计算型情报分析的进展

Progress in Computational Intelligence Analysis

大数据理念和方法的不断应用,使情报分析在理论和实践上逐渐从定量化分析向计算化分析发展。本文将迄今为止的计算型情报分析的发展历程划分为三个阶段:经验计数分析阶段、精确计量分析阶段、计算型分析阶段,并分别从分析对象、代表成果、主要贡献、主要特征、方法与工具等维度对各阶段进行论述;对情报思维、计算思维、社会计算科学进行辨析,提出了计算型情报分析的主要研究目的和研究内容,并对计算型情报分析的三个主要内容--计算型情报分析模型、计算型情报分析算法、计算型情报分析框架和系统最新的研究进展和研究趋势进行系统的论述和分析。

In the 21st century, with the wide application of network technology and the rise of big data ideas and methods, intelligence analysis has gradually evolved from simple quantitative analysis to computational analysis in theory and practice. Computational intelligence analysis ( CIA) is gradually taking shape. The development process of CIA can be divided into three stages: the empirical counting analysis stage, the accurate measuring analysis stage, and the computational analysis stage. This paper discusses these stages from 5 aspects: the analysis objects, the representative results, the main contributions, the main features, the methods and tools. By summarizing the concept "computation" in the area of computational thinking, social computation and social computing science, this paper puts forward the main contents of CIA which include CIA model, CIA framework and system, and CIA algorithm. The CIA model is a kind of mathematical models, which is used to abstract and simulate the behavior from the specific task-oriented intelligence analysis subject, the specific intelligence object and the specific procedure of the intelligence transmission. With the introduction of big data analysis models, social network models, complex theoretical models and system dynamics models into intelligence analysis and their fruitful achievements, the construction and application of the CIA models have shown the following characteristics and trends in recent years: First, the CIA model is no longer limited to simple descriptive statistics modes, but tends to be a kind of more comprehensive or combined mathematical mode. Secondly, the construction of the CIA model is transformed from a static model to a complicated dynamic model, that is, variables changing over time and their relationships are introduced into the model. Thirdly, the construction and application of CIA models are based on massive data or big data. Fourthly, the model construction's focuses shift from evaluating past achievements to predicting and interpreting future outcomes and scientific discoveries. The CIA framework mainly refers to the technical framework for intelligence analysis, which consists of abstract components and messaging methods for the components, instances. The CIA framework can be considered as a constrained support structure designed to solve the open-ended problem in the process of intelligence analysis. A CIA system is an implementation of the CIA framework. In recent years, the development and research of CIA framework and system have shown the following characteristics and trends: First, the analytical framework and system are evolving from a single architecture to a distributed, cloud computing architecture. Such framework and system are more in line with the requirements of storage, computing, analysis of multi-source, heterogeneous, massive data. Secondly, the CIA framework and system have changed from a traditional relatively closed, dedicated mode to an open, military-civilian integration mode. Thirdly, the intelligence analysis processes in the framework or system can be customized and automated. Fourthly, the CIA framework and system are becoming more intelligent with the application of artificial intelligence technologies. The CIA algorithm is a series of computation steps and instructions formed under specific intelligence analysis tasks or intelligence analysis scenarios by the guidance of intelligence thinking. The CIA algorithm is the prerequisites for automated intelligence analysis so that the intelligence analysis model can be calculated by machine and finally obtains the computational results. The CIA algorithm has recently shown some characteristics and trends: First, the explain ability of intelligence analysis algorithm is being initiated. Secondly, the open source intelligence-oriented perception algorithm is increasingly being valued. Thirdly, the algorithm for the fusion of human analyzed results and computer analyzed results has got more attention. The CIA is an extension of traditional intelligence analysis and now it becomes one of the important directions of future development of information science. The CIA will gradually develop into a new discipline that integrates computational science, social science, cognitive science, complex science, biochemical science, engineering, and many other academic subjects. For researchers, opportunities and challenges will coexist. This requires us to meet the challenges of the future with more innovative, open and bold ideas.