LSR-for-CTI-classification

This repository supports the paper "LSR: Latent Space Refinement for Cyber Threat Intelligence Classification", which proposes an unsupervised text classification method to identify and organize Cyber Threat Intelligence (CTI) text from open-source data.

📊 Datasets

Three datasets are used in the study:

1. TD Dataset

self-collected and included in this repository

2. BC Dataset

• From work Cyberthreat detection from twitter using deep neural networks
• Dionísio, N., Alves, F., Ferreira, P. M., & Bessani, A. (2019, July). Cyberthreat detection from twitter using deep neural networks. In 2019 international joint conference on neural networks (IJCNN) (pp. 1-8). IEEE.

3. STC Dataset

• From work Corpus and deep learning classifier for collection of cyber threat indicators in twitter stream
• Behzadan, V., Aguirre, C., Bose, A., & Hsu, W. (2018, December). Corpus and deep learning classifier for collection of cyber threat indicators in twitter stream. In 2018 IEEE International Conference on Big Data (Big Data) (pp. 5002-5007). IEEE.

What's included in the TD Dataset (Twitter Data)

• The TD dataset consists of tweets grouped into five topics[music, sports, cyber security, movies, health].
• These tweets were collected in 2023 using the Twitter API.
• We provide tweet IDs only, in accordance with Twitter's Developer Policy, which allows the sharing of up to 50,000 tweet IDs for non-commercial academic research purposes.
• To access the full tweet content, users must rehydrate the tweets using their own Twitter API credentials.

TD Dataset Embeddings

• To enable reproducibility of our results, we provide SentenceBERT embeddings generated from the TD dataset using the sentence-t5-base model.
• These embeddings are non-reversible and do not contain any text, metadata, or identifiers linking back to individuals.
• The embeddings are strictly derived for academic use and are provided for research reproducibility only.

Anonymity and Privacy

• No personally identifiable information (PII) is included.
• Tweet text, usernames, timestamps, and other metadata are not shared.
• Embeddings are not linked to reconstructable content.

✅ Ethics Approval

This study was approved by the Queensland University of Technology (QUT) Human Research Ethics Committee under Low-Risk Application No. LR 2022-6079-10144.