Emotion-Aware Embedding Fusion in Large Language Models (Flan-T5, Llama 2, DeepSeek-R1, and ChatGPT 4) for Intelligent Response Generation DOI Creative Commons
Abdur Rasool, Muhammad Khurram Shahzad,

Hafsa Aslam

и другие.

AI, Год журнала: 2025, Номер 6(3), С. 56 - 56

Опубликована: Март 13, 2025

Empathetic and coherent responses are critical in automated chatbot-facilitated psychotherapy. This study addresses the challenge of enhancing emotional contextual understanding large language models (LLMs) psychiatric applications. We introduce Emotion-Aware Embedding Fusion, a novel framework integrating hierarchical fusion attention mechanisms to prioritize semantic features therapy transcripts. Our approach combines multiple emotion lexicons, including NRC Emotion Lexicon, VADER, WordNet, SentiWordNet, with state-of-the-art LLMs such as Flan-T5, Llama 2, DeepSeek-R1, ChatGPT 4. Therapy session transcripts, comprising over 2000 samples, segmented into levels (word, sentence, session) using neural networks, while these pooling techniques refine representations. Attention mechanisms, multi-head self-attention cross-attention, further features, enabling temporal modeling shifts across sessions. The processed embeddings, computed BERT, GPT-3, RoBERTa, stored Facebook AI similarity search vector database, which enables efficient clustering dense spaces. Upon user queries, relevant segments retrieved provided context LLMs, their ability generate empathetic contextually responses. proposed is evaluated practical use cases demonstrate real-world applicability, AI-driven chatbots. system can be integrated existing mental health platforms personalized based on data. experimental results show that our enhances empathy, coherence, informativeness, fluency, surpassing baseline improving LLMs’ intelligence adaptability for

Язык: Английский

Human guided empathetic AI agent for mental health support leveraging reinforcement learning-enhanced retrieval-augmented generation DOI

Gayathri Soman,

M. V. Judy,

Aadhil Muhammad Abou

и другие.

Cognitive Systems Research, Год журнала: 2025, Номер unknown, С. 101337 - 101337

Опубликована: Фев. 1, 2025

Язык: Английский

Процитировано

0
Emotion-Aware Embedding Fusion in Large Language Models (Flan-T5, Llama 2, DeepSeek-R1, and ChatGPT 4) for Intelligent Response Generation DOI Creative Commons
Abdur Rasool, Muhammad Khurram Shahzad,

Hafsa Aslam

и другие.

AI, Год журнала: 2025, Номер 6(3), С. 56 - 56

Опубликована: Март 13, 2025

Empathetic and coherent responses are critical in automated chatbot-facilitated psychotherapy. This study addresses the challenge of enhancing emotional contextual understanding large language models (LLMs) psychiatric applications. We introduce Emotion-Aware Embedding Fusion, a novel framework integrating hierarchical fusion attention mechanisms to prioritize semantic features therapy transcripts. Our approach combines multiple emotion lexicons, including NRC Emotion Lexicon, VADER, WordNet, SentiWordNet, with state-of-the-art LLMs such as Flan-T5, Llama 2, DeepSeek-R1, ChatGPT 4. Therapy session transcripts, comprising over 2000 samples, segmented into levels (word, sentence, session) using neural networks, while these pooling techniques refine representations. Attention mechanisms, multi-head self-attention cross-attention, further features, enabling temporal modeling shifts across sessions. The processed embeddings, computed BERT, GPT-3, RoBERTa, stored Facebook AI similarity search vector database, which enables efficient clustering dense spaces. Upon user queries, relevant segments retrieved provided context LLMs, their ability generate empathetic contextually responses. proposed is evaluated practical use cases demonstrate real-world applicability, AI-driven chatbots. system can be integrated existing mental health platforms personalized based on data. experimental results show that our enhances empathy, coherence, informativeness, fluency, surpassing baseline improving LLMs’ intelligence adaptability for

Язык: Английский

Процитировано

0