Quality in simultaneous interpreting is a frequently discussed concept. In the enterprise setting, earnings conference call remains a rarely explored field. This thesis offers a descriptive study on assessing interpre...Quality in simultaneous interpreting is a frequently discussed concept. In the enterprise setting, earnings conference call remains a rarely explored field. This thesis offers a descriptive study on assessing interpreting quality from perspectives of fidelity, fluency, and appropriacy. As the corpus, Tencent 2022 Third Quarter Result Announcement provides an ideal transcript to the author to conduct its analysis. Interpreting is frequently done without bearing in mind the multitude of factors that can affect the quality of interpreting. Drawing a conclusion that the interpreter does make a lot of omissions, pauses and hesitations posing a negative effect on the fidelity, fluency and accuracy of the interpreting, the present author suggests that more preparation should be done for improving performance, such as terminologies, company background information, a reasonable speech rate, good image and acoustic quality, and so on.展开更多
IP的语音传输(Voice over Internet Protocol,VoIP)技术是一种基于网络的语音技术,它可以将语音信号转换成数据流,并在互联网上传输。VoIP技术与Call Center的融合使用户在通话时能更加方便、快捷地使用互联网,节省时间和费用,实现语音...IP的语音传输(Voice over Internet Protocol,VoIP)技术是一种基于网络的语音技术,它可以将语音信号转换成数据流,并在互联网上传输。VoIP技术与Call Center的融合使用户在通话时能更加方便、快捷地使用互联网,节省时间和费用,实现语音信号与数据流的同时传输,提高通话质量和效率。同时,技术融合可以解决传统呼叫中心面临的一些挑战,如服务质量下降、成本上升等问题。展开更多
微服务架构因具有良好的可扩展性和可维护性越来越受到云应用软件的青睐.与此同时,微服务之间复杂的交互使得系统的性能异常检测变得更加困难.现有的微服务性能异常检测方法均不能很好地建立跨不同调用路径的微服务及其对应的响应时间...微服务架构因具有良好的可扩展性和可维护性越来越受到云应用软件的青睐.与此同时,微服务之间复杂的交互使得系统的性能异常检测变得更加困难.现有的微服务性能异常检测方法均不能很好地建立跨不同调用路径的微服务及其对应的响应时间之间的复杂关系,导致异常检测准确率不高、根因定位不准确.提出了一种基于Transformer的微服务性能异常检测与根因定位方法TTEDA(Transformer trace explore data analysis).首先将调用链构建为微服务调用序列和对应的响应时间序列,然后借助自注意力机制捕捉微服务之间的调用关系,并通过编码器-解码器建立微服务的响应时间与其调用路径之间的关联关系,从而获得微服务在不同的调用链上的正常响应时间分布.基于学习到的正常模式判断调用链的异常,并可将异常精确到微服务级别.进一步地,利用微服务之间的调用关系以及异常的传播方式,对出现性能异常的微服务进行反向拓扑排序,实现了准确快速的根因定位.在开源基准微服务系统Train-Ticket的数据集和AIops挑战赛数据集评估了TTEDA的有效性,相比于同类异常检测方法AEVB,Multi-LSTM,TraceAnomaly,精确率平均提高了48.6%,30.2%,3.5%,召回率平均提高了34.7%,1.1%,4.1%.相比于根因定位算法MonitorRank和TraceAnomaly,根因定位的准确率分别提高了35.4个百分点和6.1个百分点.展开更多
文摘Quality in simultaneous interpreting is a frequently discussed concept. In the enterprise setting, earnings conference call remains a rarely explored field. This thesis offers a descriptive study on assessing interpreting quality from perspectives of fidelity, fluency, and appropriacy. As the corpus, Tencent 2022 Third Quarter Result Announcement provides an ideal transcript to the author to conduct its analysis. Interpreting is frequently done without bearing in mind the multitude of factors that can affect the quality of interpreting. Drawing a conclusion that the interpreter does make a lot of omissions, pauses and hesitations posing a negative effect on the fidelity, fluency and accuracy of the interpreting, the present author suggests that more preparation should be done for improving performance, such as terminologies, company background information, a reasonable speech rate, good image and acoustic quality, and so on.
文摘IP的语音传输(Voice over Internet Protocol,VoIP)技术是一种基于网络的语音技术,它可以将语音信号转换成数据流,并在互联网上传输。VoIP技术与Call Center的融合使用户在通话时能更加方便、快捷地使用互联网,节省时间和费用,实现语音信号与数据流的同时传输,提高通话质量和效率。同时,技术融合可以解决传统呼叫中心面临的一些挑战,如服务质量下降、成本上升等问题。
文摘微服务架构因具有良好的可扩展性和可维护性越来越受到云应用软件的青睐.与此同时,微服务之间复杂的交互使得系统的性能异常检测变得更加困难.现有的微服务性能异常检测方法均不能很好地建立跨不同调用路径的微服务及其对应的响应时间之间的复杂关系,导致异常检测准确率不高、根因定位不准确.提出了一种基于Transformer的微服务性能异常检测与根因定位方法TTEDA(Transformer trace explore data analysis).首先将调用链构建为微服务调用序列和对应的响应时间序列,然后借助自注意力机制捕捉微服务之间的调用关系,并通过编码器-解码器建立微服务的响应时间与其调用路径之间的关联关系,从而获得微服务在不同的调用链上的正常响应时间分布.基于学习到的正常模式判断调用链的异常,并可将异常精确到微服务级别.进一步地,利用微服务之间的调用关系以及异常的传播方式,对出现性能异常的微服务进行反向拓扑排序,实现了准确快速的根因定位.在开源基准微服务系统Train-Ticket的数据集和AIops挑战赛数据集评估了TTEDA的有效性,相比于同类异常检测方法AEVB,Multi-LSTM,TraceAnomaly,精确率平均提高了48.6%,30.2%,3.5%,召回率平均提高了34.7%,1.1%,4.1%.相比于根因定位算法MonitorRank和TraceAnomaly,根因定位的准确率分别提高了35.4个百分点和6.1个百分点.