1
Overview
The concept of knowledge graphs was first proposed by Google on May 17, 2012, aiming to describe concepts, entities, events, and their relationships in the objective world, serving as a core foundation for building the next generation of intelligent search engines. In simple terms, a knowledge graph is a network of relationships formed by connecting different types of information. Knowledge graphs provide the ability to analyze problems from the perspective of “relationships.” A simple knowledge graph is shown below.
Currently, well-known high-quality large-scale open knowledge graphs around the world include not only Google’s own knowledge graph but also various multilingual and broad domain knowledge graphs such as DBpedia, Wikidata, ConceptNet, and Microsoft Concept Graph, as well as the Chinese open knowledge graph platform OpenKG. Knowledge graph technology encompasses three research areas: knowledge representation, knowledge graph construction, and knowledge graph applications.
Figure 1 Example of a Simple Knowledge Graph
2
Knowledge Representation
Knowledge representation technology can be divided into two main categories: symbolic and connectionist. The specific representation methods can be classified into three types.
(1) Knowledge representation based on symbolic logic. This mainly includes logical representations, production rule representations, and frame representations. In the current era of large-scale data, these methods are inadequate for solving knowledge representation issues.
(2) Knowledge representation of web content. This mainly includes XML language, RDF description framework, and OWL language. Currently, the widely used method in industry is based on RDF triples.
(3) Representation learning. The goal of representation learning is to represent the semantic information of the research object as dense low-dimensional real-valued vectors through machine learning or deep learning. Compared to traditional methods, this approach can significantly improve computational efficiency, effectively alleviate data sparsity, and facilitate the integration of heterogeneous information from different sources.
3
Knowledge Graph Construction
Currently, most knowledge graphs are constructed using a bottom-up approach. The general process of knowledge graph construction is illustrated in the figure below, mainly involving knowledge acquisition, knowledge fusion, and knowledge processing, which is an ongoing iterative process.
(1) Information acquisition: Extracting entities (concepts), attributes, and the interrelationships between entities from various types of data sources, forming a ontological knowledge representation based on this.
(2) Knowledge fusion: After acquiring new knowledge, it needs to be integrated to eliminate contradictions and ambiguities. For example, some entities may have multiple expressions, and a specific title may correspond to multiple different entities.
(3) Knowledge processing: The newly fused knowledge must undergo quality assessment (with some requiring human involvement) before the qualified parts are added to the knowledge base to ensure its quality. After adding new data, knowledge reasoning, expanding existing knowledge, and generating new knowledge can be performed.
Figure 2 Technical Architecture of Knowledge Graphs
4
Applications of Knowledge Graphs
Knowledge graphs are a field that is both challenging and fascinating. They change the existing methods of information retrieval, enabling conceptual retrieval through reasoning on one hand, and presenting categorized structured knowledge to users in a graphical manner on the other, liberating people from the traditional model of manually sifting through web pages for answers. Services and applications based on knowledge graphs are currently a significant research hotspot. They can be categorized into semantic search, knowledge Q&A, and big data analysis and decision-making based on knowledge. An important role of knowledge graphs is knowledge reasoning, which is a crucial technical avenue in artificial intelligence.
The information technology based on knowledge graphs also has extensive application needs in the military field. Jiang Kai et al. proposed a military information search technology architecture based on knowledge graphs in their paper, detailing specific steps combined with military database searches. The military information search based on knowledge graphs has a wide range of application scenarios, supporting intelligence assurance, operational planning, and decision-making assistance. In aerospace systems, knowledge engineering technologies represented by knowledge graphs can fully utilize the vast amounts of knowledge and information present during research, achieving comprehensive and effective development and utilization of knowledge resources and knowledge innovation.
[1] Liu Qiao, Li Yang et al. Overview of Knowledge Graph Construction Technology [J]. Computer Research and Development, 2016, 53(3):582-600.
[2] Xu Zenglin, Sheng Yongpan, He Lirong et al. Overview of Knowledge Graph Technology [J]. Journal of University of Electronic Science and Technology of China, 2016, 45(4):589-606.
[3] Li Juanzi, Hou Lei. Overview of Knowledge Graph Research [J]. Journal of Shanxi University (Natural Science Edition). 2017, 40(3):454-459.
[4] Jiang Kai, Qian Kui, Zheng Xuan. Military Information Search Technology Architecture Based on Knowledge Graphs [J]. Command Information Systems and Technology. 2016, 7(1):47-52.
end
