Fosca Giannotti

Chandana Sree Mala, Gizem Gezici, Fosca Giannotti

Large Language Models (LLMs) excel in language comprehension and generation but are prone to hallucinations, producing factually incorrect or unsupported outputs. Retrieval Augmented Generation (RAG) systems address this issue by grounding LLM responses with external knowledge. This study evaluates the relationship between retriever effectiveness and hallucination reduction in LLMs using three retrieval approaches: sparse retrieval based on BM25 keyword search, dense retrieval using semantic search with Sentence Transformers, and a proposed hybrid retrieval module. The hybrid module incorporates query expansion and combines the results of sparse and dense retrievers through a dynamically weighted Reciprocal Rank Fusion score. Using the HaluBench dataset, a benchmark for hallucinations in question answering tasks, we assess retrieval performance with metrics such as mean average precision and normalised discounted cumulative gain, focusing on the relevance of the top three retrieved documents. Results show that the hybrid retriever achieves better relevance scores, outperforming both sparse and dense retrievers. Further evaluation of LLM-generated answers against ground truth using metrics such as accuracy, hallucination rate, and rejection rate reveals that the hybrid retriever achieves the highest accuracy on fails, the lowest hallucination rate, and the lowest rejection rate. These findings highlight the hybrid retriever's ability to enhance retrieval relevance, reduce hallucination rates, and improve LLM reliability, emphasising the importance of advanced retrieval techniques in mitigating hallucinations and improving response accuracy.

Benedetta Muscato, Praveen Bushipaka, Gizem Gezici et al.

Prior studies show that adopting the annotation diversity shaped by different backgrounds and life experiences and incorporating them into the model learning, i.e. multi-perspective approach, contribute to the development of more responsible models. Thus, in this paper we propose a new framework for designing and further evaluating perspective-aware models on stance detection task,in which multiple annotators assign stances based on a controversial topic. We also share a new dataset established through obtaining both human and LLM annotations. Results show that the multi-perspective approach yields better classification performance (higher F1-scores), outperforming the traditional approaches that use a single ground-truth, while displaying lower model confidence scores, probably due to the high level of subjectivity of the stance detection task.

Benedetta Muscato, Lucia Passaro, Gizem Gezici et al.

In the realm of Natural Language Processing (NLP), common approaches for handling human disagreement consist of aggregating annotators' viewpoints to establish a single ground truth. However, prior studies show that disregarding individual opinions can lead can lead to the side effect of underrepresenting minority perspectives, especially in subjective tasks, where annotators may systematically disagree because of their preferences. Recognizing that labels reflect the diverse backgrounds, life experiences, and values of individuals, this study proposes a new multi-perspective approach using soft labels to encourage the development of the next generation of perspective aware models, more inclusive and pluralistic. We conduct an extensive analysis across diverse subjective text classification tasks, including hate speech, irony, abusive language, and stance detection, to highlight the importance of capturing human disagreements, often overlooked by traditional aggregation methods. Results show that the multi-perspective approach not only better approximates human label distributions, as measured by Jensen-Shannon Divergence (JSD), but also achieves superior classification performance (higher F1 scores), outperforming traditional approaches. However, our approach exhibits lower confidence in tasks like irony and stance detection, likely due to the inherent subjectivity present in the texts. Lastly, leveraging Explainable AI (XAI), we explore model uncertainty and uncover meaningful insights into model predictions.

Benedetta Muscato, Praveen Bushipaka, Gizem Gezici et al.

Subjective NLP tasks usually rely on human annotations provided by multiple annotators, whose judgments may vary due to their diverse backgrounds and life experiences. Traditional methods often aggregate multiple annotations into a single ground truth, disregarding the diversity in perspectives that arises from annotator disagreement. In this preliminary study, we examine the effect of including multiple annotations on model accuracy in classification. Our methodology investigates the performance of perspective-aware classification models in stance detection task and further inspects if annotator disagreement affects the model confidence. The results show that multi-perspective approach yields better classification performance outperforming the baseline which uses the single label. This entails that designing more inclusive perspective-aware AI models is not only an essential first step in implementing responsible and ethical AI, but it can also achieve superior results than using the traditional approaches.

Daniele Gambetta, Gizem Gezici, Fosca Giannotti et al.

As synthetic content increasingly infiltrates the web, generative AI models may be retrained on their own outputs: a process termed "autophagy". This leads to model collapse: a progressive loss of performance and diversity across generations. Recent studies have examined the emergence of model collapse across various generative AI models and data types, and have proposed mitigation strategies that rely on incorporating human-authored content. However, current characterizations of model collapse remain limited, and existing mitigation methods assume reliable knowledge of whether training data is human-authored or AI-generated. In this paper, we address these gaps by introducing new measures that characterise collapse directly from a model's next-token probability distributions, rather than from properties of AI-generated text. Using these measures, we show that the degree of collapse depends on the complexity of the initial training set, as well as on the extent of autophagy. Our experiments prompt a new suggestion: that model collapse occurs when a model trains on data that does not "surprise" it. We express this hypothesis in terms of the well-known Free Energy Principle in cognitive science. Building on this insight, we propose a practical mitigation strategy: filtering training items by high surplexity, maximising the surprise of the model. Unlike existing methods, this approach does not require distinguishing between human- and AI-generated data. Experiments across datasets and models demonstrate that our strategy is at least as effective as human-data baselines, and even more effective in reducing distributional skewedness. Our results provide a richer understanding of model collapse and point toward more resilient approaches for training generative AI systems in environments increasingly saturated with synthetic data.