| 摘要: |
| 情报分析是情报工作的核心，情报预测是情报分析的难点，是进行决策和行动的重要依据。探讨情报预测的概念内涵与技术发展，对大数据和人工智能等先进技术的科学应用以及未来研究具有重要意义。通过描述性研究、概念分析、资料总结等方法，对情报预测的概念和流程进行了剖析，提炼总结了一种包括预测对象、预测内容的情报预测问题描述框架，并按照理论-方法-工具三个层次对现有的机器定量预测方法体系进行了分析。情报预测包括业务和技术两个维度，清晰明确的问题定义是大数据和人工智能等先进技术发挥作用的重要前提。目前，机器定量预测方法以机器学习、深度学习为主，大大提高了机器解决情报预测问题的能力，未来将向神经符号计算方向拓展。 |
| 关键词: 情报预测 预测问题描述 情报预测方法体系 机器定量预测 神经符号计算 |
| DOI：10.20079/j.issn.1001-893x.220522003 |
|
| 基金项目: |
|
| Intelligence Prediction:Connotation and Technology Development |
| ZHAHG Haiying,DAI Lican,LIU Xin,WANG Chenggang |
| (Southwest China Institute of Electronic Technology,Chengdu 610036,China) |
| Abstract: |
| Intelligence analysis is the core of intelligence work.Intelligence prediction is the difficulty of intelligence analysis and an important basis for decision-making and action.Discussing the concept connotation and technology development of intelligence prediction is of great significance to the scientific application and future research of advanced technologies such as big data and artificial intelligence.Through the methods of descriptive research,concept analysis and data summary,the authors analyzes the concept and process of intelligence prediction,summarizes a description framework of information prediction problems including prediction object and prediction content,and analyzes the existing machine quantitative prediction methodology according to the three levels of theory,method and tool.Intelligence forecasting includes two dimensions,business and technology.A clear definition of business issues is an important prerequisite for the role of advanced technologies such as big data and artificial intelligence.At present,machine quantitative prediction methods mainly focus on machine learning and deep learning,which greatly improves the ability of machines to solve information prediction problems.In the future,it will expand to the direction of neural symbol computing. |
| Key words: intelligence prediction prediction problem description intelligence prediction methodology machine quantitative prediction neural symbol computing |
