Social Structure Construction from the Forums Using Interaction Coherence

Kazuaki Kashihara, Jana Shakarian, Chitta Baral

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Scopus citations


Social network analysis is one of the important tasks for cybersecurity field to identify some potentially important members in the hackers’ forums and communities. To create the social network of the forums and communities, we need to understand the social structure of them. Once the metadata from each forum is collected, the social structure is constructed based on the users’ interactions such as who replies to whose post and when the response is posted. However, most of the hackers’ forums are unstructured, and there is no way to find the interaction between the users. We introduce the Next Paragraph Prediction method that returns true if a response is a direct response of the previous post. This method helps to create the social structure from an unstructured forum. We evaluated our method with 10 different topics’ threads about the cybersecurity related field, and the method shows above accuracy on average with the largest training dataset.

Original languageEnglish (US)
Title of host publicationProceedings of the Future Technologies Conference, FTC 2020, Volume 3
EditorsKohei Arai, Supriya Kapoor, Rahul Bhatia
PublisherSpringer Science and Business Media Deutschland GmbH
Number of pages14
ISBN (Print)9783030630911
StatePublished - 2021
EventFuture Technologies Conference, FTC 2020 - San Francisco, United States
Duration: Nov 5 2020Nov 6 2020

Publication series

NameAdvances in Intelligent Systems and Computing
ISSN (Print)2194-5357
ISSN (Electronic)2194-5365


ConferenceFuture Technologies Conference, FTC 2020
Country/TerritoryUnited States
CitySan Francisco


  • BERT
  • Cybersecurity
  • Nest paragraph prediction
  • Social structure

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Computer Science(all)


Dive into the research topics of 'Social Structure Construction from the Forums Using Interaction Coherence'. Together they form a unique fingerprint.

Cite this