| 摘 要: 目前，大部分内置对话功能的终端只是用户文字语音传输的“中转站”，并不承担计算功能。这导致用于计算的云服务需要承担较高的网络负载和计算负载。随着硬件性能的不断提升，终端设备也能运行部分自然语言处理算法，分担云服务的压力。本文讨论了使用边缘计算技术实现对话机器人终端部署的可行性，并设计了基于云+边缘协同计算的对话系统架构。通过将部分对话机器人部署在终端，可以降低云服务的访问频率，从而降低网络和计算负载。 |
| 关键词: 对话机器人；边缘计算；深度学习模型部署 |
|
中图分类号: TP183
文献标识码: A
|
|
| Terminal Deployment for Edge Computing-driven Chatbot |
|
MA Zhuang1,2, YANG Wei3
|
( 1. Dalian Neusoft University of Information, Dalian 116023, China;
2. Research Institute, Dalian Neusoft Education Technology Group Co. Limited, Dalian 116023, China;
3. Product Development Center, Dalian Neusoft Education Technology Group Co. Limited, Dalian 116023, China)
mazhuang@neuedu.com; yangwei@neuedu.com
|
| Abstract: At present, most information terminals with built-in dialogue function are only transit stations for users' text and voice transmission, and they do not undertake calculation functions. As a result, cloud services for computing have to carry high network load and computing load. With hardware improvement, terminal devices can also locally run some natural language processing algorithms to share pressure of cloud services. This paper discusses feasibility of using edge computing technology to realize deployment of chatbot terminals, and designs a dialogue system architecture based on cloud + edge collaborative computing. By deploying some chatbots on terminals, access to cloud services can be reduced, thereby lowering network and computing load. |
| Keywords: chatbot; edge computing; deep learning model deployment |
