Evaluating the Novelty of Information

In today's digital space, vast amounts of information circulate, including propositions about individuals. While most research examines whether a proposition is true or false, this study focused on novelty, the extent to which a proposition introduces new information into the public sphere. Treating novelty as a distinct preliminary layer can improve sensitivity evaluation and enhance privacy-protective systems by identifying new, potentially harmful disclosures. This work presents a fully automated seven-step pipeline that combines web search, text extraction, and semantic similarity using Sentence Bidirectional Encoder Representations from Transformers to evaluate novelty. Negated propositions were assessed based on the novelty of their underlying issue rather than the literal negation. To test the approach, a dataset of 54 propositions was constructed, evenly split between previously known and newly fabricated statements. The method achieved 96.3% accuracy, with perfect precision for detecting novel claims and strong performance on known ones.