Year: 2025
Type: article
Abstract: Hallucination in textual data generated by language models poses significant challenges in ensuring the accuracy and reliability of information. This research aims to explore various hallucination detection and mitigation methods to address this issue. The study investigates techniques such as hallucination detection using prompting methods and classification models like Vectara. By analyzing these approaches, the research provides insights into their effectiveness and limitations in identifying and mitigating hallucinated content. The findings highlight the importance of robust detection mechanisms to enhance the credibility of generated text, offering valuable contributions to the field of natural language processing.
Authors: Harshaprabha Shetty, Hrishikesh Bhagawati -, Dhruv Patel -, Bhaskara Sai Kumar Rongali -
Institution: Accenture (Switzerland)
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4409646498