Construction of Journal Knowledge Graph Based on Deep Learning and LLM DOI Open Access
Jinyu Zuo,

Jiaojiao Niu

Electronics, Год журнала: 2025, Номер 14(9), С. 1728 - 1728

Опубликована: Апрель 24, 2025

Knowledge graphs are powerful tools for representing the relationships between concepts and entities in real world through triples. Due to their superior knowledge representation efficient reasoning abilities, have gained widespread attention across various fields, leading development multiple domains. However, research on construction of journal remains relatively limited, posing challenges integration utilization domain. To address this gap, study explores effective methods constructing develops a graph-based question answering system. Specifically, datasets were collected from sources using Scrapy framework, encompassing structured, semi-structured, unstructured data. A BERT-BiLSTM-CRF framework was then employed extract entities, attributes, semi-structured In addition, constructed graph integrated with large language models (LLMs) build journal-related system, facilitating querying utilization. Finally, Neo4j used storing graph.

Язык: Английский

Auto-summarization of the texts of construction dispute precedents DOI
Wonkyoung Seo, Youngcheol Kang

Advanced Engineering Informatics, Год журнала: 2025, Номер 65, С. 103381 - 103381

Опубликована: Апрель 22, 2025

Язык: Английский

Процитировано

0

Construction of Journal Knowledge Graph Based on Deep Learning and LLM DOI Open Access
Jinyu Zuo,

Jiaojiao Niu

Electronics, Год журнала: 2025, Номер 14(9), С. 1728 - 1728

Опубликована: Апрель 24, 2025

Knowledge graphs are powerful tools for representing the relationships between concepts and entities in real world through triples. Due to their superior knowledge representation efficient reasoning abilities, have gained widespread attention across various fields, leading development multiple domains. However, research on construction of journal remains relatively limited, posing challenges integration utilization domain. To address this gap, study explores effective methods constructing develops a graph-based question answering system. Specifically, datasets were collected from sources using Scrapy framework, encompassing structured, semi-structured, unstructured data. A BERT-BiLSTM-CRF framework was then employed extract entities, attributes, semi-structured In addition, constructed graph integrated with large language models (LLMs) build journal-related system, facilitating querying utilization. Finally, Neo4j used storing graph.

Язык: Английский

Процитировано

0