Article Preview
Top1. Introduction
As an important feature of knowledge graph, semantic representation and reasoning of structured data has become an important foundational research field of artificial intelligence and been widely studied (Cai, Ma, Wei, & Jiang, 2023; P. Li, Li, et al., 2022; Stylianou, Vlachava, Konstantinidis, Bassiliades, & Peristeras, 2022; Zhou, Zeng, Xu, & Zhao, 2023).
By using RDF Schema (RDFS) (Boneva, Gayo, & Prud’Hommeaux, 2017) which represents knowledge as facts (also known as triples) with the encoding form of subject-predicate-object (s, p, o), knowledge graph provides a pioneering chance to gain insight and value from structured data. However, the bottleneck is that classical knowledge graph contain some implicit semantics between different entities, while the existing RDFS reasoning rules only focus on the semantic properties of individual predicate, but ignore the deep implicit semantics in different adjacency predicates (P. Li, Jiang, Wang, & Yin, 2017; P. Li, Xiao, Aftab, Jiang, & Zhang, 2018). Moreover, the relationships between entities are usually unclear, which cannot be measured in numeric or Boolean values. Unfortunately, the existing RDF triplets cannot well describe such fuzzy semantic information (P. Li, Wang, et al., 2022; Yang, Wang, Zhang, & Wang, 2020). Aiming to solve the aforementioned limitations of the RDFS model, this work investigates the semantically-enhanced knowledge discovery method for knowledge graph by introducing fuzzy theory into consideration.
The main contributions of the paper were highlighted as following:
- •
Fuzzy knowledge graph is defined which is a more general description of classical knowledge graph.
- •
The fuzzy semantic between adjacency predicates is analyzed and the corresponding mathematical model is designed.
- •
Some fuzzy reasoning rules which can provide better process support for reasoning results are presented to realize fuzzy semantic extension.
- •
Performance of the strategy discovers more implicit valid knowledge with fuzzy semantic.
The rest of the paper is organized as follows. Section 2 briefly reviews related works in semantic discovery about knowledge graph. Section 3 introduces the limitations of current reasoning rules by providing a simple example as the motivation and present some new annotating and reasoning strategies to depict the fuzzy semantic between different entities in knowledge graph. In Section 4 we define a well-defined framework as the formalized expression of our fuzzy semantic discovery model for knowledge graph and design the details of processing algorithms. Section 5 is devoted to evaluating our method. Finally, we draw our conclusion and outline the future work in Section 6.
TopIn 2012, Google firstly launched its knowledge graph project, called Knowledge Vault (Dong et al., 2014), and take it as the foundation to construct the next generation of intelligent search engine. In recent years, a growing number of graph-structured data sources such as (Färber, Bartscherer, Menne, & Rettinger, 2018; Frber, 2019; Ruqian et al., 2020) have been published and joined in Web of Linked Data.