Published April 10, 2022 | Version v1
Journal article Open

AGRICULTURE STARTUPS (AGTECHS): A BIBLIOMETRIC STUDY

Description

ABSTRACT

Purpose:  Conduct a bibliometric study on agricultural startups (AgTech) and the main concepts related to them in the literature.

Theoretical framework: The agribusiness sector has the challenge of producing food sustainably to ensure food security for the planet's population by 2050. In this context, there is an exponential growth in investments in agriculture technology (Kakani et al., 2020). Most of these technologies are developed and marketed by AgTechs, the technological startups in agribusiness. AgTechs are expressive in the 4.0 agriculture scenario, where more precise and environmentally sustainable technologies are sought (Dutia, 2014. However, despite the growing number of AgTechs, few studies present their main concepts in the scientific literature.

Design/methodology/approach: The Web of Science (WoS) and Scopus databases were used together with softwares: SciMAT and VOSviewer to develop the bibliometric study. The SciMAT was used to clean up raw bibliographic data, analyze, and configure the analysis. The maps generated were produced at VOSviewer and based on co-citation for the periods defined in the SciMAT.

Findings: The results showed that the theme is not well consolidated in the literature, but it is in a dizzying growth, with 71.3% of the articles having been published in the last three years in 79 journals and with publications covering 44 countries.

Research, Practical & Social implications: the AgTech theme is consolidating in literature where digital and disruptive technologies are concerned, however, the human factors, business models, and management aspects involved in this topic are being neglected, which resulted in the proposal of a Research Agenda that can help both academics and practitioners to analyze AgTechs aspects that appear to not be in focus right now.

Originality/value:  The study brought important contributions to a better understanding of the term AgTech in the literature and to the improvement of concepts related to this ecosystem.

METHODOLOGY

The Web of Science (WoS) and Scopus databases were used to search the strings: “AgTech” OR “Agritech” OR “Agrotech” OR “Agriculture Startup” OR “Agricultural Startup” OR “Agriculture Startup” OR “Agricultural Startup” to find all publications that addressed the following topic: "startups aimed at agribusiness." After, SciMAT and VOSviewer software were used to carry out the other steps of the bibliometric study described below.

SciMAT

Developed by Secaba Lab at the University of Granada (Spain), SciMAT is an Open-Source software (GPLv3) that incorporates the necessary functionalities to carry out all steps of a bibliometric study, from loading the data to interpreting the outputs, incorporating methods, algorithms, and measures to obtain the different analyzes and visualizations. (Moral-Muñoz et al., 2020). In addition, the SciMAT creates scientific maps by analyzing the co-occurrence of keywords that characterize the publication, allowing the monitoring of the scientific field, delimiting the areas of investigation, and providing an understanding of the intellectual, social, conceptual, and cognitive development, as well as the analysis of its structural evolution over time (Martinez et al., 2014).

The software was developed based on the scientific mapping approach divided into four stages (Cobo et al., 2011): (1) through a bibliometric analysis for each studied period, detect the substructures contained in the research field; (2) visually display the results of the first step (clusters); (3) analyze the evolution of the clusters detected over the different periods studied to detect the main areas of evolution in the field of investigation, their origins and their interrelationships; and (4) carry out a performance analysis of the different periods, clusters and areas of evolution, through bibliometric measures.

In addition, the SciMAT software has three critical features: (a) a powerful pre-processing module to clean raw bibliographic data that makes it possible to detect duplicate files and spelling errors, organize the data chronologically, among other important pre-processing functions; (b) the use of bibliometric measures to study the impact of each element studied as maximum, minimum and average citations, as well as the use of advanced bibliometric indices, such as h-index (Alonso et al., 2009; Hirsch, 2005), G-index (Egghe, 2006), HG-index (Alonso et al., 2010) and q²-index (Cabrerizo et al., 2010); and (c) an analysis setup wizard that allows the analyst to easily select algorithms, methods, and measures to be used in the bibliometric analysis (Cobo et al., 2012).

Regarding the SciMAT software, in this work, (a) the exclusion of duplicate articles, correction of spelling errors in keywords, the addition of keywords with the same meaning (example, "IOT," "Internet of Things" and “Internet-of-Things”) and combination of plural and singular keywords (example: “Change” and “Changes”); (b) the presentation of results was performed according to the h-index, as indicated by Rincon-Patino et al. (2018); and (c) the parameters proposed by Van Eck and Waltman (2007), Cobo et al. (2012) and Rincon-Patino et al. (2018), as shown by Table 1.

Table 1 - Parameters used on the SciMAT software

Analysis Period:

1996 to 2021

Unit of Analysis:

Keywords

Frequency for Data and Network Reduction:

Minimum standard frequency (1)

Type of Network:

Co-occurrence

Standardization Measure:

Strength of association

Clustering Algorithm:

Simple Centers Algorithm, with a maximum network size of 10 and a minimum of 1

Documment Mapper:

K-mapper of 1

Bibliometric Quality and Performance measures:

H-Index and Sum of Citations

Measure for Construction of Evolution and overlay maps:

Association Strength

 

VOSviewer

VOSviewer is a software developed by the Center for Science and Technology Studies (CWTS) of Leiden University (Netherlands) and created for the construction and visualization of bibliometric networks, with individual researchers, journals, or publications as leading actors, based on - citation, bibliographic coupling or co-authorship relationships (Van-Eck; Waltman, 2010). The VOSviewer builds maps based on a three-step co-occurrence matrix (Van-Eck; Waltman, 2010): (1) similarity matrix to apply the VOS mapping technique (Waltman; Van Eck; Noyons, 2010) using the strength of association (Van-Eck; Waltman, 2007); (2) VOS mapping technique, to build a map reflecting the similarity between items; and (3) translation, rotation, and reflection, to correct the optimization problem described in the literature (O'Connell; Borg; Groenen, 1999).

The tool has network, overlay, and density as three map visualization resources, which can be saved as different file formats, facilitating the editing and handling by the analyst. Noteworthy is the zoom and scroll option that enables a detailed examination of the generated map. The maps generated in this work were based on co-citation and used the publications found between 2013 to 2021 (174 publications).

Files

Files (34.4 kB)

Name Size Download all
md5:adf2721c3248839bfc6d269a84fb7fa1
34.4 kB Download