Trend Detection Using NLP as a Mechanism of Decision Support

Trend Detection Using NLP as a Mechanism of Decision Support

The purpose of this article is to present the principles of a developed algorithm for identifying trends based on the analysis of big text data and presenting the result in formats that are convenient for decision makers to be implemented in the iFORA Big Data Mining System. The paper provides an ov...

Full description

Saved in:
Bibliographic Details
Published in:Scientific and technical information processing Vol. 50; no. 5; pp. 440 - 448
Main Authors: Lobanova, P. A., Kuzminov, I. F., Karatetskaia, E. Yu, Sabidaeva, E. A., Anpilogov, V. V.
Format: Journal Article
Language:English
Published: Moscow Pleiades Publishing 01-12-2023
Springer Nature B.V
Subjects:
Algorithms
Big Data
Clustering
Computer Science
Computer Systems Organization and Communication Networks
Data mining
Decision analysis
Decision support systems
Trends
iFORA
NLP
decision making
text analytics
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The purpose of this article is to present the principles of a developed algorithm for identifying trends based on the analysis of big text data and presenting the result in formats that are convenient for decision makers to be implemented in the iFORA Big Data Mining System. The paper provides an overview of existing text analytics algorithms; outlines the mathematical basis for identifying terms that mean trends, which is proposed and tested for dozens of implemented projects; describes approaches to clustering terms based on their vectors in the Word2vec space; and provides examples of two key visualizations (semantic, trend maps) that outline the range of topics and trends that characterize a particular area of study, as a way to adapt the results of the analysis to the tasks of decision makers. The limitations and advantages of using the proposed approach for decision support are discussed, and directions for future research are suggested.
ISSN:0147-6882
1934-8118
DOI:10.3103/S0147688223050106