Knowledge grounded medical dialogue generation using augmented graphs

Article Computer Science, Artificial Intelligence

FlauBERT vs. CamemBERT: Understanding patient's answers by a French medical chatbot

Corentin Blanc et al.

Summary: This study found that in a French corpus of clinical cases, the FlauBERT language model paired with a simple linear network architecture performed the best for intent and slot prediction by a chatbot.

ARTIFICIAL INTELLIGENCE IN MEDICINE (2022)

Add to Collection

Article Computer Science, Artificial Intelligence

THINK: A novel conversation model for generating grammatically correct and coherent responses

Bin Sun et al.

Summary: This study proposes a conversation model called THINK, which avoids generating duplicated and self-contradicting responses by using a complex decoder and increases the coherence of responses by simplifying context vectors in a reasonable way.

KNOWLEDGE-BASED SYSTEMS (2022)

Add to Collection

Article Biochemical Research Methods

BioGPT: generative pre-trained transformer for biomedical text generation and mining

Renqian Luo et al.

Summary: This paper introduces a domain-specific generative Transformer language model BioGPT pre-trained on large-scale biomedical literature, which performs well on biomedical natural language processing tasks, especially achieving high accuracy in relation extraction tasks.

BRIEFINGS IN BIOINFORMATICS (2022)

Add to Collection

Article Mathematical & Computational Biology

MKA: A Scalable Medical Knowledge-Assisted Mechanism for Generative Models on Medical Conversation Tasks

Ke Liang et al.

Summary: This paper proposes a scalable medical knowledge-assisted mechanism (MKA) to help general neural generative models perform better in medical conversation tasks. The mechanism includes a medical-specific knowledge graph with various medical-related information and a defined token concatenation policy. Evaluation results show that models combined with the MKA mechanism outperform original methods in multiple automatic evaluation metrics, with MKA-BERT-GPT achieving state-of-the-art performance.

COMPUTATIONAL AND MATHEMATICAL METHODS IN MEDICINE (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

Multi-goal multi-agent learning for task-oriented dialogue with bidirectional teacher-student learning

Wanwei He et al.

Summary: In this work, a multi-goal multi-agent learning framework is proposed for task-oriented dialogue generation, which includes KB-oriented and context-oriented teacher agents, as well as a student agent. A two-to-one teacher-student learning method coordinates these three networks to help the student integrate expert knowledge and achieve comprehensive performance. Additionally, dialogue memory network and KB memory network are employed to filter and extract information dynamically, improving system effectiveness.

KNOWLEDGE-BASED SYSTEMS (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

Heterogeneous Relational Graph Neural Networks with Adaptive Objective for End-to-End Task-Oriented Dialogue

Qingbin Liu et al.

Summary: End-to-end task-oriented dialogue systems face challenges in effectively incorporating external knowledge bases and addressing entity imbalance. The proposed HRGNN-AO method utilizes heterogeneous relational graphs, shared-private parameterization, and hierarchical attention mechanism to improve response generation, while introducing an adaptive objective to handle entity imbalance and outperform state-of-the-art dialogue systems.

KNOWLEDGE-BASED SYSTEMS (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

KEPLER: A Unified Model for Knowledge Embedding and Pre-trained Language Representation

Xiaozhi Wang et al.

Summary: The paper introduces a unified model for Knowledge Embedding and Pre-trained Language Representation (KEPLER), which integrates entity descriptions into pre-trained language models to optimize knowledge embedding and language modeling objectives, achieving state-of-the-art performances in NLP tasks and KG link prediction.

TRANSACTIONS OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS (2021)

Add to Collection

Proceedings Paper Computer Science, Information Systems

Semi-Supervised Variational Reasoning for Medical Dialogue Generation

Dongdong Li et al.

Summary: The study aims to achieve medical dialogue generation through an end-to-end variational reasoning approach, introducing patient states and physician actions as latent variables for explicit tracking and learning. Experimental results demonstrate the superior performance of the proposed method in medical dialogue generation, and the semi-supervised reasoning method shows comparable performance to fully supervised methods in physician policy learning.

SIGIR '21 - PROCEEDINGS OF THE 44TH INTERNATIONAL ACM SIGIR CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION RETRIEVAL (2021)

Add to Collection

Article Biochemical Research Methods