What Is Knowledge Engineering?

Knowledge Engineering is an emerging engineering technology discipline. It arises from the intersection of social science and natural science and the interpenetration of science and engineering. ^[1]

Is a new one

The concept of knowledge engineering is 1977

Knowledge engineering process includes 5 activities

(1)

Basic theoretical research, such as knowledge classification, structure and utility, knowledge representation, knowledge acquisition and

The process of knowledge information processing and related technologies. The term "knowledge engineering" is used by Stanford University in the United States

Comparison of Knowledge Management and Knowledge Engineering

1. Analysis of the ratio of learning to science in knowledge management and knowledge engineering ^[5]

The research on knowledge management is very hot, and the concept of knowledge management is also very many. Different concept cognitions reflect different schools. Earle analyzed the seven schools of knowledge management, including the system school, the cartography school, the engineering school, the business school, the organization school, the space school, and the strategic school. Benny divides knowledge management into communication, analysis, asset management, process, development and innovation. Zuo Meiyun summarizes knowledge management research into three schools, including the technical school, the behavior school, and the comprehensive school; Wu Jinxi summarizes the four university schools of knowledge management, including the IT technology school, the organizational behavior school, the strategic management school, and the knowledge engineering school. Sheng Xiaoping summarized eight schools, including epistemology school, strategic management school, knowledge innovation school, space school, information technology school, organizational behavior school, knowledge engineering school, and comprehensive school. These schools are generally divided into two categories, one is the enterprise knowledge management school, which focuses on the conversion and sharing of knowledge, focuses on the explicitization of tacit knowledge, and aims at improving the core competitiveness of enterprises, such as literature, which belongs to management science. The second category is the library knowledge management school, which aims at the ordering of knowledge and improves the orderliness of knowledge organization, thereby improving the level of knowledge service. It belongs to library science. Knowledge management research focuses on the fields of business management, library science and information science. Library knowledge management is divided into two categories, one is knowledge management with the goal of knowledge ordering, and the other is knowledge management with the goal of knowledge sharing and transformation. The former attaches importance to the construction of resources, and the core of management is resources. The latter regards the library as a specific institution for knowledge management, and the core of management is people. However, no matter what school, it is a typical feature of knowledge management to focus on organization and not technology. ^[5]

Research on knowledge engineering in China is focused on the field of computer science and artificial intelligence. For example, researcher Lu Ruyi of the Chinese Academy of Sciences conducted in-depth research on knowledge engineering and knowledge science. From the perspective of logic, he conducted in-depth research on knowledge engineering. Professor Pan Yunhe of Zhejiang University and others used manipulative thinking theory to study the conversion between semantic knowledge and graphic images. He also studied KQML ( Knowledge Query and Manipulation Language) Knowledge operation. Regardless of which school of knowledge engineering, focusing on technology but not organization is a common feature of knowledge engineering. The fundamental purpose of knowledge engineering is to solve the problem of knowledge acquisition in artificial intelligence, especially expert systems. ^[5]

It is not advisable to include knowledge engineering in knowledge management or knowledge management. Knowledge management pays more attention to human factors and belongs to the management category; knowledge engineering pays more attention to the realization of technology and belongs to the technical category. Therefore, no matter in terms of goals, processing methods and methods, application fields, subject categories, etc., knowledge management and knowledge engineering are very different. They are two completely different research fields.

2. Comparative analysis of core content of knowledge management and knowledge engineering ^[5]

Knowledge management mainly includes knowledge transformation and knowledge ordering. Knowledge transformation is a process of knowledge sharing, and knowledge sharing is also a prerequisite for knowledge transformation. The knowledge transformation in knowledge management includes four aspects, from the tacit knowledge to the tacit knowledge socialization process; from the tacit knowledge to the explicit knowledge externalization process; from the explicit knowledge to the explicit knowledge synthesis process; from The internalization process of explicit knowledge to tacit knowledge, these transformations are mainly the changes in the form of knowledge existence and the attached subject. Knowledge organization in knowledge management is mainly based on the ordering of knowledge, including operations such as classification, retrieval, and sorting. The traditional knowledge organization uses the method of document unit, according to the structural pattern in the retrieval language, adopts the classification method, title method, unit lexical method, keyword method, and thesaurus method. Based on these methods, various catalogs, indexes, and documents are compiled Wait. It is unrealistic to use keywords or topic words to transform knowledge from a document unit at the physical level to a knowledge unit at the cognitive level, because a word unit is not sufficient to reflect knowledge completely, and it should at least be sentence-level. The knowledge map reveals the source of knowledge and the relationship between knowledge. It points to knowledge without including knowledge itself. It is a guide rather than a collection of knowledge. So the knowledge map is actually an index of knowledge. But knowledge maps do not have the basic attribute of geographic coordinates. ^[5]

Knowledge management is not only a matter of acquiring, organizing, and retrieving information, but also involves data mining, text clustering, databases, and documents. The close correlation between knowledge and human cognition determines that knowledge management is positioned on intricate and structured content processing. Knowledge organization in knowledge management describes knowledge in a natural language. The granularity of knowledge is not uniform, ranging from large to small, from a document to a knowledge point.

Knowledge engineering is a discipline that uses knowledge to process knowledge, borrows engineering ideas, and uses the principles, methods, and technologies of artificial intelligence to design, construct, and maintain knowledge-based systems. People generally think that knowledge engineering is an application branch of artificial intelligence . Knowledge engineering includes three processes: knowledge acquisition, knowledge representation and knowledge utilization. There are three ways to acquire knowledge: non-automatic knowledge acquisition, knowledge extraction, and machine learning knowledge. Non-automatic knowledge acquisition is performed by knowledge engineers by reading relevant literature or communicating with domain experts to obtain original knowledge and analyze, summarize, and organize it, and form knowledge items expressed in natural language and input them into the database. Knowledge extraction is to identify, understand, filter, and format the knowledge contained in text documents, extract each knowledge point of the document, and store it in a knowledge base in a certain form. Machine learning knowledge directly senses the external world through machine vision, hearing, etc., inputs natural information, acquires perceptual and rational knowledge, or deduces and summarizes new knowledge from existing knowledge or examples based on system operating experience, and adds knowledge In the library. The efficiency of non-automatic knowledge acquisition is low, and machine learning knowledge is too difficult. Knowledge extraction is the most effective way to acquire knowledge. Knowledge extraction is one of the three ways of knowledge acquisition. Knowledge acquisition is one of the three major steps of knowledge engineering (including knowledge acquisition, knowledge representation and knowledge utilization). Therefore, knowledge extraction is the most effective way of knowledge engineering.

The emergence of ontology research has injected new vitality into the research of knowledge engineering, but what role does ontology play in knowledge engineering? Ontology is a way of representing knowledge? Ontology engineering will replace knowledge engineering. Ontology It is a kind of sufficiently complex vocabulary. Although it is possible to solve many problems with the ontology, how to obtain the ontology is still a major difficulty, just as knowledge acquisition has always been the bottleneck of artificial intelligence. There are three ways to obtain the ontology: manual construction, vocabulary conversion, and automatic acquisition. Ontology is a kind of epistemology. Ontology representation language is more specific and more operable than knowledge representation language.

There are nine methods of knowledge representation, namely: predicate logical representation, production representation, frame representation, script representation, process representation, semantic web representation, Petri net representation, and object-oriented representation. Different Knowledge types use different representation methods. For example, rules are suitable for production representation, experimental processes are suitable for process representation, concept features are suitable for object-oriented representation, and relationships between concepts are suitable for semantic web representation. Knowledge use includes knowledge Search and knowledge reasoning. Knowledge search determines what kind of knowledge is required under what circumstances, and whether the searched knowledge meets the current needs. After finding the appropriate knowledge, perform reasoning to get the results.

3. Comparative analysis of peripheral elements of knowledge management and knowledge engineering ^[5]

Knowledge management focuses on knowledge transfer from person to person, while knowledge engineering pays more attention to the operation of knowledge itself. The goal of Knowledge Management (KM) is to establish a knowledge base for human use, while the goal of Knowledge Engineering (KE) is to establish a knowledge base for computer use. The core of knowledge management is the ordering of disordered knowledge, explicitization of tacit knowledge, and generalization of knowledge ontology. Knowledge engineering mainly involves the three major processes of knowledge acquisition, knowledge representation and knowledge utilization. Among them, knowledge acquisition has always been the difficulty of knowledge engineering and the bottleneck of artificial intelligence. From the perspective of management science, knowledge management focuses on the explicitization of tacit knowledge, which is not very technical, and the result of management is mainly for human use. Knowledge engineering is from the perspective of engineering, focusing on knowledge acquisition and knowledge representation. It is very technical, and the results can be used by both humans and machines, mainly machines. Knowledge management revolves around people. The users of knowledge management are people, computers are auxiliary management tools, and people are the ontology in knowledge management. Knowledge engineering revolves around computers. The users of knowledge engineering are computers (systems), people and computers are tools for implementation, and computers are the ontology in knowledge engineering.

Knowledge organization in knowledge engineering describes knowledge in a computer-understandable manner, the granularity of knowledge is relatively small, and the knowledge element (or knowledge point) is used as a unit. Such as the knowledge base CYC, chess records used by IBM dark blue computers. The link between knowledge elements and knowledge elements constitutes a knowledge chain. There are three main uses of the concept of a knowledge chain. The first usage is the link between knowledge elements and knowledge, such as the links formed between multiple knowledge elements used in the process of knowledge discovery. The second usage is document knowledge linking, such as Tsinghua Tongfang's China HowNet, the knowledge link portal of Wanfang Data, and the granularity between different knowledge nodes is very different, such as from author to document and from author to institution. Link, knowledge link cannot directly perform knowledge discovery. The third usage is the action chain formed by the processing of knowledge, such as the chain formed by the processes of knowledge acquisition, knowledge reorganization, knowledge storage, and knowledge transmission. The first kind of knowledge chain emphasizes the countability of knowledge, the category of knowledge nodes in the second kind of knowledge chain is larger, and the knowledge in the third kind of knowledge chain can be large or small. The first two kinds of knowledge chains are the links formed between different knowledge elements, which are the relationships between elements, and the third kind of knowledge chains are the chains formed by the operations performed around a single knowledge element, which are between actions and actions. relationship. Knowledge grid is different from knowledge network. Grid is a computing technology that makes full use of network resources. The fundamental problem solved by this technology is computing resources (including storage and operations, especially operations). Different knowledge meta-logics are put together to form a grid.

4. Analysis on the development trend of knowledge management and knowledge engineering ^[5]

Knowledge management should aim at expliciting tacit knowledge, ordering unordered knowledge, and generalizing knowledge ontology. Knowledge engineering, which aims to establish an object-oriented knowledge base and a logical proposition knowledge base, describe the things in nature in the most natural way, and describe the rules between things in a way that people can recognize and understand in order to be able to effectively To solve problems such as information flooding and information explosion, we can filter and screen duplicate information to obtain clear and orderly knowledge that can best reflect the nature and natural laws of things. Han Kesong and others believe that knowledge discovery is the highest level of knowledge management: the primary stage is the knowledge base (you know what you have), the intermediate stage is knowledge sharing (you know you have nothing), and the advanced stage is knowledge discovery (you don't know you have what).

Knowledge engineering is also moving in the direction of clear knowledge expression, orderly data organization, and content storage ontology. With the new development of natural language processing and the mature application of object-oriented methods, especially the introduction of ontological ideas, knowledge The development of the project indicates the direction and injects new vitality into the implementation of the knowledge project. The way of knowledge representation is relatively mature and can cover most types of knowledge. The key to knowledge engineering is still knowledge acquisition. Non-automatic knowledge acquisition is too slow to meet engineering needs. Full-automatic knowledge acquisition is too difficult. Before natural language processing can achieve major breakthroughs, it is also difficult to implement engineering. Therefore, the method of semi-automatic knowledge acquisition is more operable, constructing part of the knowledge base and learning rules, then analyzing the corpus, extracting while analyzing, and then improving the rules to continuously improve the algorithm and enrich the knowledge base.

5. Future development of knowledge and technology ^[5]

Knowledge management does not include all about knowledge processing, and knowledge engineering does not include all of knowledge processing. Knowledge management and knowledge engineering have their own division of responsibilities, each with its own responsibility. If knowledge management and knowledge engineering overlap, it is in the construction of the knowledge base. The knowledge base constructed in knowledge management is generally in natural language, while the knowledge base constructed in knowledge engineering is generally in artificial language. Although the representation is different from the use of objects, building a knowledge base is a key part. The premise of knowledge base construction is knowledge acquisition. The effective way of knowledge acquisition is knowledge extraction. The goal of knowledge extraction is to form a knowledge base with knowledge elements as a unit. Knowledge acquisition is a key problem to be solved by knowledge engineering. Therefore, knowledge extraction is a key part of knowledge engineering. On the other hand, knowledge extraction realizes a kind of knowledge ordering, which organizes knowledge at different granularities, and knowledge organization is a key part of knowledge management. Therefore, knowledge extraction is not only conducive to the knowledge acquisition of knowledge engineering, but also to the knowledge organization of knowledge management. Both knowledge management and knowledge engineering involve knowledge organization.

Whether it is knowledge management or knowledge engineering, acquiring knowledge through analysis is bound to be the focus of research. After acquiring knowledge, the analysis of knowledge itself and the relationship between knowledge will inevitably become a new research hotspot. Knowledge acquisition through analysis mainly refers to knowledge extraction. The analysis of knowledge itself includes knowledge representation, knowledge transformation and knowledge mapping, and knowledge between The relationship analysis is embodied in knowledge mining and knowledge discovery. Intelligence scientists are somewhere between knowledge management and knowledge engineering.

The management of people is not as good as a management scientist, and the study of computers is not as good as a computer scientist. Therefore, the position of information science in knowledge management is more oriented to knowledge services. Information scientists are taking the crossroads of knowledge management and knowledge engineering, both knowledge ordering and knowledge transformation. Pure information may generate intelligence, and pure knowledge is difficult to generate intelligence. Most information is the result of the combined effect of information and knowledge, that is, the analysis of new information through knowledge, the analysis of situations and opportunities, and the provision of solutions for decision-making. It is the essence of intelligence activities. Therefore, how to acquire knowledge and use knowledge effectively becomes the key to knowledge processing. There are many technologies involved in knowledge processing, including knowledge organization, knowledge management, knowledge service, knowledge discovery, knowledge mining, knowledge retrieval, etc., but the core of knowledge processing is knowledge acquisition, representation and utilization. Some of these processes are manual, such as explicit tacit knowledge; some are automated by computers, such as extracting knowledge from the literature; and some are human-computer interactions, such as knowledge representation. Solving the coming and going of knowledge and the intermediate analysis process are the three major processes of knowledge processing, and they are also the core. Knowledge processing must be based on the summary of the characteristics of the academic literature, with academic literature as the main processing object, and appropriate use of natural language processing technology to analyze the content structure and semantic expression of the literature, and use knowledge elements as processing units to extract and organize And use it to realize the automatic processing of knowledge, improve the knowledge dimension and intelligent components of the analysis process, and promote the rapid development of library and information science