Knowledge Tree
prompt engineering
Facts (50)
Sources
Survey and analysis of hallucinations in large language models frontiersin.org Sep 29, 2025 10 facts
claimUnderstanding whether hallucinations are caused by prompt formulation or intrinsic model behavior is essential for designing effective prompt engineering strategies, developing grounded architectures, and benchmarking Large Language Model reliability.
claimPrompt engineering is a cost-effective, model-agnostic approach to reduce hallucinations at inference time without altering the underlying model parameters.
claimPrompt engineering, particularly Chain-of-Thought (CoT) prompting, reduces hallucination rates in large language models but is not universally effective.
measurementStructured prompting using Chain-of-Thought reduced CPS values to 0.06, demonstrating the effectiveness of structured prompt engineering as noted by Zhou et al. (2022).
claimPrompt engineering is a superficial fix that cannot fully eliminate model-intrinsic hallucinations, particularly when dealing with deceptive prompts or ambiguous tasks.
claimModel-based mitigation techniques require more infrastructure and training resources than prompt engineering but offer more robust mitigation for model-intrinsic hallucinations.
perspectiveEffective hallucination mitigation requires targeted strategies including prompt engineering improvements, robust factual grounding, and careful model selection based on specific deployment needs and risk tolerance.
claimDeepSeek-67B demonstrates strong internal consistency and confidence, but its hallucinations are primarily caused by internal factual gaps in its training data or architecture that prompt engineering cannot resolve.
claimPrompt engineering is not a universal solution for mitigating hallucinations in large language models, particularly for models with strong internal biases.
procedureThe prompt engineering protocol used in the study involves five categories: Zero-shot (basic instruction), Few-shot (2-3 input-output examples), Instruction (structured natural language), Chain-of-thought (step-by-step reasoning), and Vague/misleading (intentionally unclear).
A Survey on the Theory and Mechanism of Large Language Models arxiv.org Mar 12, 2026 8 facts
claimPrompt engineering steers a Large Language Model at inference time by modifying the input sequence without updating the model parameters.
referenceThe paper 'Unleashing the potential of prompt engineering in large language models: a comprehensive review' is an arXiv preprint, identified as arXiv:2310.14735.
referenceThe paper 'Theoretical foundations of prompt engineering: from heuristics to expressivity' is an arXiv preprint (arXiv:2512.12688) cited in section 6.2.1 of 'A Survey on the Theory and Mechanism of Large Language Models'.
referenceThe paper 'A systematic survey of prompt engineering in large language models: techniques and applications' is an arXiv preprint (arXiv:2402.07927) cited in section 4.3.2 of 'A Survey on the Theory and Mechanism of Large Language Models'.
claimPrompt engineering provides insight into a Large Language Model's internal behavior because minor modifications to the prompt can result in significant changes in the generated distribution.
claimPrompt engineering in deployed systems requires principled auditing of how instruction conflicts are represented and resolved inside the forward pass, according to Rossi et al. (2024) and Yao et al. (2024b).
referenceThe paper 'Claude 2.0 large language model: tackling a real-world classification problem with a new iterative prompt engineering approach' describes an iterative prompt engineering method applied to the Claude 2.0 model for classification tasks.
claimPrompt engineering serves as the primary interface for translating user intent into a form a Large Language Model can follow, often determining whether the model utilizes prior knowledge, adheres to constraints, or generates structured outputs.
Practices, opportunities and challenges in the fusion of knowledge ... frontiersin.org 3 facts
claimPrompt engineering is a crucial component in knowledge graph-guided large language model reasoning methods.
claimPrompt engineering methods for knowledge graph completion, such as ProLINK and TAGREAL, suffer from information loss because they must split complex entity names into subword fragments.
claimPrompt engineering for Knowledge Graph (KG) completion involves designing input prompts to guide Large Language Models (LLMs) in inferring and filling missing parts of KGs, which enhances multi-hop link prediction and allows handling of unseen cues in zero-sample scenarios.
Medical Hallucination in Foundation Models and Their ... medrxiv.org Mar 3, 2025 3 facts
measurementThe prompt engineering method using ChEBI identifiers maintained 92% terminological consistency with FDA drug labeling databases during pharmacological report generation.
measurementWhen applied to pharmacological report generation, the prompt engineering method using ChEBI identifiers reduced attribute hallucinations, specifically incorrect dosage or formulation details, by 33% compared to baseline prompts.
claimDetection and mitigation strategies for medical hallucinations in Foundation Models include factual verification, consistency checks, uncertainty quantification, and prompt engineering.
Combining large language models with enterprise knowledge graphs frontiersin.org Aug 26, 2024 3 facts
claimIn-context learning offers greater flexibility for adapting to the rapidly evolving field of Large Language Models (LLMs), though prompt engineering is time-consuming and requires methods that are not universally applicable across models, as reported by Zhao et al. (2024).
referenceRecent literature identifies two primary approaches to named entity recognition and relation extraction: creating large training sets with hand-curated or extensive automatic annotations to fine-tune large language models, or using precise natural language instructions to replace domain knowledge with prompt engineering.
claimPrompt engineering for full Knowledge Graph Extraction (KGE) is impractical because the structural mismatch between natural language and knowledge graphs complicates the creation of automated prompts for large knowledge graphs.
A Comprehensive Benchmark and Evaluation Framework for Multi ... arxiv.org Jan 6, 2026 2 facts
referenceA comparative analysis of medical AI implementation methods indicates that Prompt Engineering has very low implementation cost but low consistency, RAG has moderate implementation cost and high consistency, Fine-Tuning has high implementation cost and moderate consistency, and Multi-Agent systems have very high implementation cost and very high consistency.
claimThe Basic setup of the Patient Agent framework, which relies solely on prompt engineering without constraints, exhibits a high hallucination rate and suboptimal behavioral consistency.
Evaluating RAG applications with Amazon Bedrock knowledge base ... aws.amazon.com Mar 14, 2025 2 facts
procedureTo optimize RAG systems, developers should analyze patterns in lower-performing responses to adjust retrieval parameters, refine prompts, or modify knowledge base configurations.
claimThe Amazon Bedrock knowledge base evaluation feature allows users to assess RAG application performance by analyzing how different components, such as knowledge base configuration, retrieval strategies, prompt engineering, model selection, and vector store choices, impact metrics.
KG-RAG: Bridging the Gap Between Knowledge and Creativity - arXiv arxiv.org May 20, 2024 2 facts
claimPrompt engineering techniques, including Chain of Thought (CoT), Tree of Thought (ToT), Graph of Thoughts (GoT), and ReAct (Reason and Act), have demonstrated significant improvements in the reasoning abilities and task-specific actions of Large Language Models.
claimPrompt engineering techniques, including Chain of Thought (CoT), Tree of Thought (ToT), Graph of Thoughts (GoT), and ReAct (Reason and Act), have demonstrated significant improvements in the reasoning abilities and task-specific actions of Large Language Models.
LLM-Powered Knowledge Graphs for Enterprise Intelligence and ... arxiv.org Mar 11, 2025 2 facts
claimThe relation extraction component utilizes Large Language Models (LLMs) with advanced prompt engineering, incorporating both contextual data from the Contextual Retrieval Module (CRM) and extracted entities as input to enhance the precision and relevance of relationship extraction.
claimThe entity extraction component improves precision and consistency by using Large Language Models (LLMs) with prompt engineering and contextual data retrieved from the Contextual Retrieval Module (CRM).
Integrating Knowledge Graphs into RAG-Based LLMs to Improve ... thesis.unipd.it 2 facts
claimCustom prompt engineering strategies are necessary for fact-checking systems because different LLMs benefit from different types of contextual information provided by knowledge graphs.
claimEffective fact-checking performance requires custom prompt engineering strategies because different Large Language Models benefit from different types of contextual information.
Detecting hallucinations with LLM-as-a-judge: Prompt ... - Datadog datadoghq.com Aug 25, 2025 2 facts
claimPrompt engineering for LLM agents involves defining a logical flow across multiple LLM calls and using format restrictions like structured output to create guidelines for LLM output.
perspectiveDatadog posits that gains achieved through prompt engineering can transfer to fine-tuned models.
Unlocking the Potential of Generative AI through Neuro-Symbolic ... arxiv.org Feb 16, 2025 1 fact
claimPrompt engineering techniques, including Chain-of-Thought (CoT) prompting, zero-shot prompting, and few-shot prompting, enable Large Language Models (LLMs) to reason and generalize across diverse tasks without requiring extensive retraining.
A framework to assess clinical safety and hallucination rates of LLMs ... nature.com May 13, 2025 1 fact
claimBaseline prompts (Experiments 1 and 2) served as the initial benchmark for comparing later prompt engineering strategies in the study.
Re-evaluating Hallucination Detection in LLMs - arXiv arxiv.org Aug 13, 2025 1 fact
claimPrompt engineering and dataset-specific post-processing techniques often lack scalability and generalizability across different models and datasets when attempting to improve ROUGE scores.
Hallucinations in LLMs: Can You Even Measure the Problem? linkedin.com Jan 18, 2025 1 fact
procedurePrompt engineering mitigates LLM hallucinations by refining instructions to ensure the model understands the task and restricts its output to verified concepts.
The construction and refined extraction techniques of knowledge ... nature.com Feb 10, 2026 1 fact
referenceChen, B. et al. published 'Unleashing the potential of prompt engineering in large Language models' in Patterns 6 (6), 101260 (2025).
Stanford Study Reveals AI Limitations at Scale - LinkedIn linkedin.com Mar 16, 2026 1 fact
perspectiveThe author asserts that prompt engineering cannot resolve structural AI failures, as these issues are inherent to how current AI systems are designed.
Reference Hallucination Score for Medical Artificial ... medinform.jmir.org Jul 31, 2024 1 fact
referenceNg (2025) provided a practical guide for traditional, complementary, and integrative medicine researchers on prompt engineering for generative artificial intelligence chatbots, published in Integrative Medicine Research.
LLM-KG4QA: Large Language Models and Knowledge Graphs for ... github.com 1 fact
referenceThe paper 'A Prompt Engineering Approach and a Knowledge Graph based Framework for Tackling Legal Implications of Large Language Model Answers' (arXiv, 2024) proposes a framework combining prompt engineering and knowledge graphs to address legal implications in Large Language Model outputs.
On Hallucinations in Artificial Intelligence–Generated Content ... jnm.snmjournals.org 1 fact
claimPrompt engineering is a technique that seeks to improve AI output accuracy by optimizing the structure and content of input instructions.
Applying Large Language Models in Knowledge Graph-based ... arxiv.org Jan 7, 2025 1 fact
referenceBarn et al. developed a prompt engineering meta-model for the 4EM method, which includes domain concepts and modeling language elements, to enable prompt-based interactions in enterprise modeling.
Detect hallucinations in your RAG LLM applications with Datadog ... datadoghq.com May 28, 2025 1 fact
procedureDatadog's hallucination detection feature utilizes an LLM-as-a-judge approach combined with prompt engineering, multi-stage reasoning, and non-AI-based deterministic checks.