期刊文献+
共找到3篇文章
< 1 >
每页显示 20 50 100
融合时间流特征和传播结构特征的谣言检测
1
作者 董苏军 钱忠 +1 位作者 李培峰 朱巧明 《中文信息学报》 CSCD 北大核心 2024年第9期167-176,共10页
现存关于谣言检测的研究方法要么只关注谣言在社交媒体上传播的时间流特征,要么仅关注传播结构特征,并且使用了大量的辅助信息。实际上,谣言传播的时间流和传播结构特征均有助于提升谣言检测模型的性能,并且能够形成互补作用。与此同时... 现存关于谣言检测的研究方法要么只关注谣言在社交媒体上传播的时间流特征,要么仅关注传播结构特征,并且使用了大量的辅助信息。实际上,谣言传播的时间流和传播结构特征均有助于提升谣言检测模型的性能,并且能够形成互补作用。与此同时,源用户的自我描述相比于其他辅助信息更为重要,并且源推文的语义信息在整个会话线程中起到了关键作用。为解决上述问题,该文提出了一个新颖的谣言检测模型TPSS。该模型融合了时间流和传播结构特征。同时,仅采用源用户的自我描述作为辅助信息,并且提出了一种协同注意力机制来增强源推文的作用。该机制基于源推文特征来增强时间流特征和传播结构特征。在Twitter15、Twitter16和PHEME数据集上的实验结果表明TPSS优于基准系统。 展开更多
关键词 谣言检测 时间流特征 传播结构 增强机制
下载PDF
基于混合方法的IPSec VPN加密流量识别 被引量:10
2
作者 周益旻 刘方正 王勇 《计算机科学》 CSCD 北大核心 2021年第4期295-302,共8页
文中提出了一种混合方法,将指纹识别与机器学习方法相结合,实现了IPSec VPN加密流量的识别。该方法首先基于负载特征从网络流量中筛选出IPSec VPN流量;接着,基于时间相关的流特征,利用随机森林算法建立了IPSec VPN流量分类模型,通过参... 文中提出了一种混合方法,将指纹识别与机器学习方法相结合,实现了IPSec VPN加密流量的识别。该方法首先基于负载特征从网络流量中筛选出IPSec VPN流量;接着,基于时间相关的流特征,利用随机森林算法建立了IPSec VPN流量分类模型,通过参数优化以及特征选择,整体流量识别的准确率达到了93%。实验结果验证了通过流特征提取的机器学习方法识别IPSec VPN流量的可行性;同时表明了该方法能够有效均衡识别精度与识别速度,达到了高效识别IPSec VPN加密流量的效果。 展开更多
关键词 IPSec VPN 加密量识别 随机森林 时间相关特征 参数优化
下载PDF
Epidemiology of traffic crash mortality in west of Iran in a 9 year period
3
作者 Behrooz Hamzeh Farid Najafi +3 位作者 Behzad Karamimatin Tuoraj Ahmadijouybari Aresh Salari Mehdi Moradinazar 《Chinese Journal of Traumatology》 CAS CSCD 2016年第2期70-74,共5页
Purpose: In Iran, the most common cause of injuries and the second leading cause of deaths are traffic accidents, and those problems impose a substantial financial burden on the society. This study aims to determine ... Purpose: In Iran, the most common cause of injuries and the second leading cause of deaths are traffic accidents, and those problems impose a substantial financial burden on the society. This study aims to determine traffic accident mortality trends and their epidemiologic characteristics in the Kermanshah province, west of lran. Methods: In a cross sectional study, road traffic fatality data from 2004 to 2013 were analyzed to determine the epidemiological pattern of traffic accident mortality. Trend assessment was performed to ascertain the decreasing or increasing status. Chi-square and one-way analysis of variance (ANOVA) tests, as well as Poisson regression were used to determine the significance of the data in time. Data were analyzed using Excel and statistical package of SPSS version 19. Results: Out of 5110 people that died in traffic accidents, 4024 (78.7%) were males. The state of accidents indicated that 404 (43.8%) female pedestrians died as a result of car crashes, and 1330 (41.4%) males died because of car collisions. 1554 (31.9%) deaths happened to pedestrians and 1556 (32.1%) to vehicle drivers, and the rest belonged to vehicle passengers. Head trauma was the cause of death for as much as 3400 (69.9%) cases. Fatal crashes in which pedestrians were involved mostly occurred between the hours 13:00 to 15:00, while the time for vehicle drivers was between 16:00 to 18:00. 2882 people (59.1%) died before reaching to health care facilities. Traffic crash mortality trend for pedestrians follows a linear pattern with a gentle downward slope, but the trend shows various swings when it comes to vehicle drivers. Conclusion: The number of traffic crash deaths from 2004 to 2013 indicates a decreasing trend in two groups of road users: vehicle drivers and car occupants. This can be due to some interventions such as modification of traffic rules and enhancement of police control which has been implemented in recent years. Moreover, more attention should be paid to promote the optimal health care services to save the lives of the injured from traffic accidents. 展开更多
关键词 Accidents trafficMotor vehiclesMortality[ran
原文传递
上一页 1 下一页 到第
使用帮助 返回顶部