Classifying Research Publications
— by Fanny Liu
Introduction
Researchers seek to understand how scientific knowledge is thematically structured. However, this is difficult because both knowledge and research communities are constantly evolving. These shifting structures are usually called fields, specialties, or topics.
Under different classification systems (schemas), a single publication could be classified into areas such as Engineering, Public Health, Environmental and Occupational Health, Human Factors and Ergonomics, Rehabilitation, Medicine, and Health Sciences.

This post briefly introduces three widely used classification schemas, Essential Science Indicators, Scopus All Science Journal Classification, and OpenAlex Topics, using a small case study.
Case study
Data
Articles and reviews authored by scholars in the Department of Data and Systems Engineering at the University of Hong Kong published in 2025 were retrieved from SciVal (Scopus), InCites (Web of Science), and OpenAlex on 10 March 2026. Only publications indexed in these databases were included.
Results and discussion
Essential Science Indicators and Highly Cited Researchers
In the Web of Science Core Collection, 108 publications were indexed and assigned to a research area under the Essential Science Indicators schema. These publications were distributed across 11 research areas. The majority were categorised under Engineering (66.67%), followed by Computer Science (8.33%), and Social Sciences, general (7.41%).
Figure 1 Research areas under the Essential Science Indicators schema
Under this schema, research publications are assigned to a research area based on the journal (InCites Benchmarking & Analytics). Each journal is assigned to one of the 22 research areas. As a special case, publications in journals categorised as multidisciplinary, such as Nature and Science, are assigned to a field based on the cited references. For example, a paper citing mostly publications in Engineering journals will be categorised as Engineering.
This schema is used in Clarivate’s Highly Cited Papers, Hot Papers, and Highly Cited Researchers lists.
In contrast, in the Web of Science Categories schema, journals are assigned to one or more of the approximately 250 subject categories, and publications inherit those categories (InCites Benchmarking & Analytics).
Scopus All Science Journal Classification (ASJC) and Quacquarelli Symonds (QS) Subject Area
In Scopus, the 108 items were categorised into 75 All Science Journal Classification (ASJC) subject areas. The largest shares of ASJC assignments were Industrial and Manufacturing Engineering (6.38%), Management Science and Operations Research (6.09%), Computer Science Applications (5.80%), and Electrical and Electronic Engineering (5.80%).
Figure 2 All Science Journal Classification Codes (ASJC) subject areas
Similar to Essential Science Indicators, Scopus’ All Science Journal Classification (ASJC) schema classifies research based on the journal (Elsevier). However, unlike Essential Science Indicators, where each journal is assigned to only one subject field, under the ASJC schema, each journal is assigned to one or more subject areas.
ASJC codes are matched to Quacquarelli Symonds (QS)’s taxonomy of broad faculty areas and narrow subjects. As one publication can be linked to multiple ASJC subject areas, it can also contribute to multiple QS subjects. For example, an article was categorised under four (ASJC) fields:
- Business and International Management,
- Management of Technology and Innovation,
- Strategy and Management, and
- Management Science and Operations Research,
and contributed to two QS subjects:
- Business & Management Studies, and
- Statistics & Operational Research.
As for the 108 publications, they contributed to 27 QS subjects. The largest share contributed to Computer Science & Information Systems (15.10%), closely followed by Engineering – Mechanical (11.41%) and Engineering – Petroleum (9.73%).
Figure 3 Quacquarelli Symonds (QS) Subject Areas
See Journal Explorer by HKU Libraries (Log-in required. Current HKU staff/students only) for a handy journal discovery tool with QS subject area information.
However, journal-based classification systems have been criticized. Researchers argue against classifying publications based on journal classification schemes, as they have limitations such as difficulties handling multidisciplinary journals and considerably lower accuracy than topic-level taxonomies (Klavans & Boyack, 2017; Shu et al., 2019; Waltman & van Eck, 2012).
OpenAlex Topics
Researchers have developed new methods to delineate fields and identify thematic structures, involving more complex and efficient algorithms for language processing, network analysis, and visualization, supported by the growth of computing capacity (Gläser et al., 2017). Document clustering includes methods such as direct citation, bibliographic coupling (articles A and B both citing article C, so A and B are bibliographically coupled), and co-citation (articles A and B both appearing in the reference list of article C, so A and B are co-cited) (Klavans & Boyack, 2017; Rousseau et al., 2018). A more recent development in topic identification is stochastic topic modelling. It assumes that documents contain mixtures of topics in different proportions and analyses the full text of documents to detect these topics (Gläser et al., 2017).
OpenAlex trained a deep learning model based on the citation network (OpenAlex Support). The model assigns a primary topic to each work based on title, abstract, citations, and journal name. The topics are arranged in a four-level hierarchy: domains (4), fields (26), subfields (254), and topics (~4,500).
The 108 publications were associated with 73 primary topics across 41 subfields and 13 fields. Each publication was assigned to one primary topic. The topics with the highest number of publications were Transportation and Mobility Innovations (6.48%), Reliability and Maintenance Optimization (5.56%), and Supply Chain and Inventory Management (4.63%).
Figure 4 OpenAlex Primary Topic Fields
Similar approaches are also used in other databases. For example, a SciVal topic is a collection of publications with a common intellectual interest based on direct citations (Elsevier). The emerging topics in InCites are identified algorithmically through co-citation analysis (InCites Benchmarking & Analytics).
See Where Research Is Heading: Identifying Emerging Topics for a discussion on the benefits of emerging topic identification and relevant tools.
Conclusion
Researchers have developed classification schemes throughout history, from journal-based mapping to bottom-up topic modelling (Rafols, 2025). Journal-based classifications provide stable but coarse categories, while algorithmic topic mappings offer evolving and more granular representations of research activity. Multiple mappings exist, and some may be more suitable for particular tasks.
Extended Reading
Tracking Research Excellence: HKU’s Performance in Essential Science Indicators (ESI) 2026
Where Research Is Heading: Identifying Emerging Topics
A Comparative Approach to Explore HKU Publication Trends Using Data from OpenAlex and Peer Databases
Declaration of Generative AI Use
I acknowledge the use of Generative AI tools in writing this post.
I used:
- GPT-5.5 to refine the language.
I declare that I reviewed and edited the contents as needed, and take full responsibility for the contents of the post; And the information provided is complete and accurate.
References
Elsevier. SciVal Topics. Retrieved 18 March 2026 from https://www.elsevier.com/products/scival/overview/topics
Elsevier. What are Scopus subject area categories and ASJC codes? Retrieved 18 March 2026 from https://service.elsevier.com/app/answers/detail/a_id/12007/supporthub/scopus/
Gläser, J., Glänzel, W., & Scharnhorst, A. (2017). Same data—different results? Towards a comparative approach to the identification of thematic structures in science. Scientometrics, 111(2), 981–998. https://doi.org/10.1007/s11192-017-2296-z
InCites Benchmarking & Analytics. Emerging Topics Research Area Schema. Retrieved 18 March 2026 from https://incites.zendesk.com/hc/en-gb/articles/31654156088081-Emerging-Topics-Research-Area-Schema
InCites Benchmarking & Analytics. Essential Science Indicators. Retrieved 18 March 2026 from https://incites.zendesk.com/hc/en-gb/articles/22514286958097-Essential-Science-Indicators
InCites Benchmarking & Analytics. Web of Science Research Areas. Retrieved 18 March 2026 from https://incites.zendesk.com/hc/en-gb/articles/22586272202513-Web-of-Science-Research-Areas
Klavans, R., & Boyack, K. W. (2017). Which Type of Citation Analysis Generates the Most Accurate Taxonomy of Scientific and Technical Knowledge? Journal of the Association for Information Science and Technology, 68(4), 984–998. https://doi.org/10.1002/asi.23734
OpenAlex Support. Topics. Retrieved 17 March 2026 from https://help.openalex.org/hc/en-us/articles/24736129405719-Topics
Rafols, I. (2025). Towards multiple ontologies in science mapping. A tribute to Loet Leydesdorff. Scientometrics, 130(6), 3229–3255. https://doi.org/10.1007/s11192-025-05323-0
Rousseau, R., Egghe, L., & Guns, R. (2018). Publication and Citation Analysis. In R. Rousseau, L. Egghe, & R. Guns (Eds.), Becoming Metric-Wise (pp. 99–153). Chandos Publishing. https://doi.org/10.1016/B978-0-08-102474-4.00005-4
Shu, F., Julien, C.-A., Zhang, L., Qiu, J., Zhang, J., & Larivière, V. (2019). Comparing journal and paper level classifications of science. Journal of Informetrics, 13(1), 202–225. https://doi.org/10.1016/j.joi.2018.12.005
Waltman, L., & van Eck, N. J. (2012). A new methodology for constructing a publication-level classification system of science. Journal of the American Society for Information Science and Technology, 63(12), 2378–2392. https://doi.org/10.1002/asi.22748
