arXiv:2509.10746v2 Announce Type: replace 
Abstract: Large language models in healthcare often miss critical emotional cues, delivering medically sound but emotionally flat advice. Such responses are insufficient in clinical encounters, where distressed or vulnerable patients rely on empathic communication to support safety, adherence, and trust. We present RECAP (Reflect-Extract-Calibrate-Align-Produce), an inference-time framework that guides models through structured emotional reasoning without retraining. RECAP decomposes patient input into appraisal-theoretic stages, identifies psychological factors, and assigns Likert-based emotion likelihoods that clinicians can inspect or override, producing nuanced and auditable responses. Across EmoBench, SECEU, and EQ-Bench, RECAP improves emotional reasoning by 22-28% on 8B models and 10-13% on larger models over zero-shot baselines. In blinded evaluations, oncology clinicians rated RECAP's responses as more empathetic, supportive, and context-appropriate than prompting baselines. These findings demonstrate that modular, principled prompting can enhance emotional intelligence in medical AI while maintaining transparency and accountability for clinical deployment.

تم تقديم إطار عمل جديد يسمى RECAP لتحسين التفكير العاطفي في أنظمة الحوار الطبي، مما يعالج قيود النماذج اللغوية الكبيرة التي غالبًا ما تقدم استجابات عاطفية مسطحة. من خلال تحليل مدخلات المرضى إلى مراحل عاطفية منظمة والسماح للأطباء بفحص أو تجاوز احتمالات العواطف، أظهر RECAP تحسينًا في التفكير العاطفي بنسبة 22-28% على النماذج ذات 8 مليارات و10-13% على النماذج الأكبر عبر عدة معايير.

Se ha presentado un nuevo marco llamado RECAP para mejorar el razonamiento emocional en los sistemas de diálogo médico, abordando las limitaciones de los grandes modelos de lenguaje que a menudo ofrecen respuestas emocionalmente planas. Al descomponer la entrada del paciente en etapas emocionales estructuradas y permitir que los clínicos inspeccionen o anulen las probabilidades emocionales, RECAP ha demostrado mejorar el razonamiento emocional en un 22-28 % en modelos de 8 mil millones y en un 10-13 % en modelos más grandes en varios benchmarks.

Un nouveau cadre appelé RECAP a été introduit pour améliorer le raisonnement émotionnel dans les systèmes de dialogue médical, répondant aux limites des grands modèles de langage qui fournissent souvent des réponses émotionnellement plates. En décomposant les entrées des patients en étapes émotionnelles structurées et en permettant aux cliniciens d'inspecter ou de remplacer les probabilités émotionnelles, RECAP a montré une amélioration du raisonnement émotionnel de 22 à 28 % sur des modèles de 8 milliards et de 10 à 13 % sur des modèles plus grands à travers divers benchmarks.

A new framework called RECAP has been introduced to enhance emotional reasoning in medical dialogue systems, addressing the limitations of large language models that often provide emotionally flat responses. By decomposing patient input into structured emotional stages and allowing clinicians to inspect or override emotion likelihoods, RECAP has shown to improve emotional reasoning by 22-28% on 8B models and 10-13% on larger models across various benchmarks.

RECAP: Transparent Inference-Time Emotion Alignment for Medical Dialogue Systems

arXiv:2509.04472v2 Announce Type: replace 
Abstract: Understanding user intent is essential for effective planning in conversational assistants, particularly those powered by large language models (LLMs) coordinating multiple agents. However, real-world dialogues are often ambiguous, underspecified, or dynamic, making intent detection a persistent challenge. Traditional classification-based approaches struggle to generalize in open-ended settings, leading to brittle interpretations and poor downstream planning. We propose RECAP (REwriting Conversations for Agent Planning), a new benchmark designed to evaluate and advance intent rewriting, reframing user-agent dialogues into concise representations of user goals. RECAP captures diverse challenges such as ambiguity, intent drift, vagueness, and mixed-goal conversations. Alongside the dataset, we introduce an LLM-based evaluator that assesses planning utility given the rewritten intent. Using RECAP, we develop a prompt-based rewriting approach that outperforms baselines, in terms of plan preference. We further demonstrate that fine-tuning two DPO-based rewriters yields additional utility gains. Our results highlight intent rewriting as a critical and tractable component for improving agentic planning in open-domain dialogue systems.

تم تقديم RECAP (إعادة كتابة المحادثات لفهم النوايا في التخطيط الوكالي) مؤخرًا بهدف تعزيز فهم النوايا في المساعدات المحادثية المدعومة بنماذج اللغة الكبيرة (LLMs). يتناول هذا المعيار التحديات المتعلقة بالحوار الغامض والديناميكي، مقترحًا طريقة لإعادة كتابة محادثات المستخدم-الوكيل إلى تمثيلات واضحة لأهداف المستخدم، مما يحسن فعالية التخطيط.

La reciente introducción de RECAP (REwriting Conversations for Agent Planning) tiene como objetivo mejorar la comprensión de las intenciones en los asistentes conversacionales impulsados por grandes modelos de lenguaje (LLMs). Este benchmark aborda los desafíos de los diálogos ambiguos y dinámicos, proponiendo un método para reescribir las conversaciones entre usuarios y agentes en representaciones claras de los objetivos del usuario, mejorando así la efectividad de la planificación.

La récente introduction de RECAP (REwriting Conversations for Agent Planning) vise à améliorer la compréhension des intentions dans les assistants conversationnels alimentés par de grands modèles de langage (LLMs). Ce benchmark aborde les défis des dialogues ambigus et dynamiques, proposant une méthode pour réécrire les conversations utilisateur-agent en représentations claires des objectifs utilisateur, améliorant ainsi l'efficacité de la planification.

The recent introduction of RECAP (REwriting Conversations for Agent Planning) aims to enhance intent understanding in conversational assistants powered by large language models (LLMs). This benchmark addresses the challenges of ambiguous and dynamic dialogues, proposing a method to rewrite user-agent conversations into clear representations of user goals, thereby improving planning effectiveness.

RECAP: REwriting Conversations for Intent Understanding in Agentic Planning

One More Thing in AI – Your Shortcut to AI Mastery

RECAP: Transparent Inference-Time Emotion Alignment for Medical Dialogue Systems

Was this article worth reading? Share it

One More Thing in AI

LucidQuery AI

Nudge AI

Supanote

DBTAI

Prepin

Ready to build your own newsroom?