Feng Zhiwei, Pei Yajun
Abstract: Terminology research and knowledge engineering are the two pillars of “knowledge infrastructure”. Terminology crystallizes human scientific and technological knowledge in natural language, and throughout the history of science, the emergence of new concepts and the obsolescence of old ones are realized through terminology. Chinese terminology has a long history and its own characteristics. Since the emergence of knowledge engineering in the 1950s, it has gone through the stages of expert systems, the semantic web, knowledge graphs, and large language models. In the future, terminology research and knowledge engineering can mutually reinforce each other and develop together.
Keywords: terminology; knowledge engineering; expert systems; semantic web; knowledge graphs; large language models.
Introduction
Terminology research and knowledge engineering are the two pillars of “knowledge infrastructure”: the former addresses how to name, standardize, and align concepts and terms, and the latter addresses how to represent, acquire, and reason over knowledge. The two are highly complementary in terms of goals, methods, and outputs, and this relationship can be viewed as one between “inside” and “outside”—terminology provides unified and authoritative conceptual labels, while knowledge engineering weaves these concepts into a computable network.
1. Terminology Research
1.1 The Past and Present of Terminology
Terminological activities can be traced back to the 18th century, such as Lavoisier’s chemical nomenclature and Linnaeus’s systematic attempts at biological classification. China had already practiced terminology compilation in “Erya” (尔雅, early Han Dynasty) and “Compendium of Materia Medica” (本草纲目, Ming Dynasty); during the Qing Dynasty, Chinese scholars created a large number of scientific and technological terms by translating Western works.
In 1931, the Russian scholar Lotte proposed the scientific basis of technical terminology, marking the birth of modern terminology. The Austrian scholar E.Wüster established a systematic approach in the 1930s and 1960s, proposing “General Terminology” and laying the disciplinary framework (Wüster2011).
Since the 1970s, computer technology has promoted the construction of terminology databases. In 1975, the first Terminology Knowledge Engineering Symposium was held. China established the China National Committee for Terminology in Science and Technology (CNTERM) in 1985 to systematically promote terminology standardization (Pei2018).
Globally, early terminology was largely studied as a subdiscipline of applied linguistics. Since the 1970s, terminology has remained closely connected to linguistics while increasingly taking on an interdisciplinary character. Terminology studies the components and structure of concepts, their characteristics, methods of definition, conceptual intersections, and concept systems; therefore, it is closely related to logic. Terminology studies the methods and techniques for classifying concepts and conceptual systems; therefore, it is closely related to taxonomy (classification). Terminology studies the relationship between terms and real-world entities; therefore, it is closely related to ontology. Terminology studies linguistic signs and their connections; therefore, it is closely related to semiotics. Terminology studies terminological issues in the processing of scientific and technical literature and uses them as a basis for information retrieval; therefore, it is closely related to information science. Terminology studies computer-based processes such as term storage, retrieval, extraction, and indexing; therefore, it is closely related to computational linguistics and natural language processing (NLP). Terminology also studies the terms of various disciplines; therefore, it is closely related to the natural sciences, the humanities, and the social sciences. In this way, terminology gradually became independent of traditional applied linguistics and developed into a broad and profound discipline involving almost all areas of human knowledge, with its theories and methods continually improving. Not only are linguists concerned with terminology, but experts in almost all disciplines also need to pay attention to terminology. Among the various subdisciplines of applied linguistics, few have attracted as much attention and concern from experts as terminology.
Without terminology, there is no science; without terminology, there is no knowledge. Where there is knowledge, there is terminology. In this sense, terminology and knowledge are closely intertwined, and terminology research and knowledge engineering have become the two pillars of “knowledge infrastructure”.
This shows that terminology plays an important role in the overall structure of modern human knowledge, and that terminology crystallizes scientific knowledge in natural language. In the history of science, the emergence of new concepts and the obsolescence of old concepts are achieved through terminology. Whenever a new scientific concept is generated, a new term must be created to represent it; when an old concept becomes obsolete or is proven wrong in practice, the corresponding term dies out or becomes obsolete and is cited only in discussions of the history of science. Terminology research is becoming increasingly linked to knowledge engineering.
1.2 Trends in Terminology Research
At present, there are the following trends in terminology research:
①Theoretical diversification
From the prescriptive paradigm to the descriptive paradigm, branches such as cognitive terminology (studying conceptual structure) and socioterminology (focusing on language planning) have emerged. In China, Prof. Feng Zhiwei proposed that terminology is the crystallization of human scientific knowledge in natural language and published the monograph “Introduction to Modern Terminology” in 1997 (Feng 2011).
②Deepening technological applications
Corpus linguistics and computational linguistics enable automated term extraction and machine translation, while discourse analysis helps disambiguate polysemous terms through contextual analysis.
③Interdisciplinary research
Globalization has given rise to new terminologies (such as artificial intelligence and biomedical terminologies), and terminology research increasingly requires integrating methods from linguistics, knowledge engineering, and other fields to conduct interdisciplinary research.
1.3 New Challenges to Terminology Research
The development of science and technology has posed the following new challenges to terminology research:
① Accelerate the pace of terminology updates: The emergence of new technological concepts is accelerating, while terminology updates lag behind; therefore, terminology updates must be accelerated.
② Research on cultural differences: Because semantic biases exist in cross-language term correspondence (for example, in translations of Traditional Chinese Medicine terms), terminology research should pay more attention to cultural differences across languages.
③Improve the level of digitization: In the era of big data, terminology research should develop smarter terminology management tools to raise the level of digitization and adapt to the big-data environment.
In summary, terminology research is evolving from standardization to an integrated cognitive–social model, and technology-driven and interdisciplinary collaboration will be key directions for the future.
Modern terminology can be divided into four schools: the German-Austrian school, the Russian school, the Czechoslovak school, and the Canadian-Quebec school. These schools are all schools of Western terminology. These Western schools of terminology have made outstanding contributions to the development of terminology throughout history.
However, these Western schools were formed in Europe and Canada, and during their formation, due to linguistic and geographical barriers, European and Canadian scholars had little understanding of the terminological work carried out in China for thousands of years. As a result, many theories and methods proposed by these Western schools do not fully meet the practical needs of Chinese terminology research.
In fact, long before the Western terminology schools in Europe and Canada took shape, China had already carried out substantial terminological research. Chinese terminological research should be included in the global landscape of terminological studies and recognized, with full confidence, as an important part of world terminology research. In international terminology studies, any tendency to overlook Chinese terminology is one-sided and therefore misguided.
1.4 Characteristics of Terminology Research in China
Prof. Feng Zhiwei believes that compared with European and Canadian terminological research, Chinese terminology research has the following eight characteristics:
① The language is distinctive: For Western terminology schools, Chinese terminology research involves a distantly related language, whereas most Western terminology research concerns closely related languages. Chinese terminology uses Chinese characters, whereas Western terminology uses alphabetic scripts (Latin alphabet, Greek alphabet, Slavic alphabet); the writing systems are fundamentally different. Term formation in closely related languages can often rely on transliteration. However, for distantly related languages written in Chinese characters, term formation cannot rely on transliteration alone.
The borrowing of foreign terms through “alphabetic transcription” often involves a complex process, from phonetic transcription to semantic translation; therefore, the transcription of Chinese terms is more difficult than that of Western terms (Feng 2004).
②The history is long: Chinese terminology research has a long history. As far back as the “Zhou Li” (周礼), there is a record of “Xiangxu” (象胥). The so-called “Xiangxu” refers to an ancient interpreter/translator responsible for translating the languages of foreign countries or ethnic minorities into Chinese, which naturally involved terminology translation. In Chinese history, there have been three major translation climaxes: the translation of Buddhist scriptures from the Eastern Han Dynasty to the Tang and Song dynasties (the first), the translation of Western scientific and technological works in the late Ming and early Qing dynasties (the second), and the translation of Western academic masterpieces from the Opium War to the May Fourth Movement (the third). During these three translation climaxes, substantial terminological work was carried out and many foreign terms were translated, making outstanding contributions to the world’s terminology research. In the period of reform and opening up in contemporary China, it has been necessary to introduce a large number of foreign terms in the natural sciences, social sciences, and engineering, and to translate traditional Chinese ideological and cultural terms into foreign languages (Feng 1992).
③Respect tradition: In addition to studying the translation of foreign terms, it is also necessary to study the translation of traditional Chinese ideological and cultural terms in order to promote Chinese culture to the world. For example, Prof. Feng Zhiwei recently wrote the book “Chinese Characters” in English, which involves translating traditional Chinese script terms such as “oracle script, bronze script, great seal script, small seal script, clerical script, regular script, running script, and cursive script”. These terms are distinctive in traditional Chinese culture, and in terminology translation it is necessary to “seek truth” and express the original meaning accurately so that international readers can understand it; this is very difficult (Feng, Zhang 2017).
China has also made major achievements in translating classical science-and-technology works. Classics such as “Huang Di Nei Jing”(黄帝内经) and “Tian Gong Kai Wu” (天工开物）have now been well translated and published. Respecting the tradition of Chinese thought and culture is a major feature of Chinese terminology.
④ The system is mature: The research and work of Chinese terminology—from institutional arrangements to academic publishing and professional specialization—have developed into a mature system. Since its establishment, the China National Committee for Terminology in Science and Technology (CNTERM) has established more than 70 sub-committees. A total of more than 3,000 experts (including more than 300 academicians) have participated in the validation of more than 300,000 scientific and technical terms in nearly 70 disciplines, initially forming a relatively complete system that covers the natural sciences, engineering, medical sciences, agriculture and forestry sciences, social sciences, and many interdisciplinary fields. This system has played an important role in promoting and safeguarding scientific research, education, and academic exchange in China. Such a mature and complete scientific and technical terminology system is unique in the world and has distinctive characteristics.
⑤Focus on structure: Western terminology attaches great importance to the study of terminological concepts, while Chinese terminology pays special attention to the study of term structure in addition to the study of concepts. As early as 1985, Prof. Feng Zhiwei used a computer to analyze the hierarchical structures of phrase-type terms, and found potential ambiguity in them. He creatively proposed the Potential Ambiguity (PA) theory and introduced it at international conferences. When Chinese scholars analyze term structures with computers, they also find that the number of phrase-type terms in a terminology system is usually significantly higher than that of word-type terms (Feng1991). After carefully studying this phenomenon, Prof. Feng Zhiwei proposed nearly ten measures—such as the frequency of word formation and the average length of terms—studied the relationships among them, and used mathematical methods to describe them. He ultimately formulated the economic law of term formation and used the FEL formula to express it; this work was published in the top journal “Chinese Social Sciences” in 1988 (Feng1988). He presented this important finding to the international academic community. Ten years later(1998), the International Conference on Computational Linguistics organized the world's first symposium on computational terminology. Research on computational terminology in China far exceeds international research in this area, representing an important contribution by Chinese scholars to the theory of world terminology. Subsequently, Chinese scholars proposed the “systematic economic rate of terminology translation” for cross-language situations, which further extends and deepens this theory. This research style that focuses on term structure is distinctive in international terminology (Feng 2008).
⑥Function-oriented: Traditional terminology has made great achievements in term standardization, static studies of terms, and qualitative research. However, it has often overlooked descriptive research, dynamic studies of terms, and quantitative research. Chinese terminology combines qualitative and quantitative approaches, breaks through the limitations of traditional terminology, and reflects the functionalist tendency in contemporary linguistic research (Feng1995).
⑦Data-driven: As early as the 1980s, Chinese scholars used computers to build large-scale terminology databases in machinery manufacturing, the chemical industry, agriculture, and other disciplines, which provided convenience for terminology retrieval and terminology translation in these fields (Feng 1992).
⑧Fruitfulresults: Chinese terminologists have carried out serious theoretical exploration, written the monograph with Chinese characteristics “Introduction to Modern Terminology”, published the specialized journal “China Terminology”, compiled more than 70 discipline-specific terminology dictionaries, established online terminology websites, and achieved fruitful results. This is also rare in international terminological research (Pei 2018).
The above are the eight characteristics of Chinese terminology research. Obviously, China’s efforts have overcome limitations of the German-Austrian, Russian, Czechoslovak, and Canadian-Quebec schools and have developed distinctive characteristics. These characteristics can be attributed to structural functionalism in terminology research, which has in effect formed a Chinese school of terminology. Compared with the four major terminological schools in the world, the Chinese school of terminology is no less significant. The Chinese school of terminology can be regarded as the structural functional school in terminology（Felber2011）.
Terminology is the crystallization of human scientific knowledge in natural language, and terminology research is closely related to knowledge engineering (Jie, Feng 2009).
2. The Synergistic Relationship Between Terminology Research and Knowledge
Engineering
2.1 Four Stages of Knowledge Engineering Development
Since the emergence of Knowledge Engineering (KE) in the 1950s, it has roughly gone through four stages:
① Expert systems stage;
② Semantic web stage;
③ Knowledge graph stage;
④ Large language model stage.
In recent years, knowledge engineering has pursued research on temporal knowledge graphs, multimodal knowledge extraction, retrieval-augmented generation (RAG), and agent applications. With the development of national projects such as real-time data-driven platforms and intellectual-property public service platforms, knowledge engineering is accelerating toward a new stage—one that is dynamic, explainable, and autonomous.
2.2 Synergistic Relationship Between Terminology Research and Knowledge Engineering
The following illustrates the synergistic relationship between terminology research and knowledge engineering in six aspects.
① The common goal of terminology research and knowledge engineering is to build a computable model of the world: Terminology is pursued through standardization, definitions, and multilingual control. It represents human natural-language descriptions of entities/attributes as “unique, unambiguous” conceptual symbols. Knowledge engineering further formalizes these symbols into entities, classes, and relationships, and then uses knowledge graphs or ontologies for computer reasoning. Terminology research helps eliminate homonyms and synonyms, while knowledge engineering helps eliminate heterogeneous models. Together, they support the effective application of trusted artificial intelligence and knowledge-based systems.
② Terminology research and knowledge engineering methodologies complement each other: Knowledge engineering requires appropriate “conceptual granularity”, and terminology can provide expert-approved term concepts and definitions, mapping them directly to classes (owl:Class) and attribute relationships (owl:DatatypeProperty) in OWL.
③ Terminology research requires “context” and “usage examples”, while knowledge engineering can extract entity co-occurrence, semantic relations, and contextual information from massive corpora, providing empirical evidence for term selection, updates, and the discovery of synonymous terms.
For example, ISO 18650 “Architectural Information Modeling Terminology” defines more than 1,200 concepts. When constructing an architectural knowledge graph, these terms can be used directly as suffixes of entity URIs, enabling standards-to-graph integration.
④ Terms provide “node labels” to a knowledge graph, while knowledge engineering provides “edges” and “rules”, thereby constituting the ontology of domain knowledge. For example, the W3C Simple Knowledge Organization System (SKOS) serves as an official “term–ontology bridge”: it uses skos:prefLabel, skos:altLabel, and skos:broader to handle preferred terms, synonyms, and broader terms, thereby elevating a traditional termbase into a lightweight ontological knowledge system. For instance, a “Meteorological Terminology Knowledge Graph” can take 40,000 national standard terms from the China Meteorological Administration as core nodes and extract another 1.2 million weather-disaster-crop relations to support disaster early warning. A “TCM (Traditional Chinese Medicine) Knowledge Graph” can be built on “TCM Terms”, mapping 3,800 syndrome terms into categories and combining relationships among prescriptions and medicinal materials to establish a domain knowledge ontology.
⑤ Terminology and knowledge graphs can work together in technical processes to model knowledge ontologies based on terms. Term definitions provide “class scope”, helping avoid a modeler's subjective assumptions and improving the level of terminology standardization.
⑥ Knowledge graphs can feed back into terminology research: By statistically estimating the co-occurrence strength of edges in a knowledge graph, terminologists can identify which synonymous terms are active and which concepts are becoming obsolete. This enables dynamic maintenance of terminology.
In standards at home and abroad, increasing attention has been paid to combining terminology research with knowledge engineering. Both ISO 704 “Working Principles for Terminology” and ISO 23386 “Model of Architectural Terminology” explicitly recommend RDF/OWL to express terms so that they can be processed by computers. China’s national standard GB/T 10112-2019 “Principles and Methods of Terminology Work” adds an appendix on “Interconnection of Terminology and Knowledge Systems”, which proposes requirements for terminology URIs, version control, and persistent resolution, directly connecting with knowledge engineering. The 2025 edition of the “Guidelines for Artificial Intelligence Knowledge Engineering” lists “terminology standardization” as the first activity domain of knowledge modeling, alongside data cleaning and ontology design (Feng 2025).
3. Future Development Trends
In the future, terminology research and knowledge engineering can promote each other’s development, and key trends include:
① Large language models have a high hallucination rate when generating text. Using termbase-constrained decoding that allows only standard concepts to be generated can significantly reduce hallucinations.
② Neural networks can be used to extract candidate terms, which are then confirmed by experts to achieve “discovery, standardization, and reasoning”.
③ In the “Belt and Road Initiative”, multilingual terminology centers and knowledge graphs can work together to support risk reasoning for multilingual, cross-border supply chains.
Conclusion
Terminology research provides an “authoritative conceptual dictionary”, and knowledge engineering provides a “computable network and reasoning engine”. The former helps the machine “be correct”, and the latter enables the machine to “think deeply”. Only through the integration of terminology and knowledge engineering can we build a truly credible, explainable, and evolvable domain knowledge system—one that is also a key prerequisite for implementing knowledge-driven AI and automatic Q&A systems (Feng 2025).
Works Cited
Felber, Helmut. Terminology, Epistemology and Knowledge Technology. Translated by Bihua Qiu, reviewed by Zhiwei Feng, Commercial Press, 2011.
Feng, Zhiwei. “An Emerging Terminology Discipline—Computational Terminology.” Terminology Standardization and Information Technology, no. 4, 2008, pp. 18–22.
———. Introduction to Modern Terminology. Revised ed., Commercial Press, 2011.
———. “Language Resources and Language Technology in China.” Proceedings of TELRI Seminar, Tihany, 1995.
———. Natural Language Processing in the Era of Big Data. Commercial Press, 2025.
———. “On Potential Ambiguity in Chinese Terminology.” Proceedings of TSTT’91, Beijing, 1991.
———. “Semantic Loanwords and Phonemic Loanwords in Chinese Language.”
Proceedings of 11th International Symposium of NIJLA, Tokyo, Mar. 2004, pp. 200–229.
———. “A Terminological Databank in Chinese Language—GLOT-C.” Proceedings of ICCIP’92, Beijing, 1992.
Feng, Zhiwei, and Zhan Hongwei. Chinese Characters. Foreign Language Teaching and Research Press, 2017.
Jie, Chunyu, and Feng Zhiwei. “Definition of Terms Based on Knowledge Ontology.” Terminology Standardization and Information Technology, no. 2, 2009, pp. 10–15.
Pei, Yajun. “Attaching Importance to Local Characteristics and Establishing a Chinese School of Terminology Research.” China Terminology, no. 4, 2018, pp. 10–12.
Wüster, Eugen. Introduction to General Terminology and Terminology Lexicography. Translated by Bihua Qiu, reviewed by Zhiwei Feng, Commercial Press, 2011.
The Authors
Feng Zhiwei, Professor at Xinjiang University, (Urumqi, 830046) and Heilongjiang University (Harbin, 150080).
Email: zwfengde2020@163.com
Pei Yajun,Research Fellow and Executive Deputy Chairman of the China National Committee for Terminology in Science and Technology (CNTERM), where he also serves as Director of its Executive Office. Currently, he is vice president to China Association for Lexicography, deputy chairman of the Terminology Theory and Application Committee, SAC/TC62, and associate editor-in-chief of Chinese Science & Technology Translators Journal. His principal research interests include terminology management, natural language processing, knowledge graphs, data governance and open data sharing. He has led or participated in nearly 10 projects funded by the National Natural Science Foundation of China and the National Social Science Fund of China.
Email: peiyj@cnterm.cn