Hallucinate or Memorize? The Two Sides of Probabilistic Learning in Large Language Models

• URL: http://arxiv.org/abs/2511.08877v1
• Date: Thu, 13 Nov 2025 01:13:33 GMT
• Title: Hallucinate or Memorize? The Two Sides of Probabilistic Learning in Large Language Models
• Authors: Junichiro Niimi,
• Abstract summary: Large language models (LLMs) have been increasingly applied to a wide range of tasks, from natural language understanding to code generation.<n>While they have also been used to assist in citation recommendation, the hallucination of non-existent papers remains a major issue.<n>This study hypothesizes that an LLM's ability to correctly produce records depends on whether the underlying knowledge is generated or memorized.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models (LLMs) have been increasingly applied to a wide range of tasks, from natural language understanding to code generation. While they have also been used to assist in citation recommendation, the hallucination of non-existent papers remains a major issue. Building on prior studies, this study hypothesizes that an LLM's ability to correctly produce bibliographic records depends on whether the underlying knowledge is generated or memorized, with highly cited papers (i.e., more frequently appear in the pretraining corpus) showing lower hallucination rates. We therefore assume citation count as a proxy for training data redundancy (i.e., the frequency with which a given bibliographic record appears in the pretraining corpus) and investigate how citation frequency affects hallucinated references in LLM outputs. Using GPT-4.1, we generated and manually verified 100 citations across twenty computer-science domains, and measured factual consistency via cosine similarity between generated and authentic metadata. The results revealed that (i) citation count is strongly correlated with factual accuracy, (ii) bibliographic information becomes almost verbatim memorized beyond roughly 1,000 citations, and (iii) memory interference occurs when multiple highly cited papers share similar content. These findings indicate a threshold where generalization shifts into memorization, with highly cited papers being nearly verbatim retained in the model.

Related papers

• CiteAudit: You Cited It, But Did You Read It? A Benchmark for Verifying Scientific References in the LLM Era [51.63024682584688]
  Large language models (LLMs) introduce a new risk: fabricated references that appear plausible but correspond to no real publications.<n>We present the first comprehensive benchmark and detection framework for hallucinated citations in scientific writing.<n>Our framework significantly outperforms prior methods in both accuracy and interpretability.
  arXiv  Detail & Related papers  (2026-02-26T19:17:39Z)
• Hallucinations in Bibliographic Recommendation: Citation Frequency as a Proxy for Training Data Redundancy [0.0]
  Large hallucination models (LLMs) have been increasingly applied to a wide range of tasks.<n>This study hypothesizes that an LLM's ability to correctly produce information depends on whether the underlying knowledge is generated or memorized.
  arXiv  Detail & Related papers  (2025-10-29T10:51:35Z)
• Cite Pretrain: Retrieval-Free Knowledge Attribution for Large Language Models [44.31597857713689]
  We introduce Active Indexing for the first stage, which creates generalizable, source-anchored bindings.<n>Experiments with Qwen-2.5-7B&3B show that Active Indexing consistently outperforms a Passive Indexing baseline.<n> internal citations complement external ones by making the model more robust to retrieval noise.
  arXiv  Detail & Related papers  (2025-06-21T04:48:05Z)
• How Deep Do Large Language Models Internalize Scientific Literature and Citation Practices? [1.130790932059036]
  We show that large language models (LLMs) reinforce the Matthew effect in citations by consistently favoring highly cited papers.<n>We analyze 274,951 references generated by GPT-4o for 10,000 papers.
  arXiv  Detail & Related papers  (2025-04-03T17:04:56Z)
• The Noisy Path from Source to Citation: Measuring How Scholars Engage with Past Research [20.649638393774048]
  We introduce a computational pipeline to quantify citation fidelity at scale.<n>Using full texts of papers, the pipeline identifies citations in citing papers and the corresponding claims in cited papers.<n>Using a quasi-experiment, we establish the "telephone effect" - when citing papers have low fidelity to the original claim, future papers that cite the citing paper and the original have lower fidelity to the original.
  arXiv  Detail & Related papers  (2025-02-27T22:47:03Z)
• Demystifying Verbatim Memorization in Large Language Models [67.49068128909349]
  Large Language Models (LLMs) frequently memorize long sequences verbatim, often with serious legal and privacy implications. We develop a framework to study verbatim memorization in a controlled setting by continuing pre-training from Pythia checkpoints with injected sequences. We find that (1) non-trivial amounts of repetition are necessary for verbatim memorization to happen; (2) later (and presumably better) checkpoints are more likely to memorize verbatim sequences, even for out-of-distribution sequences.
  arXiv  Detail & Related papers  (2024-07-25T07:10:31Z)
• Large Language Models Reflect Human Citation Patterns with a Heightened Citation Bias [1.7812428873698407]
  Citation practices are crucial in shaping the structure of scientific knowledge, yet they are often influenced by contemporary norms and biases. The emergence of Large Language Models (LLMs) introduces a new dynamic to these practices. Here, we analyze these characteristics in an experiment using a dataset from AAAI, NeurIPS, ICML, and ICLR.
  arXiv  Detail & Related papers  (2024-05-24T17:34:32Z)
• Deep Graph Learning for Anomalous Citation Detection [55.81334139806342]
  We propose a novel deep graph learning model, namely GLAD (Graph Learning for Anomaly Detection), to identify anomalies in citation networks. Within the GLAD framework, we propose an algorithm called CPU (Citation PUrpose) to discover the purpose of citation based on citation texts.
  arXiv  Detail & Related papers  (2022-02-23T09:05:28Z)
• Quantifying Memorization Across Neural Language Models [61.58529162310382]
  Large language models (LMs) have been shown to memorize parts of their training data, and when prompted appropriately, they will emit the memorized data verbatim. This is undesirable because memorization violates privacy (exposing user data), degrades utility (repeated easy-to-memorize text is often low quality), and hurts fairness (some texts are memorized over others). We describe three log-linear relationships that quantify the degree to which LMs emit memorized training data.
  arXiv  Detail & Related papers  (2022-02-15T18:48:31Z)
• Towards generating citation sentences for multiple references with intent control [86.53829532976303]
  We build a novel generation model with the Fusion-in-Decoder approach to cope with multiple long inputs. Experiments demonstrate that the proposed approaches provide much more comprehensive features for generating citation sentences.
  arXiv  Detail & Related papers  (2021-12-02T15:32:24Z)
• Enhancing Scientific Papers Summarization with Citation Graph [78.65955304229863]
  We redefine the task of scientific papers summarization by utilizing their citation graph. We construct a novel scientific papers summarization dataset Semantic Scholar Network (SSN) which contains 141K research papers in different domains. Our model can achieve competitive performance when compared with the pretrained models.
  arXiv  Detail & Related papers  (2021-04-07T11:13:35Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.