Prediction of trending topics using ANFIS and deterministic models

TL;DR

This paper presents a hybrid approach combining deterministic epidemic models with ANFIS to predict trending topics and rumor spread in social networks, addressing data limitations with estimated variables.

Contribution

It introduces a method integrating deterministic epidemic models with ANFIS for better prediction of trending topics, especially when data is limited.

Findings

ANFIS effectively models rumor spreading dynamics.

Deterministic models provide useful input variables for ANFIS.

Hybrid approach improves prediction accuracy for trending topics.

Abstract

Trending topics are often the result of the spreading of information between users of social networks. These special topics can be regarded as rumors. The spreading of a rumor is often studied with the same techniques as in epidemics spreading. It is common that many datasets may not have enough measured variables, so we propose a method for studying the general behavior of the spreaders by selecting estimated variables given by the deterministic model. In order to provide a good approximation, we implemented an adaptative neuro fuzzy inference system (ANFIS). So, in our numerical experimentations, a deterministic approach using SIR and SIRS models (with delay) for two different topics is used. Thus, the authors just applied the ANFIS model for their application and the deterministic model as the preprocessing input variable.