Feature Computation Procedure for Fake News Detection: An LLM-Based Extraction Approach
IntelITSIS 2025, pp. 112-124
Editorial summary
Language-model embeddings are combined with numerical text indicators, including sentiment and paraphrasing features, to classify news articles. Evaluation uses English FakeNewsNet and Ukrainian Fake vs. True datasets and compares the combined representation with baseline classifiers.
Bibliographic reference
Use the publisher’s record for the citation format required by your journal or organisation.
Andrii Shupta, Pavlo Radiuk, Iurii Krak. "Feature Computation Procedure for Fake News Detection: An LLM-Based Extraction Approach". IntelITSIS 2025, pp. 112-124, 2025.
Test a method against your business task.
Describe the decision your system needs to support and the data you have. We assess whether a method fits, define a prototype and agree how to measure its quality, response time and operating cost.