Presentation: 2025 ND EPSCoR Annual conference 

October 21, 2025, NDSU Memorial Union, Fargo, North Dakota

CyberTweetGrader&Labeler: Social Media Analytics for Cyberattack Intelligence

I present CyberTweetGrader&Labeler, a domain-specific NLP pipeline for detecting, prioritizing, and labeling social-media discourse related to cyberattacks. The method combines an incident-focused data-curation protocol with targeted feature groups—event-specific and related entities, cybersecurity terms, and law-enforcement/media references and applies a composite relevance score with tiered labeling to surface higher-value signals. I describe the end-to-end pipeline (normalization, filtering, scoring, labeling), report empirical results from a healthcare cyberattack incident case study, and compare performance with general-purpose baseline classifiers using precision, recall, and F1. Error analyses highlight contextual ambiguity, media resharing, and class imbalance; brief ablations illustrate the contribution of feature families to ranking stability. I invite interested colleagues to collaborate on two concrete next steps that fit small research allocations: (1) independent validation of the pipeline on a second incident and (2) assembling a compact, multi-incident evaluation set suitable for future joint manuscripts. The goal is to make the approach portable across North Dakota research settings while keeping the time commitment realistic for teaching-focused faculty.