Decomposing an information stream into the principal components

TL;DR

This paper introduces a novel method for decomposing thematic information streams into principal components using an analogy with Fourier transform and multifractal analysis, demonstrated on Brexit data.

Contribution

It presents a new approach combining Fourier analogy and multifractal analysis to identify and analyze principal topics within information streams.

Findings

Principal components align with related topics from Google Trends.

Multifractal analysis effectively identifies similar topics.

Decomposition reveals thematic structures in Brexit stream.

Abstract

We propose an approach to decomposing a thematic information stream into principal components. Each principal component is related to a narrow topic extracted from the information stream. The essence of the approach arises from analogy with the Fourier transform. We examine methods for analyzing the principal components and propose using multifractal analysis for identifying similar topics. The decomposition technique is applied to the information stream dedicated to Brexit. We provide a comparison between the principal components obtained by applying the decomposition to Brexit stream and the related topics extracted by Google Trends.