在最大限度减少信息损失的前提下实现了文本特征的有效降维。对文本的分类实验结果表明,该特征降维方法具有良好的分类效果。     

一种基于熵的文本相似性计算方法

文本比较是求解两个文本间相似度的过程,文本间的相似度越高代表两个文本越趋于类似。传统的相似度算法主要从字符的角度度量文本的相似性,而忽略了文本内多个共同文本串对于文本相似度的影响。针对此问题,本文提出一种基于熵的相似度求解方法。该方法在对文本间字符信息的提取基础上,建立共同子文本串度量维度,然后采用熵的方法进行相似度度量。实验表明,该方法具有更平滑的相似度曲线,验证了算法的有效性和准确性。     

一种结合上下文语义的短文本聚类算法

短文本因具有特征信息不足且高维稀疏等特点,使得传统文本聚类算法应用于短文本聚类任务时性能有限。针对上述情况,提出一种结合上下文语义的短文本聚类算法。首先借鉴社会网络分析领域的中心性和权威性思想设计了一种结合上下文语义的特征词权重计算方法,在此基础上构建词条-文本矩阵;然后对该矩阵进行奇异值分解,进一步将原始特征词空间映射到低维的潜在语义空间;最后通过改进的K-means聚类算法在低维潜在语义空间完成短文本聚类。实验结果表明,与传统的基于词频及逆向文档频权重的文本聚类算法相比,该算法能有效改善短文本特征不足及高维稀疏性,提高了短文的本聚类效果。     

Unfriend or ignore tweets?: A time series analysis on Japanese Twitter users suffering from information overload

In recent years, with increased opportunities to post content on social media, a number of users are experiencing information overload in relation to social media use. This study addresses how Japanese Twitter users suffering from information overload cope with their stress, focusing on two actions: (1) The “unfriending” activities and (2) The changes in tweet processing methods. Objective data, such as numbers of friends, were collected through Twitter's open Application Programming Interfaces (APIs), and subjective data, such as perceived information overload and tweet processing methods, were collected through a web-based survey as a panel dataset (n = 778). The results demonstrated that although users experience information overload, they continue to increase their number of friends, and that the users who experience information overload modify their usage habits to avoid seeing all received tweets. In short, users do not choose a strategy to reduce the absolute number of received tweets, but only a strategy that involves changing the processing method of the received tweets.     

Prevalence and personality correlates of Facebook bullying among university undergraduates

The purpose of the present study was to examine the prevalence of cyber-bullying through Facebook in a sample of 226 Greek university undergraduates, and to explore whether big five personality characteristics, narcissism, as well as attitudes toward Facebook, technological knowledge and skills were predictive of such behavior. Participants completed a self-report questionnaire measuring the above constructs. Results indicated that almost one third of the sample reported Facebook bullying engagement at least once during the past month, with male students reporting more frequent involvement than females. Bullying through Facebook was predicted by low Agreeableness and more time spent on Facebook only for males, whereas for females none of the studied variables predicted engagement in Facebook bullying. Findings are discussed in terms of prevention and intervention strategies.     

Nonlinear aircraft sensor fault reconstruction in the presence of disturbances validated by real flight data

This paper proposes an approach for Inertial Measurement Unit sensor fault reconstruction by exploiting a ground speed-based kinematic model of the aircraft flying in a rotating earth reference system. Two strategies for the validation of sensor fault reconstruction are presented: closed-loop validation and open-loop validation. Both strategies use the same kinematic model and a newly-developed Adaptive Two-Stage Extended Kalman Filter to estimate the states and faults of the aircraft. Simulation results demonstrate the effectiveness of the proposed approach compared to an approach using an airspeed-based kinematic model. Furthermore, the major contribution is that the proposed approach is validated using real flight test data including the presence of external disturbances such as turbulence. Three flight scenarios are selected to test the performance of the proposed approach. It is shown that the proposed approach is robust to model uncertainties, unmodeled dynamics and disturbances such as time-varying wind and turbulence. Therefore, the proposed approach can be incorporated into aircraft Fault Detection and Isolation systems to enhance the performance of the aircraft.     

基于词向量模型的情感分析

随着移动互联网的发展,以商品评论等带有主观性的短文本信息急剧增加.海量的文本信息使得人工管理越来越困难.本文以商品评论为研究对象进行情感分析.针对商品评论为短文本的特点,本文在词向量的基础上提出了词向量叠加方法和加权词向量方法进行文本特征的提取,从而更深层次的提取短文本特征.在进行评论情感分析模型性能的比较中,说明了本文所提方法的有效性.基于情感分析技术可以解决人工难以胜任的海量商品评论的分类,方便用户快速获取有效信息.