Loading...
Paper Type
ERF
Description
Metaverse has gained widespread attention in both academia and industry. As a result, many new concepts, methodologies, and applications are emerging in recent years. Constructing a Metaverse knowledge graph can help with knowledge management and facilitate various intelligent tasks such as recommendations, information retrieval, and chatbots in this emerging field. However, existing knowledge graph construction approaches are not directly applicable in the Metaverse domain due to a lack of labeled dataset, making existing supervised relation extraction approaches impracticable. In this study, we develop a novel graph construction framework by integrating the deep learning approach (i.e., BERT-BiLSTM-CRF) with an unsupervised relation extraction approach based on semantic mining. The framework we propose in this paper can automatically construct knowledge graphs while significantly reducing labeling efforts. Further, we plan to implement the proposed framework on a Metaverse corpus to construct a Metaverse knowledge graph. To test our approach, we will employ an expert-survey method to evaluate the quality of triple sets.
Paper Number
1186
Recommended Citation
Li, Litao; Xu, Ruiyun; and Zhao, J. Leon, "Metaverse Knowledge Graph Construction: An Unsupervised Relation Extraction Approach based on Semantic Mining" (2023). AMCIS 2023 Proceedings. 6.
https://aisel.aisnet.org/amcis2023/conf_theme/conf_theme/6
Metaverse Knowledge Graph Construction: An Unsupervised Relation Extraction Approach based on Semantic Mining
Metaverse has gained widespread attention in both academia and industry. As a result, many new concepts, methodologies, and applications are emerging in recent years. Constructing a Metaverse knowledge graph can help with knowledge management and facilitate various intelligent tasks such as recommendations, information retrieval, and chatbots in this emerging field. However, existing knowledge graph construction approaches are not directly applicable in the Metaverse domain due to a lack of labeled dataset, making existing supervised relation extraction approaches impracticable. In this study, we develop a novel graph construction framework by integrating the deep learning approach (i.e., BERT-BiLSTM-CRF) with an unsupervised relation extraction approach based on semantic mining. The framework we propose in this paper can automatically construct knowledge graphs while significantly reducing labeling efforts. Further, we plan to implement the proposed framework on a Metaverse corpus to construct a Metaverse knowledge graph. To test our approach, we will employ an expert-survey method to evaluate the quality of triple sets.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
Conference Theme