Understanding Knowledge Graphs Clearly

01 Origin of Knowledge Graphs

In 1977, American computer scientist Feigenbaum officially named knowledge engineering. He received the Turing Award in 1994 and is hailed as the father of expert systems and the founder of knowledge engineering. Knowledge engineering is a top-down approach that heavily relies on expert intervention, with the fundamental goal of imparting expert knowledge to machines to utilize them for problem-solving.

In traditional knowledge engineering, it is essential to have experts in the relevant field who can express their knowledge. Additionally, knowledge engineers are required to convert the knowledge expressed by experts into a format that computers can process.

With the advent of the internet, knowledge engineering has entered the era of big data. Although traditional knowledge engineering methods are forward-looking, their representational capacity is limited, making it difficult to meet the demands of large-scale open applications in the internet age.

To address these issues, researchers in academia and industry began to seek new solutions, turning their attention to the data itself, and proposed the concept of linked data.

Linked data not only needs to be published on the Semantic Web but also requires establishing connections between data to form a vast network of linked data. Google’s search engine product has made significant breakthroughs in this technology, which they named “Knowledge Graph”.

02 Definition of Knowledge Graphs

The goal of a knowledge graph is to describe various entities or concepts that exist in the real world and the relationships between them. It constitutes a vast semantic network diagram where nodes represent entities or concepts, and edges represent attributes or relationships. Nowadays, knowledge graphs are widely used in various large-scale knowledge bases. Below is an example of a knowledge graph for the entire product lifecycle, as shown in Figure 1.5. Generally, a knowledge graph contains three types of nodes:

Understanding Knowledge Graphs Clearly

▲ Figure 1.5 Knowledge Graph of Product Lifecycle

Entities or concepts refer to distinguishable and independently existing things. For example, in Figure 1.5, products, Product 1, research and design, production, procurement, and quality are all independently existing entities. Everything in the world is composed of multiple concrete things, and entities play the most fundamental role in a knowledge graph, with various relationships existing between different entities.

Attributes and their corresponding values are used to describe the inherent characteristics of entities, forming a directional relationship between attributes and attribute values. Different types of attributes correspond to different types of attribute edges. Attribute values primarily refer to the specific values of the attributes that an object possesses. In Figure 1.5, “procurement,” “production,” and “quality” are several different types of attributes, while the attribute values indicate the quantity and price of procured materials, production quantity and progress, as well as the quality indicators for procurement and production.

Relationships connect two entities and describe their associations. A knowledge graph can be viewed as a vast relational network diagram, where the nodes represent entities or concepts, and the edges are formed by attributes or relationships.

03 Technical Architecture of Knowledge Graphs

The technical architecture of a knowledge graph refers to the structure of the construction model, as shown in the figure below. The figure displays the construction and updating processes of the knowledge graph.

Understanding Knowledge Graphs Clearly

▲ Figure 1.6 Technical Architecture of Knowledge Graphs

The construction of a knowledge graph starts from raw data, including structured, semi-structured, and unstructured data. Through a series of automated or semi-automated techniques, knowledge is extracted from raw databases and third-party databases and stored in the data layer and schema layer of the knowledge base. This process includes five steps: data collection, knowledge extraction, knowledge fusion, knowledge processing, and knowledge application, with each update iteration containing these four phases.

The construction methods of knowledge graphs are primarily top-down and bottom-up.

Top-down construction defines the ontology and data schema of the knowledge graph first, and then adds entities to the knowledge base. This method requires utilizing some existing structured knowledge bases as foundational knowledge bases; for example, the Freebase project adopts this method, sourcing most of its data from Wikipedia.

Bottom-up construction extracts entities from open linked data, selects entities with a high level of confidence to add to the knowledge base, and then constructs the top-level ontology schema.

For most manufacturing enterprises, due to a lack of substantial empirical data, the top-down construction method is primarily used in the early stages of application.

04 Differences Between Knowledge Graphs and Big Data

Knowledge graphs utilize new technologies and methodologies to enhance the efficiency of information conversion into knowledge and its utilization. Both knowledge graphs and big data involve “structuring” and “associating” abstract work. Big data primarily focuses on the structuring of data and data-level associations, while knowledge graphs emphasize the structuring of knowledge and knowledge-level associations.

In knowledge graph technology, knowledge structuring is modeled through a triplet data structure for entities and relationships. When addressing analytical insights, knowledge graphs handle “relationships” more intuitively and efficiently. Their goal is to transfer manual processes to computers to accomplish this work more effectively.

The goal of big data is to transform unstructured data into structured data so that it can be analyzed by computers. In this sense, traditional enterprise big data platforms, data governance, and knowledge graphs can all share the enterprise’s big data.

Understanding Knowledge Graphs ClearlySourceInternet, Copyright belongs to the original author or platform. This account respects originality, and reprinting is intended for sharing. If there are errors in source attribution or violations of your legal rights, please inform us for correction or deletion.

Winter
Break
Course
Program

Understanding Knowledge Graphs Clearly

Understanding Knowledge Graphs Clearly

Understanding Knowledge Graphs Clearly

Leave a Comment