The secondary based database was collected from the digital library of Harcourt Butler Technical University Kanpur (U.P.), in between the time period of July 2024 to October 2024.
Over the past two decades, Scopus has emerged as a leading database for abstracts and citations, providing quantitative analyses of academic literature. The field of bibliometrics analysis has been enhanced through various tools, including citation counts, keyword analysis, co-citation counts, co-occurrence measures and bibliographic coupling (
Lwesya and Achanta, 2022). This paper aims to develop techniques for extracting data from both published and unpublished articles using VOSviewer for visualisation and Biblioshiny applications. Researchers prefer these applications because they offer a secure web interface for data importing, conversion, gathering and filtering from sources like Scopus. The primary focus is on identifying research papers related to the keyword phrase “Artificial Intelligence” within the multidisciplinary databases of Scopus.
The systematic literature review (SLR) involves three main steps: Data collection (which includes data loading and conversion), analysis and the development of descriptive and network matrices (co-citation, collaboration and co-occurrence). The final step is visualisation, which enables the creation of result maps. Relevant articles cited in the SLR contribute to a bibliometrics overview. Co-citation is defined as the occurrence of two documents being cited together by other documents (
Small, 1973). The greater the number of co-citations that two documents receive, the stronger their co-citation relationship, indicating a higher likelihood of semantic connection. In this context, counting the frequency of keyword appearances allows for the analysis of research hotspots within disciplines (
Huai and Chai, 2016). Co-authorship is a common practice in academia; thus, papers with multiple authors serve as indicators of collaborative activity within a field (
Wagner and Leydesdorff, 2005). Table 1 shows the bibliometrix information of database.
Synthesis and analysis
The purpose of this paper is to conduct a descriptive and quantitative analysis of the role of Artificial Intelligence in dairy farming over the past twenty years, specifically from 2000 to 2020. Scopus, a citation and abstract platform established in 2004 and owned by Elsevier, has compiled an extensive collection of bibliographic databases, citations and references related to scientific publications across various fields, including science, technology, humanities and social sciences.Subsequently, each of the aforementioned categories are thoroughly analysed using the following elements (1) The document type, (2) Author impact analysis (3) Citation analysis (4) Trend analysis , (5) Word frequency over period of time (6) Word cloud (7) Co-occurence and (8) Thematic analysis.
Fig 1 shows author local impact by H-index and Fig 2 shows most cited authors in documents using vosviewer.
Author impact analysis
Refer to Fig 1, among the thirty authors in the database, the top ten authors who contributed the most to research on AI and modernisation in dairy farming have been identified. The complete publications of the three most productive authors during this period are from India, the United Kingdom and Italy. Researchers previously observed significant contributors who collaborated on multiple documents to identify the shared objectives of their partnerships. This approach is the most common method of investigating academic relationships in systematic research. The study was also conducted with the assistance of Biblioshiny. It was found that there are several strong co-author associations, with some pairs having collaborated on four publications together see in Fig 3 Cited Analysis for most cited countries.
Citation analysis
The term “citation,” also known as a bibliographic reference, refers to a formal acknowledgment of sources in research. Citation analysis is the study of how these references appear in bibliographies, lists or catalogs, particularly from sources outside the original context. This analysis is a commonly used bibliometric method that involves constructing citation graphs and networks, allowing researchers to explore the impact within their academic fields. Citations serve as a primary link between two documents: the one being cited and the one doing the citing. They represent a unique relationship between the cited and citing papers, establishing a connection between authors whose work can be measured through literature. In citation analysis, the commonalities among associations are based on subject areas, methodologies used and research domains. The main reasons for citing documents include acknowledging previous work and providing critical commentary on it. In this paper explains the most cited top ten countries in publishing articles on artificial intelligence in dairy farming. The top most is Australia with forty five citations in this domain (Fig 3).
Trend analysis
The trend analysis is illustrated in Fig 4, which shows the distribution of publication years from 2022 to 2024. The analysis highlights that topics related to “Artificial Intelligence” are expected to have the highest frequency in 2024. Other topics covered in the literature include milk production, machine learning, dairy farming and agriculture related to these domains. The shift towards incorporating Artificial Intelligence (AI) in various operational and supply chain activities within the dairy farming sector presents both new opportunities and challenges. Additionally, AI-enabled technology is proving to be an effective solution for avoiding disruptions in the supply chain and minimising risks. It generates positive outcomes, leading to the development of more solutions for logistics and supply chain issues. It is evident that digital dairy farming practices will continue to evolve and grow over time (Fig 4).
Thematic analysis
A thematic map is organised into four quadrants, divided along the x and y axes, which represent centrality and density, respectively. Co-occurrence network clusters are visually represented as bubbles on a graph. This innovative visualisation technique is based on Callon’s centrality and density rankings. Each bubble varies in size, reflecting the frequency of word occurrences within its cluster, which provides valuable insights into the prominence of specific themes in the research landscape. Each quadrant illustrates a different type of theme. The first quadrant focuses on niche themes related to animal husbandry and animal production, emphasising their relevance in terms of centrality. The second quadrant highlights main themes such as decision-making and future food supply, which are significant regarding development density. This quadrant includes prominent themes like artificial intelligence and milk production. The third quadrant identifies emerging and decoding themes, featuring subjects like dairies and the adoption of digital technology. Lastly, the fourth quadrant presents foundational themes associated with modernisation and human health. Both the first and third quadrants share common themes, such as dairying, dairy cows and milking machines, which have a notable degree of centrality. In contrast, artificial intelligence and milk production exhibit a low development degree yet possess a high degree of centrality (Fig 5).
Tree map analysis
The TreeMap below (Fig 6) illustrates the combination of potential keywords related to Artificial Intelligence and modernisation in dairy farming. This milk production and agriculture theme is also prominent to address the issue of the article.
Word frequency
Word frequency is the intensity that author keywords refer to a selection of terms selected by authors to represent the relevant content of their articles. The frequency of the keywords is a valuable tool for researchers, search engines and indexers to discover relevant studies. The five most commonly used author keywords in this field are Artificial Intelligence, Agriculture, dairy farming and machine learning (as illustrated in Fig 7). Artificial Intelligence, an emerging concept, gained importance in the database in 2000 and has since risen to prominence as the most frequently used keyword in a database spanning from 1956 to 2024. Furthermore, while Digitalisation was recognised as an important keyword in the past, its significance has notably increased since 2015, Fig 7 shows word frequency over time (Source: Authors elucidation using biblioshiny and Fig 8 shows occurrence of most cited words.
Word cloud
The most frequently visualised keywords, presented in Fig 9, were generated using the Biblioshiny application interface in R-studio. This application extracts specific keywords based on the co-occurrence of author keywords, helping to identify the most significant and influential keywords based on their popularity. The top 20 keywords, which are strongly linked and high-impact, are more likely to be cited by future researchers compared to less-cited keywords. Important keywords such as “Artificial Intelligence,” “milk production,” and “Agriculture” should be explored alongside “modernisation,” “female animals” and “dairy farming” to uncover new insights and relationships. Countries like India, Australia and Italy connect to the “Dairy Farming” keyword. Bibliometric analysis is a visualisation tool for identifying current and future research trends
(Pesta et al., 2018) Fig 9 shows the word cloud of most frequent keywords [Software: Biblioshiny (R-studio)].
Co-occurrence
Content co-occurrence analysis involves studying the content of publications, which can include topic areas such as titles and abstracts, as well as metadata like keywords. Literature reviews that utilise publication content allow researchers to distill various findings and contributions. This type of analysis enables the extraction and identification of theories, methods, contributions, samples, contexts, trending themes, highly cited themes, concepts and other valuable research data.
Using readily available software that employs algorithms to analyse data, such as VOSviewer and Bibliometrics, researchers can easily create visual representations of scholarship through co-occurrence maps. These maps demonstrate both inter- and intra-disciplinary research by revealing clusters or research streams related to the selected topic and dataset seen in Fig 10 Co-occurrence of sources in database using VOSViewer.
Bradford’s law
Bradford’s law is a vital concept that illustrates the distribution of articles on specific subjects across various periodicals. Since its introduction in 2006, the term “information scattering” has been recognised as a key phenomenon associated with information collections in our increasingly complex information landscape. The pattern of scientific articles dispersed in journals aligns remarkably with Bradford’s Law. When we categorise journals such that each category contains an equal number of articles on a given topic, the number of journals in each subsequent group typically follows a geometric progression.
To align perfectly with this law, certain criteria must be met: The bibliography should be comprehensive, cover a specific time frame and target a clearly defined subject. Despite this, the law often holds true even when these conditions are not fully satisfied. Brookes provides an engaging discussion on the implications of this law within library systems, emphasising its relevance. Currently, there is no simple model that clarifies the mechanics behind this law, high-lighting the importance of further exploration in this field.
Bradford’s Law describes how articles on a specific subject are distributed throughout various periodicals. Since 2006, the term “information scattering” has been used more generally to refer to a common phenomenon associated with information collections. The distribution of scientific articles in journals closely aligns with Bradford’s law. When journals are grouped such that each group contains the same number of articles on a particular subject, the number of journals in each succeeding group follows a geometric progression.
For strict conformity with this law, certain conditions must be met: the bibliography should be complete, cover a limited time frame and it can be easily shown that a frequency distribution J(p) of the number of journals with p articles follows the form: