Early estimation of protest time spans: A novel approach using topic modeling and decision trees

Satyakama Paul, Madhur Hasija, Ravi Vishwanath Mangipudi, Tshilidzi Marwala

Research output: Chapter in Book/Report/Conference proceedingChapterpeer-review

Abstract

Protests and agitations have long been used as means for showing dissident toward social, political, and economic issues in civil societies. In recent years, we have witnessed a large number of protests across various geographies. Not to be left behind by similar trends in the rest of the world, South Africa in recent years has witnessed a large number of protests. This paper uses the English text description of the protests to predict their time spans/durations. The descriptions consist of multiple causes of the protests, courses of actions, etc. Next, we use unsupervised (topic modeling) and supervised learning (decision trees) to predict the duration of protests. The results are very promising and close to 90% of accuracy in early prediction of the duration of protests. We expect the work to help public services departments to better plan and manage their resources while handling protests in future.

Original languageEnglish
Title of host publicationAdvances in Intelligent Systems and Computing
PublisherSpringer Verlag
Pages107-116
Number of pages10
DOIs
Publication statusPublished - 2018

Publication series

NameAdvances in Intelligent Systems and Computing
Volume758
ISSN (Print)2194-5357

Keywords

  • Decision trees
  • Duration
  • Early prediction
  • Latent Dirichlet allocation
  • Perplexity
  • Protests and agitations
  • South Africa
  • Topic modeling

ASJC Scopus subject areas

  • Control and Systems Engineering
  • General Computer Science

Fingerprint

Dive into the research topics of 'Early estimation of protest time spans: A novel approach using topic modeling and decision trees'. Together they form a unique fingerprint.

Cite this