Scopus AI Beta: functional analysis and cases [open access report]

Click to download PDF report

Academic databases are a fundamental source for identifying relevant literature in a field of study. Scopus contains more than 90 million records and indexes around 12,000 documents per day. However, this context and the cumulative nature of science itself make it difficult to selectively identify information. In addition, academic database search tools are not very intuitive, and require an iterative and relatively slow process of searching and evaluation. In response to these challenges, Elsevier has launched Scopus AI, currently in its Beta version.

As the product is still under development, the current user experience is not representative of the final product. Scopus AI is an artificial intelligence that generates short synthesis of the documents indexed in the database, based on instructions or prompts. This study examines the interface and the main functions of this tool and explores it on the basis of three case studies. The functional analysis shows that the Scopus AI Beta interface is intuitive and easy to use. Elsevier’s AI tool allows the researcher to obtain an overview of a problem, as well as to identify authors and approaches, in a more agile search session than conventional search. Scopus AI Beta is not a substitute for conventional search in all cases, but it is an accelerator of academic processes. It is a valuable tool for literature reviews, construction of theoretical frameworks and verification of relationships between variables, among other applications that are actually impossible to delimit.

INTRODUCTION

The term artificial intelligence (hereinafter AI) is attributed to John McCarthy, founder of
this field of study. In 1955 the scientist defined artificial intelligence as “the science and
engineering of creating intelligent machines” (McCarthy, 1955). In its beginnings, scientific
studies related to AI were focused on the field of physical sciences, but over time it has
expanded to cover other disciplines.

In fact, the number of scientific publications related to AI has increased exponentially. In 1960 only 14% of the subject areas of the Scopus All Science Journal Classification (ASJC) system featured AI-related publications. However, currently, this figure is higher than 98% (Hajkowicz et al., 2023). In the field of social sciences, recent research has studied how AI constitutes a valuable resource, both for the design of systematized literature reviews and in university teaching (Lopezosa, Codina and Ferrán-Ferrer, 2023; Codina and Garde, 2023).

The academic environment is characterized by large amounts of published research and
diverse databases. These peculiarities make it difficult for researchers to discover valuable
information, despite being a fundamental part of their work. This is especially true for young
researchers, as the advancement of science is a cumulative process. Current tools are limited,
as they do not present direct results, but rather lists of documents, which require significant
amount of time to navigate. Furthermore, it is necessary to apply various inclusion and
exclusion criteria before even approaching a document bank that can respond to a specific
information need. This search approach is necessary in some contexts, but there are others in
which an AI solution using direct responses may be a better solution, as it can help accelerate
some processes.

In this context, Elsevier has developed its own AI for the Scopus database. Scopus AI,
currently in its Beta version, generates evidence syntheses by proposing short texts that
assume direct answers, and can be used through natural language instead of search
equations.

At the same time, Scopus AI Beta is an easy-to-use search tool since, in addition to the
synthesis of evidence, it provides reference lists. Specifically, the user introduces a need for
information as an instruction – or «prompt» in its most widespread meaning – and Scopus AI
Beta generates a response based on the analysis and synthesis of summaries of quality
research published since 2013 (in the current version). That is, it is based on validated research
that has undergone a review process by experts before being published and subsequently
indexed in the database. In this way, and unlike other generative AI, the researcher has the
guarantee that the results consulted have been carried out on highly reliable a priori bases,
although it is advisable to always verify them.

Scopus AI Beta is based on a synthesis structure that can be expanded by iteration until the
user makes the discretionary decision to conclude the interactions, either because they
consider that they have an answer that is already optimal or because the new results no
longer provide notable improvements. It is a tool that generates focused syntheses using
natural language, thus reducing search time and evaluation of results.

(…)


Citation

Aguilera-Cora, Elisenda; Lopezosa, Carlos; Codina, Lluís (2024). Scopus AI Beta: functional analysis and cases. Barcelona: Universitat Pompeu Fabra, Departament de Comunicació, 2023. 46 p. (Serie Editorial DigiDoc. DigiDoc Reports). http://hdl.handle.net/10230/58658



References

  • Aguilera Cora, E., & Codina, L. (2023). Use of Scopus and WoS in literature reviews for doctoral theses: a case study illustration.
  • Aydın, Ö., & Karaarslan, E. (2022). OpenAI ChatGPT generated literature review: Digital twin in healthcare. Available at SSRN 4308687.
  • Codina, L. (2022). Cómo utilizar ChatGPT en el aula con perspectiva ética y pensamiento crítico: una proposición para docentes y educadores, https://www.lluiscodina.com/chatgpt-educadores/
  • Codina, L; Garde, C. (2023). Uso de ChatGPT en la docencia universitaria: fundamentos y propuestas. https://repositori.upf.edu/handle/10230/57015
  • Codina, L., Lopezosa, C., & Freixa Font, P. (2021). Scoping reviews en trabajos académicos en comunicación: frameworks y fuentes. In Larrondo Ureta A, Meso Ayerdi K, Peña Fernández S, editores. Información y Big Data en el sistema híbrido de medios-XIII Congreso Internacional de Ciberperiodismo; 15-17 nov 2021; País Vasco.[Leioa]: Universidad del País Vasco; 2021. p. 67-85.. Universidad del País Vasco.
  • Comisión Europea (2020). Libro blanco sobre la inteligencia artificial: un enfoque europeo orientado a la excelencia y la confianza https://eur-lex.europa.eu/legal-content/ES/TXT/?uri=CELEX:52020DC0065
  • Elsveier. (n.d.). Scopus AI: Change the way you view knowledge  https://www.elsevier.com/products/scopus/scopus-ai
  • Franganillo, J. (2023). La inteligencia artificial generativa y su impacto en la creación de contenidos mediáticos. methaodos. revista de ciencias sociales11(2), 15.
  • Guida, G.; Mauri, G. (1986). Evaluation of natural language processing systems: Issues and approaches. Proceedings of the IEEE. 74 (7): 1026–1035. doi:10.1109/PROC.1986.13580 
  • Hajkowicz, S., Sanderson, C., Karimi, S., Bratanova, A., & Naughtin, C. K. (2023). Artificial intelligence adoption in the physical sciences, natural sciences, life sciences, social sciences and the arts and humanities: A bibliometric analysis of research publications from 1960-2021. Technology in Society74, 102260. https://doi.org/10.1016/j.techsoc.2023.102260
  • King, M. R., & chatGPT. (2023). A Conversation on Artificial Intelligence, Chatbots, and Plagiarism in Higher Education. Cellular and Molecular Bioengineering, 1-2.
  • Kung, T. H., Cheatham, M., Medinilla, A., ChatGPT, Sillos, C., De Leon, L., … & Tseng, V. (2022). Performance of ChatGPT on USMLE: Potential for AI-Assisted Medical Education Using Large Language Models. medRxiv, 2022-12.
  • Llaneras, K.; Rizzi, A.; Álvarez, J. (2023) ChatGPT es solo el principio: la inteligencia artificial se lanza a reorganizar el mundo, elpais.com, https://elpais.com/sociedad/2023-01-29/chatgpt-es-solo-el-principio-la-inteligencia-artificial-se-lanza-a-reorganizar-el-mundo.html
  • Lopezosa, C.; Codina, L.; Ferran-Ferrer, N. (2023) ChatGPT como apoyo a las systematic scoping reviews: Integrando la inteligencia artificial con el framework SALSA Barcelona: Universitat de Barcelona
  • OpenAI (2022) https://openai.com/blog/chatgpt/
  • Scopus. (n.d.). https://scopus.com
  • Shieh, J. (2023). Best practices for prompt engineering with OpenAI API, OpenAI
  • https://help.openai.com/en/articles/6654000-best-practices-for-prompt-engineering-with-openai-api
  • Transformer, C. G. P. T., & Zhavoronkov, A. (2022). Rapamycin in the context of Pascal’s Wager: generative pre-trained transformer perspective. Oncoscience, 9, 82.
  • Van-Dis, E. A. M., Bollen, J., Zuidema, W., van Rooij, R., & Bockting Alo, C. L. (2023). ChatGPT: five priorities for research. Springer Nature, vol.614, 224-226 https://www.nature.com/articles/d41586-023-00288-7
  • Wang, S., Scells, H., Koopman, B., & Zuccon, G. (2023). Can ChatGPT Write a Good Boolean Query for Systematic Review Literature Search?. arXiv preprint arXiv:2302.03495.
  • Xataca (2023) ChatGPT ya aparece como autor de artículos de investigación científica: la debacle de si una IA puede ayudar a hacer ciencia, xataca.com https://bit.ly/3Ibz2C0