Survey-Like Dataset of 287 Cybersecurity Papers Selected from Top-Tier Conferences

This dataset accompanies the study \Searching for Diamonds: Cross-Domain Opportunities in Cyber Threat Intelligence\ and was generated through a large-scale automated triage process supported by large language models (LLMs). From an initial corpus of 4,651 papers presented at six top-tier security conferences, LLMs were used to filter and tag publications based on title and abstract similarity to target topics. The resulting subset of 287 papers was exported as a structured, survey-like dataset to support fast retrieval, semantic filtering, and citation management. While the main survey presented in the paper is based on manual reading and synthesis, this dataset was designed to assist researchers in identifying the most relevant works more efficiently. It includes data and metadata related to Cyber Threat Intelligence (CTI), Natural Language Processing (NLP), and LLM-assisted selecting strategies.