- What is knowledge
Some readers may say, isn't knowledge what we learn in school and at work, but what else is there to talk about? In fact, you only understand part of it correctly, and let me go into details. Let's start with a picture.
As can be seen from the above figure, knowledge is the processing of data to form logical information, and after precipitation and structuring into valuable information.
In the digital age, humans have a huge amount of data, does this mean that humans can use endless knowledge anytime, anywhere? The answer is no.
Knowledge is the result of people's practical knowledge of the objective world (including human beings) and includes facts, information, descriptions, and skills acquired in education and practice.
Knowledge is the systematic understanding that human beings have obtained from various ways that has been improved, summarized and condensed, the knowledge and understanding after the processing of information, and the results of the condensation and summary of data and information. Knowledge is valuable information.
The courses we study in school and the relevant content in our work, the daily short videos, and the articles we read are just to obtain information, and the information is only called knowledge when it becomes useful to us after being summarized and refined.
To give a very simple example, for example, we learn 1+1=2 in school, this is information, when we buy groceries, we pay 2 yuan to the boss, at this time 1+1=2 is knowledge.
So what is a graph, the following are graphs.
To put it simply, a diagram is a structure used to represent the interconnection between some things (objects) and other things. A diagram usually consists of a number of nodes (Vertice or Node) and edges that connect those nodes.
- What is a knowledge graph
Literally, a knowledge graph is a representation of knowledge in the form of a graph. The nodes in the diagram represent semantic entities or concepts, and the edges represent the various semantic relationships between nodes.
Professional definition 1: Knowledge Graph (KG) is a technology that describes things and the connections between things in the objective world in a structured form, and expresses the information of the Internet into a model closer to the human cognitive world through big data and artificial intelligence technology.
Professional definition 2: A knowledge graph is essentially a semantic network, consisting of nodes (points) and edges, each node represents the "entity" existing in the real world, and each edge represents the relationship between entities. It is to improve the quality of the answers returned by search engines and the efficiency of user queries, and to enhance the efficiency and quality of semantic search through the construction of a knowledge base.
To put it simply, a knowledge graph is to display things and relationships in the real world in the form of correlation graphs, so that people can see the relationship between them at a glance. Take a chestnut.
As shown in the figure, we can see that if there is a relationship between two nodes in China and Beijing, they will be connected by an undirected edge, then these two nodes are called Entities, and the edge between them is called Relationship. Let's start with the concept of knowledge graphs. Entities: correspond to specific things in the real world, such as China, Zhang San, Tiger, Software Engineer, Hefei, Pencil, etc. Relationship: It is used to express a certain connection between different entities, and different entities are connected to each other through relationships, such as competition, cooperative relations, hostile relationships, etc., father-son relationships, husband and wife relationships, classmate relationships, etc. Attributes: The depiction of abstract aspects of entities or relationships, such as a person's age, height, weight, etc., and relationship attributes such as the time of marriage of the husband and wife, the school of classmates, etc.
The knowledge graph is composed of the triples of Entity, Relationship and Property, which is abstractly described as: Entity-Relationship-Entity.
- The development milestone of the knowledge graph
Initially, the concept of knowledge graph was closely related to early AI research, especially the rise of expert systems and knowledge engineering laid the foundation for the development of knowledge graph. In 1960, three main schools of thought emerged in the early development of artificial intelligence: symbolism, connection, and behavior. Semiotics focused on using computer symbols to represent the knowledge in the human brain to simulate the process of human thinking and reasoning, while connectionists focused on simulating the physiological structure of the human brain, thus developing artificial neural networks. The behavioral school emphasizes the simulation of human behavior. The research of these three branches provides a theoretical and technical basis for the development of knowledge graphs.
The connection school, represented by deep learning, mainly solves the problem of perception and also leads the development boom of artificial intelligence. But higher-level cognitive domains, such as natural language understanding, reasoning, and association, still need the help of semiotics.
Knowledge graph is a representative of the semiotic school, which can help us build more knowledgeable artificial intelligence, so as to improve the functions of robot reasoning, understanding, association and so on. This cannot be achieved with big data and deep learning alone. Professor Geoffrey Hinton of the University of Toronto also proposed that one of the future development directions of artificial intelligence is the in-depth integration of deep neural networks and symbolic artificial intelligence.
In 2012, with the rapid development of big data, cloud computing and artificial intelligence technology, the research and application of knowledge graph ushered in a new climax. Google officially proposed the concept of "knowledge graph", which aims to create an intelligent search service that can provide richer and more accurate information by integrating open data on the Internet.
So far, knowledge graphs have been widely used, including but not limited to semantic search, question answering systems, intelligent decision-making, etc. It can help the network to be more intelligent and closer to human cognitive thinking. Knowledge graphs can also be applied to natural language processing, recommendation systems, and other fields. Knowledge graph is an important cornerstone for the realization of Artificial General Intelligence (AGI). In the process of leapfrogging from perception to cognition, the construction of large-scale high-quality knowledge graph is an important link. When artificial intelligence can understand human knowledge through more structured representations and interconnect them, it will be possible for machines to truly realize cognitive functions such as reasoning and association.
Therefore, the question arises, when AI has all the knowledge of human beings, can it form the ability to think independently? It is not yet possible to draw conclusions, and further research by experts and scholars is needed.
- Knowledge graph application scenarios
Knowledge graphs have a wide range of applications, covering multiple industries and application scenarios.
Typical applications of knowledge graphs
In the field of financial risk control, knowledge graphs can help banks and other financial institutions identify and prevent fraud by building and analyzing the network of relationships between financial entities.
By constructing the knowledge graph model of financial entities such as bank accounts and customers, using graph analysis technology to mine and refine the associated risk characteristics between accounts, and further combining traditional accounting case prevention rules to form an intelligent anti-fraud strategy.
Integrate the bank's internal and external data, deeply explore the correlation between various entities, identify account numbers, customers, enterprise risks, abnormal guarantees, suspected actual controllers, etc., and provide support for the bank's risk prevention and control. The application of knowledge graph is not limited to the three links of pre-loan, loan and post-loan, but also includes multi-dimensional and multi-level risk assessment of the subject to provide more comprehensive and accurate risk warning.
For example, in credit scenarios, knowledge graphs can be applied to the identification and management of credit risks. At the same time, in the face of the rampant development of the black credit industry, knowledge graph and graph mining technology can identify malicious loan fraud behaviors such as fraud gangs and black industry intermediaries, and effectively prevent malicious loan fraud such as gang fraud, black industry intermediaries, and agency packaging.
The application of knowledge graph in the field of financial risk control mainly helps financial institutions improve the efficiency and accuracy of risk management by building and analyzing the relationship network between financial entities, combining big data technology and graph analysis technology, to achieve early identification and prevention of fraud.
Its applications in the healthcare field are mainly reflected in assisting diagnosis and treatment decision-making, and specific cases include:
Auxiliary diagnosis: Provide diagnostic recommendations to doctors by analyzing the patient's symptoms, medical history, examination results and other information. This application can help doctors diagnose diseases more accurately and improve the accuracy of diagnosis. With the development of artificial intelligence technology, we can recently see a video of artificial intelligence replacing traditional Chinese medicine (TCM) consultation and prescribing traditional Chinese medicine, which shows that the accuracy of artificial intelligence robot diagnosis and the similarity between the prescription prescribed by the old Chinese medicine doctor who has worked for many years is about 80-90%, which is the best proof.
Image-assisted knowledge graph: The construction of image-assisted knowledge graph, as well as the diagnosis model of rare disease knowledge graph, and the application of artificial intelligence algorithms in medical decision support, prove the effectiveness of knowledge graph reasoning in the medical field.
Medical Knowledge Graph Database Based on Disease Diagnosis-related Grouping: This paper elaborates the construction method of medical knowledge graph database based on disease diagnosis-related grouping, and looks forward to its application prospect in health education, auxiliary diagnosis decision-making system, and optimization of disease diagnosis-related groupers.
Gastric cancer tumor knowledge graph: The construction and application of gastric cancer tumor knowledge graph for clinical decision support, including gastric cancer auxiliary clinical staging and diagnosis and treatment decision support, can provide support for clinical departments related to gastric cancer diagnosis and treatment.
Baidu Lingyi Zhihui Clinical Auxiliary Decision-making System CDSS: In addition to the disease diagnosis function based on the evidence-based medicine framework, it also supports a number of AI-assisted diagnosis functions such as intelligent risk control, similar medical record retrieval, treatment plan recommendation, and medical knowledge retrieval, so as to improve the diagnosis and treatment capabilities of doctors in an all-round and multi-dimensional way
Interactive guidance with natural language understanding and knowledge graph: Automatically process patient interaction information for doctors' clinical diagnosis with medical advice, use natural language understanding to associate user information and semi-automatically build a structured medical knowledge base.
Combination of high-quality clinical medical knowledge base knowledge graph and artificial intelligence: It is one of the core application scenarios of artificial intelligence technology in the medical field to allow computers to "learn" doctors' medical knowledge, simulate doctors' thinking and diagnostic reasoning, and give diagnosis and treatment plans.