| 引用本文： |
-
崔海燕,李雅文,徐欣.基于时间卷积网络的科技需求主题热度预测算法[J].广西科学,2022,29(4):627-633. [点击复制]
- CUI Haiyan,LI Yawen,XU Xin.Algorithm of Subject Heat of Science and Technology Demand Prediction Based on Time Convolution Network[J].Guangxi Sciences,2022,29(4):627-633. [点击复制]
|
|
| 摘要: |
| 得益于深度学习的快速发展，大数据分析技术不仅在自然语言处理领域应用广泛，在数值预测领域也更加成熟。为了提高科技需求数据主题热度预测的准确率，本文提出一种基于时间卷积网络（Time Convolution Network，TCN）的科技需求主题热度预测方法（Subject Heat of Science and Technology Demand Prediction Based on Time Convolution Network，SHDP-TCN），该方法融入科技需求的主题特征，并基于TCN及自注意力机制进行时序预测。实验结果表明，在真实的科技需求数据集上，本算法对科技需求主题热度的预测准确率优于自回归积分滑动平均（Auto Regressive Integrated Moving Average，ARIMA）、长短时记忆（Long Short-Term Memory，LSTM）网络、卷积神经网络（Convolutional Neural Networks，CNN）和TCN等算法。 |
| 关键词: 科技需求数据 TCN网络 自注意力机制 科技需求主题热度预测 残差块 |
| DOI：10.13656/j.cnki.gxkx.20220919.002 |
| 投稿时间：2022-04-25 |
| 基金项目:国家重点研发计划项目（2018YFB1402600）和国家自然科学基金项目（61772083，61877006，61802028，62002027）资助。 |
|
| Algorithm of Subject Heat of Science and Technology Demand Prediction Based on Time Convolution Network |
|
CUI Haiyan1, LI Yawen2, XU Xin1
|
| (1.Beijing Key Laboratory of Intelligent Telecommunication Software and Multimedia, School of Computer Science, Beijing University of Posts and Telecommunications, Beijing, 100082, China;2.School of Economics and Management, Beijing University of Posts and Telecommunications, Beijing, 100082, China) |
| Abstract: |
| Thanks to the rapid development of deep learning, big data analysis technology is not only widely used in the field of natural language processing, but also more mature in the field of numerical prediction. In order to improve the accuracy of subject heat of science and technology demand data prediction, this article proposes a method of Subject Heat of Science and Technology Demand Prediction Based on Time Convolution Network (SHDP-TCN). This method integrates the suject characteristics of science and technology demand, and makes temporal prediction based on TCN and self-attention mechanism. Experimental results show that on the real science and technology demand data set, the prediction accuracy of this algorithm is better than that of Auto Regressive Integrated Moving Average (ARIMA), Long Short-Term Memory (LSTM) network, Convolutional Neural Networks (CNN) and TCN algorithms. |
| Key words: science and technology demand data TCN network self-attention mechanism subject heat of science and technology demand prediction residual block |
