A cognitive approach to improving the quality of the customs authorities' analytic activities

2020 ◽  
Vol 19 (3) ◽  
pp. 416-429
Author(s):  
V.V. Makrusev ◽  
A.A. Sobol'

Subject. The article considers prospects for enhancing the quality of analytical activities of the Customs authorities through a cognitive approach implementation. Objectives. The aim is to formulate promising areas for improving the quality of analytical work of the Customs authorities by using a cognitive approach, to develop a concept for managing the analytical activities based on knowledge. Methods. The study rests on systems methodology and institutional theory. It also employs cognitive modeling techniques. Results. We show the process of transferring disparate data into knowledge, consider basic methods of big data processing, and identify the most acceptable method of customs data analysis. The paper discloses the contents and elements of the cognitive approach in analytical activities of on-line monitoring centers and describes an experiment with the application of data mining technology on the basis of the Federal Customs Service of Russia. We recommend the said approach to analytical and ICT units of organizations operating in the field of customs services. Conclusions. Current trends in software development, the use of electronic forms of customs documents, and continuously expanded list of analytical tools for big data processing entail the need for changing traditional approaches to information analysis to assess customs risks. The expert method should be supplemented with new, previously unused decision support tools, such as tools that enable automated big data analysis.

The real challenge for data miners lies in extracting useful information from huge datasets. Moreover, choosing an efficient algorithm to analyze and process these unstructured data is itself a challenge. Cluster analysis is an unsupervised practice to attain data insight in the era of Big Data. Hyperflated PIC is a Big Data processing solution designed to take advantage over clustering. It is a scalable efficient algorithm to address the shortcomings of existing clustering algorithm and it can process huge datasets quickly. HPIC algorithms have been validated by experimenting them with synthetic and real datasets using different evaluation measure. The quality of clustering results has also been analyzed and proved to be highly efficient and suitable for Big Data processing.


2019 ◽  
Vol 12 (1) ◽  
pp. 42 ◽  
Author(s):  
Andrey I. Vlasov ◽  
Konstantin A. Muraviev ◽  
Alexandra A. Prudius ◽  
Demid A. Uzenkov

Sign in / Sign up

Export Citation Format

Share Document