Factors That Support Grounded Responses in LLM Conversations: A Rapid Review

2025/11/24 by Iwashima, Gabriele Cesar, Rodrigues, Claudia Susie, Dipolitto, Claudio +1
#Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.2511.21762

Abstract

Large language models (LLMs) may generate outputs that are misaligned with user intent, lack contextual grounding, or exhibit hallucinations during conversation, which compromises the reliability of LLM-based applications. This review aimed to identify and analyze techniques that align LLM responses with conversational goals, ensure grounding, and reduce hallucination and topic drift. We conducted a Rapid Review guided by the PRISMA framework and the PICO strategy to structure the search, filtering, and selection processes. The alignment strategies identified were categorized according to the LLM lifecycle phase in which they operate: inference-time, post-training, and reinforcement learning-based methods. Among these, inference-time approaches emerged as particularly efficient, aligning outputs without retraining while supporting user intent, contextual grounding, and hallucination mitigation. The reviewed techniques provided structured mechanisms for improving the quality and reliability of LLM responses across key alignment objectives.

Citations

• LLM Post-Training: A Deep Dive into Reasoning Large Language Models
• Reinforcement Learning Enhanced LLMs: A Survey
• Can LLM Graph Reasoning Generalize beyond Pattern Memorization?
• Hybrid Alignment Training for Large Language Models
• CodecLM: Aligning Language Models with Tailored Synthetic Data
• Direct Large Language Model Alignment Through Self-Rewarding Contrastive Prompt Distillation
• RS-DPO: A Hybrid Rejection Sampling and Direct Preference Optimization Method for Alignment of Large Language Models
• ARGS: Alignment as Reward-Guided Search
• Linear Alignment: A Closed-form Solution for Aligning Human Preferences without Tuning and Feedback
• The Unlocking Spell on Base LLMs: Rethinking Alignment via In-Context Learning
• Nash Learning from Human Feedback
• A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions
• Beyond Imitation: Leveraging Fine-grained Quality Signals for Alignment
• Improving Generalization of Alignment with Human Preferences through Group Invariant Learning
• SALMON: Self-Alignment with Instructable Reward Models
• Automatic Pair Construction for Contrastive Post-training
• Unsupervised Large Language Model Alignment for Information Retrieval via Contrastive Feedback
• Large Language Model Alignment: A Survey
• RAIN: Your Language Models Can Align Themselves without Finetuning
• Mitigating the Alignment Tax of RLHF
• RRHF: Rank Responses to Align Language Models with Human Feedback without tears
• Aligning Large Language Models with Human: A Survey

Related

• D-SMART: Enhancing LLM Dialogue Consistency via Dynamic Structured Memory And Reasoning Tree
• Should We Fine-Tune or RAG? Evaluating Different Techniques to Adapt LLMs for Dialogue
• Highlight All the Phrases: Enhancing LLM Transparency through Visual Factuality Indicators
• High Accuracy, Less Talk (HALT): Reliable LLMs through Capability-Aligned Finetuning
• A Debate-Driven Experiment on LLM Hallucinations and Accuracy