HighlightCurated by Aramai EditorialFrontiers in Artificial Intelligence

Advancing engineering research through context-aware and knowledge graph–based retrieval-augmented generation

A study on improving the accuracy of large language models in generating technical content.

The authors propose a new Retrieval-Augmented Generation (RAG) model for engineering domains, which uses contextual information to improve relevance. The model is built on the n8n automation system and can retrieve densely linked concepts from multiple knowledge graphs. This approach aims to mitigate the shortcomings of traditional RAG techniques in treating isolated information.

Based on: Advancing engineering research through context-aware and knowledge graph–based retrieval-augmented generation · Frontiers in Artificial Intelligence

HighlightCurated by Aramai EditorialJournal of the American Medical Informatics Association

Detecting emergencies in patient portal messages using large language models and knowledge graph-based retrieval-augmented generation

Study on using large language models and a knowledge graph to triage patient messages for emergency care.

The study evaluates the effectiveness of four models in detecting emergency messages in patient portals, with a focus on integrating large language models (LLMs) with a knowledge graph. The results show that the model incorporating a global search within the knowledge graph outperformed other approaches. This research contributes to the development of AI-assisted triage systems for improving patient safety.

Based on: Detecting emergencies in patient portal messages using large language models and knowledge graph-based retrieval-augmented generation · Journal of the American Medical Informatics Association

HighlightCurated by Aramai EditorialarXiv (Cornell University)

Knowledge Graph-based Retrieval-Augmented Generation for Schema Matching

A proposed model for schema matching using knowledge graphs and retrieval-augmented generation.

The authors propose a Knowledge Graph-based Retrieval-Augmented Generation model (KG-RAG4SM) to address semantic ambiguities in schema matching. The model introduces novel vector-based, graph traversal-based, and query-based graph retrievals. Experimental results show that KG-RAG4SM outperforms state-of-the-art methods in terms of precision and F1 score on various datasets.

Based on: Knowledge Graph-based Retrieval-Augmented Generation for Schema Matching · arXiv (Cornell University)

HighlightCurated by Aramai EditorialProceedings of the AAAI Conference on Artificial Intelligence

Adapting to Non-Stationary Environments: Multi-Armed Bandit Enhanced Retrieval-Augmented Generation on Knowledge Graphs

A paper proposing a multi-objective multi-armed bandit enhanced RAG framework for knowledge graphs.

The authors introduce a framework that adapts to non-stationary environments by selecting the most suitable retrieval method based on user feedback and historical performance. This approach is applied to Retrieval-Augmented Generation (RAG) on knowledge graphs, aiming to enhance reasoning capabilities of large language models. Experiments demonstrate improved performance in both stationary and non-stationary settings.

Based on: Adapting to Non-Stationary Environments: Multi-Armed Bandit Enhanced Retrieval-Augmented Generation on Knowledge Graphs · Proceedings of the AAAI Conference on Artificial Intelligence

HighlightCurated by Aramai Editorialopenalex.org

SimGRAG: Leveraging Similar Subgraphs for Knowledge Graphs Driven Retrieval-Augmented Generation

A paper proposing a novel method called SimGRAG for knowledge graph-driven retrieval-augmented generation.

The authors propose SimGRAG, a two-stage process that aligns query texts and KG structures using an LLM to transform queries into a desired graph pattern. They also develop an optimized retrieval algorithm. Experiments show that SimGRAG outperforms state-of-the-art methods in question answering and fact verification.

Based on: SimGRAG: Leveraging Similar Subgraphs for Knowledge Graphs Driven Retrieval-Augmented Generation

HighlightCurated by Aramai EditorialProceedings of the AAAI Conference on Artificial Intelligence

Graph Neural Prompting with Large Language Models

A novel plug-and-play method to enhance pre-trained large language models using knowledge graphs.

The paper proposes Graph Neural Prompting (GNP), a method to assist LLMs in learning beneficial knowledge from KGs. GNP includes various designs, such as a graph neural network encoder and a cross-modality pooling module. Experiments demonstrate the superiority of GNP on commonsense and biomedical reasoning tasks across different LLM sizes.

Based on: Graph Neural Prompting with Large Language Models · Proceedings of the AAAI Conference on Artificial Intelligence

HighlightCurated by Aramai Editorialopenalex.org

Cross-Data Knowledge Graph Construction for LLM-enabled Educational Question-Answering System: A Case Study at HCMUT

A case study on constructing a knowledge graph to enhance the performance of large language models in educational question-answering systems.

This paper presents a case study on constructing a cross-data knowledge graph to improve the performance of large language models (LLMs) in educational question-answering systems. The authors propose integrating LLMs with knowledge graphs to provide factual context and address limitations such as remembering events and incorporating new information.

Based on: Cross-Data Knowledge Graph Construction for LLM-enabled Educational Question-Answering System: A Case Study at HCMUT

HighlightCurated by Aramai EditorialarXiv (Cornell University)

CommunityKG-RAG: Leveraging Community Structures in Knowledge Graphs for Advanced Retrieval-Augmented Generation in Fact-Checking

A novel zero-shot framework integrating community structures with RAG systems to enhance fact-checking.

This paper introduces CommunityKG-RAG, a framework that combines knowledge graphs and retrieval-augmented generation to improve fact-checking. It utilizes multi-hop community structures within KGs to enhance accuracy and relevance of information retrieval. Experimental results show that CommunityKG-RAG outperforms traditional methods in fact-checking.

Based on: CommunityKG-RAG: Leveraging Community Structures in Knowledge Graphs for Advanced Retrieval-Augmented Generation in Fact-Checking · arXiv (Cornell University)

HighlightCurated by Aramai Editorialopenalex.org

GenTKG: Generative Forecasting on Temporal Knowledge Graph with Large Language Models

A novel retrieval-augmented generation framework for temporal knowledge graph forecasting using large language models.

The paper proposes GenTKG, a framework that combines temporal logical rule-based retrieval and few-shot parameter-efficient instruction tuning to address challenges in temporal knowledge graph forecasting. Experiments show that GenTKG outperforms conventional methods with low computation resources and limited training data. The work highlights the potential of large language models in the temporal knowledge graph domain.

Based on: GenTKG: Generative Forecasting on Temporal Knowledge Graph with Large Language Models

HighlightCurated by Aramai Editorialopenalex.org

Leveraging LLM based Retrieval-Augmented Generation for Legal Knowledge Graph Completion

A paper proposing a model for legal knowledge graph completion using Large Language Models and retrieval-augmented generation.

The authors propose RA-KG-LLM, a model that combines the generative capabilities of Large Language Models with retrieval-augmented technology to enhance semantic information and interrelations mining in knowledge graphs. The model is evaluated on five real datasets, including Cail2022, where it achieves better performance than state-of-the-art related works. This paper aims to improve the completeness and accuracy of legal knowledge graphs.

Based on: Leveraging LLM based Retrieval-Augmented Generation for Legal Knowledge Graph Completion

HighlightCurated by Aramai EditorialarXiv (Cornell University)

Graph Retrieval-Augmented Generation: A Survey

A survey on GraphRAG methodologies for retrieval-augmented generation.

The paper provides a comprehensive overview of GraphRAG, a framework that leverages structural information in databases to improve the accuracy and context-awareness of large language models. It formalizes the GraphRAG workflow and outlines core technologies and training methods. The survey also examines downstream tasks, application domains, evaluation methodologies, and industrial use cases.

Based on: Graph Retrieval-Augmented Generation: A Survey · arXiv (Cornell University)

HighlightCurated by Aramai Editorialopenalex.org

TRACE the Evidence: Constructing Knowledge-Grounded Reasoning Chains for Retrieval-Augmented Generation

A paper proposing TRACE, a method for constructing knowledge-grounded reasoning chains to enhance multi-hop question answering.

The authors propose TRACE, a method that constructs knowledge-grounded reasoning chains to improve multi-hop question answering. TRACE uses a KG Generator and Autoregressive Reasoning Chain Constructor to build reasoning chains from retrieved documents. Experimental results show an average performance improvement of up to 14.03% compared to using all retrieved documents.

Based on: TRACE the Evidence: Constructing Knowledge-Grounded Reasoning Chains for Retrieval-Augmented Generation