Web Science and Digital Libraries Research Group

  Fig 1 from Rawte et al. Taxonomy for Hallucination in Large Foundation Model The rise of large language models (LLMs) has brought about accelerated advances in natural language processing (NLP), enabling powerful results in text generation, comprehension, and reasoning. However, alongside these advancements comes a persistent and critical issue: hallucination. Defined as the generation of content that deviates from factual accuracy or the provided input, hallucination presents a multifaceted challenge with implications across various domains, from journalism to healthcare. This blog post presents insights from three recent comprehensive surveys on hallucination in natural language generation (NLG) and foundation models to provide an understanding of the problem, its causes, and ongoing mitigation efforts. “ Survey of Hallucination in Natural Language Generation ” by Ji et al. (2022) provides a foundational exploration of hallucination in various NLG tasks, including abstractiv...

Figure 1. An Illustration of the transition from storytelling to advanced conceptualization. Image Generated by DALL-E. In my previous blog,— “ Do Large Language Models Agree on Entity Extraction? ”— I explored how different Large Language Models (LLMs) approach Named Entity Recognition (NER) tasks, highlighting the inconsistencies in entity identification and categorization. By leveraging techniques such as Sentence-BERT embeddings , and cosine similarity , I demonstrated how to unify outputs from multiple LLMs into a coherent and reliable list of entities, which is essential for tasks requiring consistent information extraction.  This blog transitions from entity extraction to a broader exploration of how LLMs move from narrative analysis to conceptual modeling and simulation . While extracting entities is a critical step in understanding narratives, the next challenge lies in structuring those narratives into meaningful models that can inform decision-making and support real...