Published at MetaROR
October 2, 2026
Table of contents
A mixed-methods framework for mapping graduate research contributions to the SDGs in Brazil
André Brasil1,2, Soraia de Queiroz Costa2, Gabriela da Rocha Barbosa2, Roberta Giraldi Romano3, Cláudia Terezinha Kniess4,5,6, José Carmino Gomes Junior5, Henyo Trindade Barretto Filho7, Flaviane de Magalhães Barros Bolzan de Morais8, Oswaldo E. do Amaral9, Evanilde Benedito10, Fernando Luiz Westphal11, Edna Ursulino Alves12, Maria Teresa Gomes Lopes11, Tania Mari Bellé Bresolin13, Valdir Florêncio da Veiga Junior14,11, Rômulo Dante Orrico Filho15, Paulo Ricardo Merísio16, Terezinha Valim Oliver Gonçalves17, and Carlos Alberto Cioce Sampaio18,5
1 Centre for Science and Technology Studies (CWTS), Leiden University
2 Brazilian Agency for Support and Evaluation of Graduate Education (CAPES)
3 Universidade Santa Úrsula (USU)
4 Universidade Federal de São Paulo (UNIFESP)
5 Universidade São Judas Tadeu (USJT)
6 Universidade do Sul de Santa Catarina (UNISUL)
7 Universidade de Brasília (UNB)
8 Pontifícia Universidade Católica de Minas Gerais (PUC-MG)
9 Universidade Estadual de Campinas (UNICAMP)
10 Universidade Estadual de Maringá (UEM)
11 Universidade Federal do Amazonas (UFAM)
12 Universidade Federal da Paraíba (UFPB)
13 Universidade do Vale do Itajaí (UNIVALI)
14 Instituto Militar de Engenharia (IME)
15 Universidade Federal do Rio de Janeiro (UFRJ)
16 Universidade Federal do Estado do Rio de Janeiro (UNIRIO)
17 Universidade Federal do Pará (UFPA)
18 Universidade Regional de Blumenau (FURB)
Originally published on October 9, 2025 at:
Since the adoption of the Sustainable Development Goals (SDGs) in 2015, the integration of social, economic, and environmental dimensions has become a global priority, particularly in developing regions where data gaps and governance challenges persist (Caudill et al., 2024; Zhou & Moinuddin, 2023). Earlier frameworks, such as Agenda 21 and the Millennium Development Goals, laid the foundation for participatory and integrated approaches. However, the SDGs expanded both scope and complexity, underscoring inclusivity and interdependence (Krauss, Cisneros, & Requena-i-Mora, 2022; Litre et al., 2020).
In this context, research on technical methodologies to map scientific contributions to the SDG has emerged as a critical area of inquiry. These efforts reflect the global push to align science, technology, and innovation with the 2030 Agenda, with practical implications for policy, funding, and institutional strategies (Ciarli et al., 2022; Mishra et al., 2024; Raman et al., 2023). However, mapping aligned research in the Global South presents persistent challenges, including language barriers, reliance on local publication venues, limited coverage of outputs beyond journal articles, and disciplinary imbalances in international databases such as Web of Science and Scopus (Diop et al., 2022; Pateman, Tuhkanen, & Cinderby, 2021).
Brazil illustrates these challenges clearly. A significant share of its scholarly production is published in Portuguese and appears in local or regional journals not indexed internationally. Furthermore, the unbalancing of the disciplinary coverage in international databases is even more extreme for the country than for the global average (Brasil, 2021). As a result, approaches based solely on the mapping of articles in global databases risk producing biassed or incomplete accounts of the country’s contributions to the SDG.
In this context, the Brazilian Agency for Graduate Education Support and Evaluation (CAPES) coordinated a national effort to assess and present how research conducted in Brazilian universities and research institutions contributes to the advancement of the SDG. A dedicated committee was appointed for this task, comprising CAPES staff, representatives of all 50 disciplinary evaluation panels in the Brazilian National System of Graduate Education (SNPG), the National Forum of Pro-Rectors of Graduate Studies (FOPROP), and the National Association of Graduate Students (ANPG) (CAPES, 2024). The investigation materialised in the book “Impact of Brazilian graduate education on the 2030 Agenda”, a contribution to the 30th Conference of the Parties (COP30) of the United Nations Framework Convention on Climate Change, held in the Brazilian Amazon in 2025 (Sampaio et al., 2025).
This article presents the more inclusive methodological approach developed by the committee to address the central challenge: how to comprehensively map the research contributions of a country to the SDGs when conventional bibliometric tools provide only a partial view. The strategy extended beyond international journal articles to include graduate-level theses and dissertations systematically collected through national research information systems. It builds on, but also moves beyond, international initiatives such as the Aurora SDG Classifier, Elsevier’s SDG mapping in Scopus, and the STRINGS project.
To complement this quantitative mapping, the committee incorporated qualitative inputs. A national survey of evaluation panel coordinators provided disciplinary perspectives on SDG-related research, while consultations with state departments of education examined how sustainability themes and SDGs are interpreted in basic and secondary education policies throughout Brazil. Together with case studies submitted by graduate programme directors, these components offered complementary vantage points—connecting research production, educational policy, and local practice. By documenting this process, the article presents the integrative approach of the committee, highlights its potential and limitations, and contributes to broader debates on the design of inclusive and policy-relevant methodologies to map the contributions of research to sustainable development.
Traditional methods for mapping SDG research and their limitations
Several initiatives have sought to identify and classify research outputs in relation to the United Nations Sustainable Development Goals. These methods are designed to support evidence-based policy, evaluation and strategic planning, but differ significantly in terms of transparency, coverage, and adaptability. Among the most widely used are the STRINGS project (Confraria et al., 2024), the Aurora Universities Network queries (Vanderfeesten, Otten, & Spielberg, 2020), and Elsevier’s SDG classification system (Elsevier & Group, 2020).
The STRINGS project (Steering Research and Innovation for Global Goals) is notable for its open and reproducible design. By publishing search strings for each of the SDGs and making them adaptable across bibliometric databases such as Web of Science and Scopus, STRINGS enables researchers to translate and refine queries in diverse contexts. This openness contrasts with Elsevier’s proprietary classification algorithm, which uses a combination of curated keyword sets and machine learning within Scopus and SciVal. Although Elsevier’s approach scales automatically in large datasets, its reliance on a commercial platform and limited disclosure of the methods restrict reproducibility (Kashnitsky et al., 2024; Raman et al., 2023). The Aurora University Network has focused on institutional alignment with the SDG, improving keyword-based searches, and testing multilingual extensions. However, it remains dependent on the coverage and biases of international databases. Other initiatives, such as SIRIS, SDSN and OSDG tools, also contribute alternative labelling systems with varying levels of transparency (Pukelis et al., 2022; Vanderfeesten, Spielberg, & Gunes, 2020).
Comparative studies show that these systems rarely converge on the same set of publications. The overlap between the STRINGS, Aurora, Elsevier, and SIRIS outputs is consistently low, with differences driven by query construction, keyword selection, and database coverage (Armitage, Lorenz, & Mikki, 2020; Li, Zhao, & Shen, 2024; Purnell, 2022). Such inconsistencies limit their usefulness for benchmarking or national policy applications (Wulff, Meier, & Mata, 2024). Even when more advanced methods such as clustering of citations or ensemble classifiers are applied, the results remain sensitive to the choice of training data, algorithms, and thresholds (Ingram, Banerjee, & Fox, 2024; Pal & Mukhopadhyay, 2024; Yin et al., 2023). Proprietary systems exacerbate the problem by withholding methodological details, further restricting independent validation (Kashnitsky et al., 2024; Raman et al., 2023).
Although machine learning and ensemble models offer promise, they also introduce practical obstacles. Transformer-based models, such as BERT or RoBERTa, which are increasingly used for the classification of SDG (Angin et al., 2022; Barros et al., 2024), require large, expert-labelled datasets and substantial computational resources. For national initiatives with constrained timelines or the need for reproducibility between multiple institutions, such approaches can be challenging to implement. Moreover, these tools remain tied to the same indexed sources that underrepresent research outputs from non-English languages, local journals, and formats such as theses and dissertations.
In summary, traditional methods for mapping SDG research have generated valuable tools and insights but remain limited by coverage, language, disciplinary, and transparency biases. These shortcomings are especially pronounced in contexts such as Brazil and the broader Global South, where the national research system is characterised by linguistic diversity and a firm reliance on locally orientated publication venues. Addressing these gaps required adapting existing methodologies to Brazilian realities, a process that began with systematic testing of available approaches. The following section outlines this quantitative start, focusing on the evaluation of different query-based methods and the choice of STRINGS as the basis for Brazil’s mapping strategy.
The Brazilian approach: a quantitative start
The starting point for the Brazilian mapping effort was a systematic assessment of alternative methodologies to identify research related to the SDG. Algorithms were developed to operationalise both the Aurora Universities Network queries and the STRINGS project methodology. This comparative exercise revealed important differences. Aurora queries produced a larger volume of results overall. However, manual inspection revealed that a significant proportion of these outputs represented false positives or were only tangentially related to the SDG. STRINGS, by contrast, retrieved a smaller number of works, but with greater specificity. Given the project’s objective of establishing a credible and representative baseline to evaluate the contributions of graduate research to the 2030 Agenda, precision was prioritised. STRINGS was therefore selected as the core algorithm for the national mapping.
The choice of STRINGS was also justified by its openness, methodological clarity, and replicability. As a query-based approach, STRINGS descriptors are transparent and can be adapted to different contexts. This made it possible to build algorithms in SQL that are both auditable and easily replicable across different datasets, which is particularly important, as applying these queries to international bibliometric databases would have reproduced the limitations already mentioned, particularly in relation to coverage, language, and document type. To overcome these constraints, the project relied on one of the most robust national Current Research Information Systems (CRIS) in Brazil, known as the Sucupira Platform.
CAPES-managed as part of the country’s mandatory evaluation system, Sucupira serves as a central data repository for all accredited graduate programmes in Brazil (PPG). Collect data on faculty, students, courses, projects, research outputs, and other relevant information. Since evaluation is a high-stakes process in the country, the coverage and quality of available information is significantly high (Brasil & Trevisol, 2025). To map the contributions of Brazilian graduate education to Agenda 2030, approximately 700,000 theses and dissertations submitted between 2013 and 2022 by more than 4,500 graduate programmes were obtained from Sucupira and compiled into a relational database. These relate to virtually every master’s or PhD defence in the country during the period, and the available metadata included titles, abstracts, and keywords in Portuguese, as well as, when available, English.
Although a substantial proportion of the records contained English abstracts and keywords, a significant share of the corpus was available only in Portuguese. It was therefore necessary to adapt the STRINGS descriptors through systematic translation and localisation. All descriptors were translated into Brazilian Portuguese and refined to capture local terminology, spelling conventions, and disciplinary usage. The translation process was also based on a robust set of descriptors for SDG-related research developed in Brazil (Junior et al., 2023), ensuring alignment with national policy documents and thematic vocabularies.
The final approach applied bilingual queries in parallel, matching both Portuguese and English descriptors to titles, abstracts, and keywords. The algorithms were implemented in SQL, using the Google Cloud platform for execution, which enabled efficient large-scale processing of the corpus while ensuring replicability and transparency (Brasil, 2025).
The use of Sucupira and the decision to go beyond articles not only addressed the structural under-representation of Global South research in international databases but also allowed for a more inclusive portrait of research activity, particularly in fields such as education, public health and social sciences. Since graduate education accounts for most of Brazil’s scientific output and Sucupira provides a uniquely comprehensive and standardised dataset that includes non-indexed research (Brasil, 2023), the Brazilian initiative was able to build a more complete and contextually grounded mapping of SDG-related research that combined the strengths of a robust national CRIS with an open methodological framework.
Complementary goals in the Brazilian context
An additional benefit of adapting STRINGS to prepare the mapping algorithm of the project was that the study was not limited to the 17 SDGs of the United Nations. As part of the book “Impact of Brazilian Graduate Education on the 2030 Agenda”, the mapping also covered three complementary goals proposed in the Brazilian context. These goals, described in the following along with their relevance, were developed through a consultative process and aimed to capture the dimensions of sustainable development that are critical to the social fabric of Brazil and are under-represented in the international framework of the SDGs (Cabral & Galvão, 2025).
SDG 18: Ethnic–Racial Equality was formally launched by the Brazilian government in November 2024 during the G20 Social, with ten goals focused on eliminating structural racism and guaranteeing rights in domains such as justice, health, housing, education, memory, and reparation. Its inclusion responds to Brazil’s history of racial inequality and the need to integrate anti-racist policies into the wider sustainability agenda. Mapping research output related to SDG 18 is crucial to assessing how academic knowledge supports policy measures aimed at confronting systemic racial disparities.
SDG 19: Art, Culture, and Communication emerged from longstanding debates, led by UNESCO and national cultural policy forums, on the transversal role of culture in sustainable development. Although culture is acknowledged as a crosscutting factor influencing all SDGs, it does not have a dedicated goal in the UN framework. In Brazil, the Ministry of Culture has advanced this agenda by consolidating cultural sustainability as both an end in itself and a means to achieve other goals. Incorporating SDG 19 into the mapping exercise allows recognition of the substantial body of graduate research in the arts, culture, and communication that would otherwise be invisible in SDG-focused analyses.
SDG 20: Rights of Indigenous Peoples and Traditional Communities is grounded in international commitments, such as the ILO Convention 169 and the UN Declaration on the Rights of Indigenous Peoples, but is contextualised to address Brazil’s national challenges, including territorial rights, cultural preservation, and recognition of traditional knowledge. Brazil hosts one of the largest and most diverse populations of indigenous and traditional communities in the world, whose contributions to biodiversity conservation and sustainability are increasingly recognised (Caudill et al., 2024; Krauss, Cisneros, & Requena-i-Mora, 2022). Mapping graduate research output related to SDG 20 provides a way to assess how the academic system contributes to safeguarding these rights and valorising Indigenous knowledge within the national development agenda.
These three goals were defined using the same transparent query-based logic as the STRINGS methodology, drawing also from national policy documents, expert consultations, and thematic vocabularies (Junior et al., 2023). This ensured methodological consistency while aligning the mapping framework with Brazil’s social realities and strengthening Brazil’s commitment to a decolonial and plural understanding of sustainable development. It also addresses calls within the literature to move beyond a one-size-fits-all framework by recognising diverse pathways to sustainability and the importance of epistemic diversity (Ràfols et al., 2021).
Benchmarking against international data
Although Sucupira offers a uniquely comprehensive and linguistically inclusive view of Brazilian graduate research, it was essential to situate these findings within the broader global research landscape. To achieve this, the project incorporated a benchmarking component using OpenAlex, an open and integrative bibliographic database developed as a public alternative to proprietary platforms.
OpenAlex was chosen for its comprehensive coverage across multiple data sources, including open-access repositories and doctoral theses. Unlike databases with narrower coverage, OpenAlex provides a more suitable basis for comparative analysis. The infrastructure developed by InSySPo, a research project at the University of Campinas (UNICAMP), in Brazil, supported the processing and analysis of OpenAlex data using the adapted STRINGS methodology applied to Sucupira, ensuring methodological consistency.
Benchmarking served three main purposes. First, it enabled comparative validation. By applying identical query sets to both datasets, the team was able to evaluate how Brazilian research appears in international sources and to identify discrepancies arising from differences in document types, language coverage, and indexing scope. Second, it facilitated international contextualisation. The exercise placed Brazil’s research output in relation to global trends in the 17 SDGs, revealing areas of thematic convergence and divergence. Third, the analysis assessed the applicability of Brazil’s complementary goals: SDG 18 (Ethnic–Racial Equality), 19 (Art, Culture, and Communication), and 20 (Rights of Indigenous Peoples and Traditional Communities) beyond the national context. By extending these descriptors to OpenAlex, the team explored how topics related to race, culture, and Indigenous rights are represented globally. The findings highlighted both areas of alignment, such as indigenous knowledge and biodiversity conservation, and areas where Brazil’s contributions remain uniquely positioned or underrepresented in the international literature.
The benchmarking process underscored the value of combining CRIS data with open international databases. While Sucupira ensured comprehensiveness and coverage of non-indexed, locally produced outputs, OpenAlex offered comparative reach and an external perspective. Together, they allowed a more multidimensional understanding of Brazil’s contributions to the SDGs, bridging global agendas and locally defined priorities.
Results of the quantitative mapping
The application of the adapted STRINGS methodology to the Sucupira data set enabled a systematic analysis of graduate-level research outputs in Brazil in relation to the 2030 Agenda. However, this article does not aim to present a comprehensive account of these results, as Sampaio et al. (2025) offers, in its many chapters, in-depth analyses of the large-scale mapping effort through multiple lenses: temporal, thematic, disciplinary, and geographic. Rather than detailing those results, this article focuses on the methodological approach that enabled such analyses. However, a brief look at some of the results from the quantitative analyses reveals the value in the approach taken.
To illustrate this point, Figure 1 presents the percentage distribution of theses and dissertations from Brazilian graduate programmes across the Sustainable Development Goals between 2013 and 2022. The chart uses fractional counting, ensuring that theses or dissertations related to more than one SDG are proportionally allocated rather than double-counted. This approach provides a more accurate representation of contributions across goals. The complementary SDGs (18–20) are also included.
Figure 1. Percentage of theses and dissertations from Brazilian graduate programmes related to Sustainable Development Goals
The distribution patterns revealed are particularly notable. One of the clearest trends is the decrease in the proportion of works not related to any SDG. At the beginning of the period, nearly 50% of theses and dissertations did not align with any of the goals; however, by 2022, this share had fallen to less than 35%. This decrease, represented by the dark grey bar at the bottom of the figure, suggests that graduate research is progressively becoming more engaged with sustainability-related topics.
Among the SDGs, some stand out as receiving greater attention: SDG 2 (Zero Hunger), SDG 3 (Good Health and Well-being), SDG 4 (Quality Education), SDG 5 (Gender Equality), and SDG 11 (Sustainable Cities and Communities). At the same time, relative growth across the SDGs is fairly even, indicating that attention has expanded broadly rather than being disproportionately directed to some goals.
However, the most significant strength of the analysis presented in Figure 1 lies in its completeness. The data set includes the near totality of graduate theses and dissertations produced in Brazil during the period, without language restrictions, geographical exclusions, or dependence on international indexing practices. This ensures that the analysis captures the true scope of graduate research in the country and avoids biases commonly found in bibliometric studies based on selective databases.
Furthermore, an additional advantage of mapping theses and dissertations from the Sucupira Platform is the breadth of information available, which provides the ability to track contributions at multiple levels, including individual graduate students, supervisors, research projects, graduate programmes, institutions, and states. The extensive coverage of micro-level data enables policymakers and researchers to conduct diverse analyses of interest. For example, Figure 1 showed Health and Well-being (SDG 3) as the most prominent SDG in Brazilian graduate research. To expand on this, Figure 2A presents the average number of theses and dissertations per graduate programme related to SDG 3 in all fields. The map uses the average number of products to account for the asymmetric distribution of PPG in the country, which would otherwise skew the concentration towards the southeast region.
Figure 2. Average number of theses and dissertations per programme related to SDG 3 (Health and Well-Being) in (a) all programmes; (b) life sciences; (c) social sciences and humanities; and (d) SDG 19 (Indigenous peoples and traditional communities).
Figure 2a reveals that the state of São Paulo has the highest overall contribution, with more than 40,000 theses and dissertations produced by 947 PPG in 2013–2022. At the other end of the spectrum, Rondônia registers an average of only 16 works per programme, representing roughly 300 theses and dissertations distributed among 18 programmes.
Figures 2b and 2c offer a disciplinary perspective on SDG 3. Figure 2b focuses on graduate programmes in the life sciences, where the average number of theses and dissertations per programme increases considerably. Rio de Janeiro emerges as the city with the highest concentration, producing nearly 13,000 works in 143 programmes. Figure 2c highlights contributions from the social sciences and humanities, where output levels are lower, as expected. In Espírito Santo, the state with the highest average, 741 theses and dissertations related to SDG 3 were produced by 29 programmes, resulting in an average of approximately 25 works per PPG. Finally, Figure 2d shifts the attention to the complementary SDG 19, a nationally defined goal that addresses the rights of indigenous peoples and traditional communities. The distribution of research on this topic reflects strong regional patterns. States in the northern region, particularly those connected to the Amazon, show a higher concentration of contributions. This alignment between research focus and regional priorities underscores the value of the mapping exercise in revealing not only disciplinary and institutional patterns, but also how local contexts shape the orientation of graduate research.
Integrating a qualitative layer
Quantitative mapping methods based on algorithmic and bibliometric tools provide important insights into the scale and distribution of SDG-related research. However, they are limited in their ability to capture social complexity, local engagement, and the feedback mechanisms linking research, education, and policy. As a result, these methods risk reinforcing structural exclusions and epistemic hierarchies, particularly in contexts where knowledge production takes diverse and often non-standard forms (Mbah & East, 2022). Therefore, recent literature has called for the integration of epistemic diversity and the recognition of local and Indigenous knowledge systems, as well as the inclusion of evidence from educational policy practice, to support more just and context-sensitive sustainability pathways (Caudill et al., 2024; Ghosh et al., 2021; Krauss, Cisneros, & Requena-i-Mora, 2022).
In line with these perspectives, the mapping framework developed for the book Impact of Brazilian Graduate Education on the 2030 Agenda included a qualitative layer designed to complement the large-scale analysis of theses and dissertations. The purpose of the project was to add disciplinary depth, policy relevance, and contextual specificity to the mapping process, giving visibility to research and educational practices that are not easily captured by bibliometric indicators.
Previous studies have demonstrated the value of qualitative approaches — including participatory mapping, narrative accounts, and case studies — to uncover situated forms of engagement and knowledge production that remain invisible in metadata-driven assessments (Ledezma, 2023; Oliveira et al., 2024). Thus, three interrelated components were implemented: a survey of disciplinary coordinators, a nationwide call for case studies from graduate programmes, and consultations with state departments of education. The first component provided an expert disciplinary interpretation of SDGs-related research; the second documented practical examples of research and outreach connecting universities to local sustainability challenges; and the third established a bridge between graduate education and basic education by identifying state-level initiatives that integrate Education for Sustainable Development (ESD) into curriculum and teacher training. Together, these components enriched the interpretive richness of the mapping, highlighted limitations in the coverage and granularity of bibliometric data, and illuminated practices of research engagement that are central to sustainable development but often excluded from publication-based assessments.
Survey of evaluation area coordinators
Brazil’s graduate education system is structured into 50 evaluation areas, each coordinated by a panel of scholars responsible for assessing academic programmes in their field. These panels, under the national evaluation system administered by CAPES, bring deep expertise in disciplinary research practices, epistemologies, and emerging thematic priorities.
A structured survey was sent to all 50 area coordinators with three guiding questions.
- How does research in your field relate to the 17 UN SDGs and the three complementary goals defined in the Brazilian context (SDG 18–20)?
- What research projects, training initiatives, outreach efforts, or partnerships exemplify this engagement?
- What opportunities and challenges do you have to advance the work related to the SDG within your discipline?
Responses were received from all 50 panels, producing a comprehensive cross-section of disciplinary perspectives. The results revealed substantial heterogeneity in the way the SDGs are interpreted and operationalised across fields. In health-related areas, such as Medicine and Nursing, emphasis is placed on SDG 3 (Good Health and Well-being), often linked to research on public health systems, epidemiology, and medical education. In Environmental and Agricultural Sciences, SDG 14 (Life Below Water) and SDG 15 (Life on Land) were central, particularly in association with biodiversity studies, agroecology and ecosystem restoration.
The humanities and Social Sciences panels frequently referenced SDG 4 (Quality Education), SDG 5 (Gender Equality) and SDG 10 (Reduced Inequalities), drawing connexions to anti-racist education, indigenous knowledge, cultural heritage, and community research. Legal studies, anthropology, and related fields are critically involved in SDG 20, focusing on territorial rights, intercultural dialogue, and the epistemological plurality required to support traditional communities.
In addition to identifying thematic alignments, many coordinators highlighted institutional and epistemic barriers to further integrating the SDGs into graduate research, including disciplinary silos, funding constraints, and the undervaluation of applied or locally rooted scholarship. These reflections served as a disciplinary map of how Brazilian graduate education interprets and contributes to the 2030 Agenda, complementing algorithmic mapping with situated expertise.
Consultations with state departments of education
A further layer of qualitative evidence involved consultations with state departments of education to explore how the Sustainable Development Goals are being operationalised in basic and secondary education and how graduate programmes can contribute to this process. Conducted through structured questionnaires distributed to all 27 Brazilian federative units in late 2024, this consultation aimed to identify local initiatives, partnerships, and challenges in linking education policy, teacher training, and sustainability-orientated curricula. 22 states responded, revealing that at least three-quarters of the Brazilian states had established partnerships with public universities and federal institutes to implement projects aligned with the SDGs, particularly those involving environmental education, teacher capacity building, and educational innovation (Gonçalves et al., 2025).
The inclusion of this component provided an institutional bridge between graduate education and public education policy, showing how research carried out in master’s and doctoral programmes is translated into pedagogical practices, teaching materials, and capacity building activities for teachers. It also illuminated the reciprocal side of this relationship, how the priorities emerging from basic education, such as the implementation of a national common curricular base and the integration of transversal sustainability themes, stimulate new research agendas within universities. By linking graduate research with the public-school system, this consultation underscored the systemic nature of sustainability education and demonstrated the potential for mutual learning between academic and policy communities.
The triangulation of disciplinary expertise, programme level practices, and state policy perspectives offered a richer interpretive framework for the data obtained through the national mapping. It enriched the understanding of how graduate research contributes not only to scientific knowledge but also to educational transformation, demonstrating that sustainable development depends on coherent and continuous action across the entire educational spectrum, from early schooling to advanced research training.
Call for case studies from graduate programmes
To complement the system-level disciplinary perspective with more grounded accounts of practice, a nationwide call for case studies was issued to all accredited graduate programmes in Brazil. Programme directors were invited to submit examples of research, teaching, or outreach activities that contributed to one or more of the SDGs. The submission process was deliberately open, allowing for diverse formats and thematic orientations, including completed work, ongoing projects, and long-term institutional initiatives (Sampaio et al., 2025).
The response was substantial: 1,950 case studies were submitted by 1,058 graduate programmes, covering every major disciplinary field and all five Brazilian regions. This corpus revealed the breadth of ways in which graduate programmes engage with sustainability challenges in context-specific and often innovative ways. The received cases, which illustrate how Brazilian Postgraduate studies contribute to the SDGs, were mainly selected by the 50 area coordinators. The selected cases by area were grouped by knowledge major areas, grouped into three schools: Life Sciences, Humanities, and Exact, Technological, and Multidisciplinary Sciences, totalling 36 selected cases. Each case, in each broad area, aimed to meet four different scopes: i) meet the largest possible number of SDGs, ii) meet the demands of the Amazon, with a leading role for Amazon Graduate Programmes, iii) meet the needs imposed by Climate Change (leading role in SDG 13), iv) meet the Adjacent SDGs: which meets an eminently Brazilian initiative related to SDG 18 – Ethnic–Racial Equality, SDG 19 – Art, Culture and Communication and SDG 20 – Indigenous Peoples and Traditional Communities. The criteria used in selecting the cases were: meeting at least one SDG from each of these three dimensions, social, environmental, and economic, in addition to the three adjacent SDGs, in correlation with Teaching and/or Research/Innovation and/or activities with the community, in addition to the robustness of the text and level of maturity of the initiative.
Examples include
- Engineering programmes that developed low-cost water purification systems for underserved rural communities.
- Social science projects using participatory methods to address food insecurity in urban peripheries.
- Public health initiatives focused on maternal and child health, implemented in partnership with local clinics and NGOs.
- Educational interventions designed to promote environmental literacy among primary and secondary school students in the Amazon region.
- Legal clinics working with Indigenous communities to advance territorial claims and cultural rights are aligned with SDG 20.
- The biotechnological potential for agricultural sustainability, with an impact on productivity and environmental responsibility, is explored in a network project involving several states of the country. This project has led to the development of new products and processes that have reached the market.
These case studies, a selection of which were published in the final book, illustrated the multiple and often non-linear pathways through which graduate research contributes to sustainable development. They also documented forms of impact — such as capacity building, local empowerment, and knowledge co-production — that are not easily captured by conventional bibliometric indicators. By opening space for institutional voices and grassroots perspectives, the case study component reinforced the project’s commitment to methodological pluralism and inclusion in sustainability assessment.
Conclusions
Mapping research contributions to the Sustainable Development Goals requires more than technical precision; it requires methodological pluralism, contextual sensitivity, and institutional inclusivity. This article has presented an integrative framework developed in Brazil to assess the contributions of graduate research to the 2030 Agenda. Grounded in both quantitative and qualitative components, the approach goes beyond conventional bibliometric tools by incorporating national CRIS data, bilingual query design, expert consultation, educational policy engagement, and programme-level case studies. The result is a more comprehensive and situated account of how graduate education contributes to sustainable development.
By adapting and extending the STRINGS methodology to a national corpus of nearly 700,000 theses and dissertations, the project addressed long-standing gaps in international SDG mapping practices, particularly those related to language coverage, document types, and disciplinary imbalances. The inclusion of three complementary goals — Ethnic–Racial Equality, Art, Culture and Communication, and Indigenous and Traditional Rights — further grounded the exercise in the social and epistemic realities of Brazil. Benchmarking against OpenAlex demonstrated the value of combining national and international datasets, while highlighting the continued under-representation of research from the Global South on global platforms.
Equally important was the expansion of the qualitative layer, now encompassing disciplinary, institutional, and policy perspectives. The evaluation panel survey revealed how sustainability is interpreted within different academic fields; case studies from graduate programmes illustrated concrete pathways of societal impact; and consultations with state departments of education illuminated how the results of graduate research inform and support the implementation of sustainability and environmental education across the public-school system. Together, these perspectives highlight the role of graduate education not only as a site of knowledge production, but also as a catalyst for transforming educational practices, teacher training, and policy innovation at multiple levels.
As countries strive to align science and education with global sustainability goals, the Brazilian experience provides a replicable yet adaptable model. It underscores the need for transparent and reproducible methodologies that recognise local priorities and value diverse forms of knowledge. Future work can build on this foundation by integrating new data sources, expanding comparative analyses, and improving tools for continuous monitoring and policy learning. By doing so, national research systems can contribute more effectively—and more equitably—to the global pursuit of sustainable development.
Conflicts of interest:
The authors declare they don’t have any commercial or associative interest that represents a conflict of interest in relation to the manuscript.
Data availability statement:
The data and algorithms that support the findings of this study are openly available in OSF at https://doi.org/10.17605/OSF.IO/9GYVB. Additional data are publicly available at https://dadosabertos.capes.gov.br and https://openalex.org.
Funding statement:
Part of this study was funded by the Brazilian Federal Agency for Support and Evaluation of Graduate Education (CAPES) — Financing Code 001.
CRedIT statement
Conceptualization: Brasil, A., Costa, S.Q., Barbosa, G.R., Romano, R.G., Kniess, C.T., Gomes Junior, J.C., Barretto Filho, H.T., Morais, F.M.B.B.d., Amaral, O.E.d., Benedito, E., Westphal, F.L., Alves, E.U., Lopes, M.T.G., Bresolin, T.M.B, Veiga Junior, V.F.d., Orrico Filho, R.D., Merísio, P.R., Gonçalves, T.V.O., Sampaio, C.A.C.; Data curation: Brasil, A.; Formal analysis: Brasil, A.; Methodology: Brasil, A., Costa, S.Q., Barbosa, G.R., Romano, R.G., Kniess, C.T., Gomes Junior, J.C., Barretto Filho, H.T., Morais, F.M.B.B.d., Amaral, O.E.d., Benedito, E., Westphal, F.L., Alves, E.U., Lopes, M.T.G., Bresolin, T.M.B, Veiga Junior, V.F.d., Orrico Filho, R.D., Merísio, P.R., Gonçalves, T.V.O., Sampaio, C.A.C.; Project administration: Brasil, A., Costa, S.Q., Barbosa, G.R., Romano, R.G., Sampaio, C.A.C.; Supervision: Sampaio, C.A.C.; Visualization: Brasil, A.; Writing – original draft: Brasil, A.; Writing – review & editing: Brasil, A., Costa, S.Q., Barbosa, G.R., Kniess, C.T., Benedito, E., Gonçalves, T.V.O., Sampaio, C.A.C.
References
Angin, M., Taşdemir, B., Yılmaz, C. A., Demiralp, G., Atay, M., Angin, P., & Dikmener, G. (2022). A RoBERTa approach for automated processing of sustainability reports. Sustainability, 14 (23), 16139. https://doi.org/10.3390/su142316139
Armitage, C. S., Lorenz, M., & Mikki, S. (2020). Mapping scholarly publications related to the sustainable development goals: Do independent bibliometric approaches get the same results? Quantitative Science Studies, 1 (3), 1092–1108. https://doi.org/10.1162/qss_a_00071
Barros, M. H. L. F. d. S., Neto, L. M., Santos, G. L., Leal, R. C. d. S., Silva, R. C. L. d., Lynn, T., Dourado, R. A., & Endo, P. T. (2024). Leveraging AI and data visualization for enhanced policy-making: Aligning research initiatives with sustainable development goals. Sustainability, 16 (24), 11050. https://doi.org/10.3390/su162411050
Brasil, A. (2021). Beyond the Web of Science: an overview of Brazilian papers indexed by regionally relevant databases. Proceedings of the 18th International Conference on Scientometrics & Informetrics, 193–204. https://bit.ly/3jOHQBR
Brasil, A. (2023). Advancing the evaluation of graduate education: Towards a multidimensional model in brazil. Leiden University. https://hdl.handle.net/1887/3645840
Brasil, A. (2025). Subsídios analíticos para o estudo da pós-graduação brasileira frente aos objetivos de desenvolvimento sustentável (tech. rep.). OSF. Retrieved May 30, 2025, from https://doi.org/10.17605/OSF.IO/9GYVB
Brasil, A., & Trevisol, J. V. (2025). Research evaluation in brazil and the netherlands: A comparative study. Research Evaluation, 34, rvaf013. https://doi.org/10.1093/reseval/rvaf013
Cabral, R., & Galvão, T. G. (Eds.). (2025). 2030 agenda guide. Cultura Acadêmica Editora.
CAPES. (2024). Portaria nº 166, de 4 de junho de 2024. institui grupo de trabalho para discutir o impacto da pós-graduação brasileira na agenda 2030 da ONU (Ordinance No. 166/2024). Diário Oficial da União. Brasília, DF.
Caudill, C. M., Pulsifer, P. L., Thumbadoo, R. V., & Taylor, D. R. F. (2024). Meeting the challenges of the UN sustainable development goals through holistic systems thinking and applied geospatial ethics. ISPRS International Journal of Geo-Information, 13 (4), 110. https://doi.org/10.3390/ijgi13040110
Ciarli, T., Aldoh, A. M. A. A., Arora, S., Arza, V., Asinsten, J., Assa, J., Chataway, J., Colonna, A., Confraria, H., Kombo, P. N., Mittal, N., Mulgan, G., Ndege, N., Noyons, E., Ouma-Mugabe, J., Bhuvana, N., Rafols, I., Steenmans, I., Stirling, A., . . . Yegros, A. (2022). Changing directions: Steering science, technology and innovation towards the sustainable development goals. University of Sussex. https://doi.org/10.20919/fsof1258
Confraria, H., Ciarli, T., Colonna, A., & Noyons, E. (2024). The STRINGS queries to identify documents related to the SDGs (+ country-SDG data) [data set]. Zenodo. https://doi.org/10.5281/zenodo.10470841
Diop, E. B., Chenal, J., Tekouabou, S. C. K., & Azmi, R. (2022). Crowdsourcing public engagement for urban planning in the global south: Methods, challenges and suggestions for future research. Sustainability, 14 (18), 11461. https://doi.org/10.3390/su141811461
Elsevier, & Group, R. (2020). The power of data to advance the SDGs: Mapping research for the sustainable development goals (tech. rep.). https://tinyurl.com/bdd5am7w
Ghosh, B., Ramos-Mejía, M., Machado, R. C., Yuana, S. L., & Schiller, K. (2021). Decolonising transitions in the global south: Towards more epistemic diversity in transitions research. Environmental Innovation and Societal Transitions, 41, 106–109. https://doi.org/10.1016/j.eist.2021.10.029
Gonçalves, T. V. O., Hardoim, E. L., Mendonca, A. P., Felipe, M. S., Silva, M. L. d., Moraes, A. O. d., Brasil, A., & Vásquez, F. G. (2025). Impact of brazilian graduate education on education for sustainable development: Possible pathways. In A. C. Sampaio, S. d. Q. Costa, A. Brasil, G. d. R. Barbosa, & R. G. Romano (Eds.), Ingram, W. A., Banerjee, B., & Fox, E. A. (2024). Agentic AI for improving precision in identifying contributions to sustainable development goals. arXiv. https://doi.org/10.48550/arxiv.2411.17598
Junior, J. C. G., Corbari, S. D., Kniess, C. T., Silva, G. M. N. d., Piontkewicz, S. C., Melo, M. d. S., Carbone, A. S., Mantovaneli Jr, O., Sobral, M. d. C. M., Junior, A. P., Fernandez, F., Dutra, A. R. d. A., Birch, R. S., Guerra, J. B. S. O. d. A., & Sampaio, C. A. C. (2023). Proposed mapping and evaluation model of sustainable development goals in graduate programs in environmental sciences in brazil. International Journal of Sustainability in Higher Education, 24 (7), 1452–1473. https://doi.org/10.1108/ijshe-07-2022-0230
Kashnitsky, Y., Roberge, G., Mu, J., Kang, K., Wang, W., Vanderfeesten, M., Rivest, M., Chamezopoulos, S., Jaworek, R., Vignes, M., Jayabalasingham, B., Boonen, F., James, C., Doornenbal, M., & Labrosse, I. (2024). Evaluating approaches to identifying research supporting the united nations sustainable development goals. Quantitative Science Studies, 5 (2), 408–425. https://doi.org/10.1162/qss_a_00304
Krauss, J. E., Cisneros, A. J., & Requena-i-Mora, M. (2022). Mapping sustainable development goals 8, 9, 12, 13 and 15 through a decolonial lens: Falling short of ‘transforming our world’. Sustainability Science, 17 (5), 1855–1872. https://doi.org/10.1007/s11625-022-01112-3
Ledezma, J. F. M. (2023). Rediscovering rural territories: Local perceptions and the benefits of collective mapping for sustainable development in colombian communities. Research in Globalization, 7, 100153. https://doi.org/10.1016/j.resglo.2023.100153
Li, L., Zhao, Y., & Shen, Z. (2024). Comparison of sustainable development goals labeling systems based on topic coverage. arXiv. https://doi.org/10.48550/arxiv.2408.13455
Litre, G., Bursztyn, M., Simoni, J., & Reis, R. M. (2020). Achieving the sustainable development goals through good enough governance: Lessons from argentine and brazilian municipalities. Agrociencia Uruguay, 24 (2), e139–e139. https://doi.org/10.31285/agro.24.139
Mbah, M. F., & East, L. A. (2022). How can “community voices” from qualitative research illuminate our understanding of the implementation of the SDGs? a scoping review. Sustainability, 14 (4), 2136. https://doi.org/10.3390/su14042136
Mishra, M., Desul, S., Santos, C. A. G., Mishra, S. K., Kamal, A. H. M., Goswami, S., Kalumba, A. M., Biswal, R., Silva, R. M. d., Santos, C. A. C. d., & Baral, K. (2024). A bibliometric analysis of sustainable development goals (SDGs): A review of progress, challenges, and opportunities. Environment, Development and Sustainability, 26 (5), 11101–11143. https://doi.org/10.1007/s10668-023-03225-w
Oliveira, D. R., Ribeiro, A. C. d. A., Araújo, R. V. d., Martins, R. C., Silva, V. M. A. d., Santos, A. B. C. d., Nogueira, M. L., & Costa, R. B. (2024). The importance of implementing sustainable practices in municipalities of the araguaia river valley, mato grosso, brazil. Revista de Gestão Social e Ambiental, 18 (11), e09788. https://doi.org/10.24857/rgsa.v18n11-127
Pal, A., & Mukhopadhyay, P. (2024). Categorisation of indian research publications by sustainable development goals (SDGs): A machine learning approach. Journal of Information and Knowledge, 303–311. https://doi.org/10.17821/srels/2024/v61i6/171637
Pateman, R., Tuhkanen, H., & Cinderby, S. (2021). Citizen science and the sustainable development goals in low and middle income country cities. Sustainability, 13 (17), 9534. https://doi.org/10.3390/su13179534
Pukelis, L., Bautista-Puig, N., Statulevičiūtė, G., Stančiauskas, V., Dikmener, G., & Akylbekova, D. (2022). OSDG 2.0: A multilingual tool for classifying text data by UN sustainable development goals (SDGs). arXiv. https://doi.org/10.48550/arxiv.2211.11252
Purnell, P. J. (2022). A comparison of different methods of identifying publications related to the united nations sustainable development goals: Case study of SDG 13: Climate action. arXiv. https://doi.org/10.48550/arxiv.2201.02006
Ràfols, I., Noyons, E., Confraria, H., & Ciarli, T. (2021). Visualising plural mappings of science for sustainable development goals (SDGs). Proceedings of the 18th Conference of the International Society for Scientometrics and Informetrics (ISSI). https://doi.org/10.31235/osf.io/yfqbd
Raman, R., Nair, V. K., Shivdas, A., Bhukya, R., Viswanathan, P., Subramaniam, N., & Nedungadi, P. (2023). Mapping sustainability reporting research with the UN’s sustainable development goal. Heliyon, 9 (8), e18510. https://doi.org/10.1016/j.heliyon.2023.e18510
Sampaio, C. A. C., Costa, S. d. Q., Brasil, A., Barbosa, G. d. R., & Romano, R. G. (Eds.). (2025). Impact of brazilian graduate education on the 2030 agenda: Contribution of the national system of graduate education to COP 30 in the amazon (A. Brasil, Trans.). CAPES. https://doi.org/10.21713/agenda2030eng
Vanderfeesten, M., Otten, R., & Spielberg, E. (2020). Search queries for “mapping research output to the sustainable development goals (SDGs)” v5.0.2 (tech. rep.). Zenodo.
Vanderfeesten, M., Spielberg, E., & Gunes, Y. (2020). Survey data of “mapping research output to the sustainable development goals (SDGs)”. Zenodo. https://doi.org/10.5281/zenodo.3798385
Wulff, D. U., Meier, D. S., & Mata, R. (2024). Using novel data and ensemble models to improve automated labeling of sustainable development goals. Sustainability Science, 19 (5), 1773–1787. https://doi.org/10.1007/s11625-024-01516-3
Yin, H., Aryani, A., Lambert, G., White, M., Salvador-Carulla, L., Sadiq, S., Sojli, E., Boddy, J., Murray, G., & Tham, W. W. (2023). Leveraging artificial intelligence technology for mapping research to sustainable development goals: A case study. arXiv. https://doi.org/10.48550/arxiv.2311.16162
Zhou, X., & Moinuddin, M. (2023). Interlinkages between the sustainable development goals, 129–162. https://doi.org/10.4337/9781803924946.00012
Editors
Kathryn Zeiler
Adrian Barnett
Editorial assessment
by Adrian Barnett
All three reviewers are positive and see the benefits of this research in terms of highlighting the limitations of standard bibliometric research. Stated benefits by the reviewers include the mixed method design, the use of the Sucupira platform, the use of theses, and including Portuguese as well as English. Reviewers #1 and #3 would like more details on the STRINGS approach. Reviewer #3 identifies some missing details that need to be added so that the methods are reproducible. Reviewers #1 and #3 both raise caution concerning over-stating the impact, as some SDG related studies may only be tangentially related. Reviewer #2 makes a comment on the generalisability of these findings from Brazil to other countries.
Recommendations from the editor
Page 10, The text around Figure 1 says that “The complementary SDGs (18–20) are also included”, but these are not in the legend.
Page 11, “nearly 50%” but the number in the figure is closer to 40%; and “fallen than less to 35%” but its 25% based on the figure.
Page 11, “At the same time, relative growth across the SDGs is fairly even” it’s hard to see the relative growths using this stacked plot. The authors could consider a different approach if growths are important.
Recommendations for enhanced transparency
- Add data files to the OSF link provided in the Data Availability Statement
- All data used to produce reported results should be made available for ease of result replication. Datasets should be assigned DOIs (or other persistent identifiers) and open licenses and published with proper documentation to ensure maximal reusability.
- All analytical code used to produce results, figures, and tables should be made publicly available in an appropriate repository (e.g., OSF, Zenodo, Dryad), and the article should contain a link to the code files. Analytical code files should be properly annotated. Consider including a codebook.
- Any software packages used in the research (such as for the construction of figures) should be described in the text, including version numbers, and cited in the reference section.
- Add author ORCID iDs.
- Add an email address for the corresponding author.
For more information on these recommendations, please refer to our author guidelines.
Peer review 1
This article makes a relevant contribution to mapping the relationship between Brazilian graduate research and the Sustainable Development Goals (SDGs). Its principal strength is that it addresses important limitations of conventional bibliometric approaches, particularly their dependence on English-language journal databases and their under-coverage of locally produced research, theses and dissertations, and disciplines such as education, the social sciences and the humanities. The use of the Sucupira Platform, which covers approximately 700,000 theses and dissertations produced between 2013 and 2022 across more than 4,500 graduate programmes, provides an unusually comprehensive national evidence base.
The adaptation of the open and reproducible STRINGS methodology through bilingual Portuguese–English descriptors is well justified, as is the use of OpenAlex for international benchmarking. However, I would have welcomed a brief explanation of how the English STRINGS keywords and Boolean operators were translated and adapted into Portuguese. The authors could provide, either in the main text or in a supplementary repository, the complete Portuguese and English descriptor lists, query syntax, and inclusion and exclusion rules. The citation of Junior et al. (2023) appears to provide some indication of the process, but additional details would improve transparency and reproducibility. If possible, the authors should also provide a direct link to the Portuguese queries.
The mixed-methods design is another important contribution. The survey of all 50 disciplinary evaluation panels, consultations with state education departments, and collection of 1,950 programme-level case studies add contextual and interpretive depth to the quantitative mapping. These components help capture forms of societal engagement, knowledge co-production, educational impact, and locally relevant research that are unlikely to be identified through bibliometric metadata alone.
The inclusion of three complementary Brazilian goals—SDG 18, SDG 19 and SDG 20—is a valuable addition, as it demonstrates sensitivity to national priorities and epistemic diversity. However, the article would benefit from an analysis of the intersections between these complementary goals and the 17 UN SDGs. For example, there may be substantial overlap between SDG 18, Ethnic–Racial Equality, and SDG 10, Reduced Inequalities, particularly in relation to Target 10.3, which concerns equal opportunities and the elimination of discriminatory laws, policies and practices. Such an analysis would help clarify whether the complementary goals identify distinct areas of research or provide more context-sensitive extensions of existing goals.
Relatedly, all 17 UN SDGs are associated with specific targets and indicators. It would therefore be useful to clarify whether SDGs 18–20 have similarly defined targets and indicators, or whether they currently function primarily as conceptual and thematic categories. This distinction is important for interpreting their comparability with the UN framework and for assessing their potential use in monitoring and evaluation.
The limitations of keyword-based classification also deserve a more explicit discussion. Matching queries against titles, abstracts and keywords may fail to identify relevant research that uses different terminology, while potentially capturing studies that mention an SDG-related concept only incidentally. Machine-learning, citation-based and network-based methods may help address some of these limitations, although they can also function as “black boxes,” as the authors rightly acknowledge (e.g. Bautista-Puig et al., 2021; Confraria et al., 2024; Romero Goyeneche et al., 2022) . The authors could usefully discuss the trade-offs between transparency, recall and precision in greater detail.
Figure 1 suggests that the share of research related to Gender Equality is relatively high compared with the distribution reported in some global studies of SDG-related research. It would be interesting to discuss this difference in more detail. In particular, is it attributable to differences between graduate theses and peer-reviewed journal articles, to the broader coverage of the Sucupira dataset, or to a distinctive Brazilian research specialisation shaped by the country’s history, social challenges and institutional incentives?
Minor comment: The terminology surrounding the complementary goals should be corrected and standardised. Figure 2’s caption refers to SDG 19 as “Indigenous peoples and traditional communities,” whereas page 8 defines this topic as SDG 20.
References
Bautista-Puig, N., Aleixo, A.M., Leal, S., Azeiteiro, U., Costas, R., 2021. Unveiling the Research Landscape of Sustainable Development Goals and Their Inclusion in Higher Education Institutions and Research Centers: Major Trends in 2000–2017. Frontiers in Sustainability 2, 1–18. https://doi.org/10.3389/frsus.2021.620743
Confraria, H., Ciarli, T., Noyons, E., 2024. Countries’ research priorities in relation to the Sustainable Development Goals. Research Policy 53, 104950. https://doi.org/10.1016/j.respol.2023.104950
Romero Goyeneche, O.Y., Ramirez, M., Schot, J., Arroyave, F., 2022. Mobilizing the transformative power of research for achieving the Sustainable Development Goals. Research Policy 51, 104589. https://doi.org/10.1016/j.respol.2022.104589
Peer review 2
This is a mature and well-written paper, and I read it with interest. It does something the SDG-mapping literature has been discussing for years without demonstrating. It shows that with a comprehensive national CRIS, mapping research contributions to the SDGs can be done for an entire national system, in the national language, and for document types that international databases do not cover. Nearly 700,000 theses and dissertations from more than 4,500 programmes is an enormous body of material, and it is well used in this study.
I would like to highlight three of the authors’ choices: (1) the mixed-methods approach; (2) the bilingual descriptor design, which treats Portuguese as a research language rather than as a coverage problem; and (3) SDGs 18–20. The complementary goals are the really interesting part of the paper, because they are an argument about what the UN framework does not see.
My comments concern the discussion rather than the methods. The paper is too modest about what it has produced, and it stops where the most interesting questions begin.
WHAT THIS TELLS US ABOUT KNOWLEDGE PRODUCED IN THE GLOBAL SOUTH
The under-representation of Southern research in Web of Science and Scopus is stated in the introduction as a premise. This project is in a position to turn that premise into evidence, and the OpenAlex benchmark is the place to do it: which SDGs and which disciplines lose most in the transition to an international database, and how would the national SDG profile look if it had been produced from that database alone? The difference between the two pictures is a finding in its own right, and it would be cited well beyond Brazil.
WHAT THESES TELL US THAT JOURNAL ARTICLES DO NOT
This is currently underdeveloped and, for me, the most interesting thread. A thesis is not simply an additional document type. It is a different unit of analysis. It is tied to one person and to a training relationship, produced at a fairly stable rate per student rather than skewed by individual productivity, written in the local language for a local committee, and it is where topics appear before they are filtered by journal gatekeeping and international audience expectations.
The authors could say more about what kind of knowledge becomes visible here: locally oriented, policy-adjacent, produced within the education system rather than for the international literature.
LESSONS FOR OTHER COUNTRIES
The conclusion states that Brazil offers a replicable yet adaptable model. It should say under what conditions. Brazil is not a typical case: a single agency, one mandatory platform, national coverage. Which elements are genuinely portable (the descriptor translation and localisation procedure, the logic of adding nationally defined goals, the SQL implementation as an auditable artefact) and which depend on institutional arrangements most countries do not have?
Countries with fragmented research information infrastructures, or with several official languages, will read this paper looking for exactly that distinction. A page separating the transferable method from the Brazilian preconditions would raise the paper’s international value considerably.
Peer review 3
General impression
I regard the use of the CAPES thesis and dissertation data as a particular strength of this study. Bibliometric analyses of SDG-related research generally rely on publications indexed in international bibliographic databases. Graduate theses and dissertations are therefore easily overlooked despite representing a substantial body of research that has undergone academic supervision and examination. The systematic aggregation of these outputs within the Brazilian national graduate education system offers an unusually rich source for studying research activity that may not reach internationally indexed publication venues.
This seems particularly important in the context of the present study. The graduate thesis and dissertation body of research has the potential to capture locally oriented research, Portuguese-language scholarship, and research concerning Indigenous communities along with other nationally or regionally important topics that may be poorly represented in international bibliographic databases. I therefore see the extension of SDG mapping to this material as an important contribution of the paper. Precisely because the underlying corpus is so valuable, however, it is important that the methodology used to classify these records by SDG is transparent, reproducible and appropriately validated.
Comment 1: The quantitative methodology needs to be made self-contained and reproducible
The principal issue I encountered in reviewing this manuscript concerns the description and reproducibility of the quantitative methodology used to identify SDG-related research. The manuscript reports a number of important methodological decisions and findings, but does not currently provide sufficient information for a reader to understand precisely how the classification was implemented or to reproduce it without substantial investigation of external sources.
I attempted to trace the methodology because it is central to the results presented in the paper. The manuscript directs readers to the CAPES open-data portal and, in the Data Availability Statement, to an OSF repository. The CAPES portal enables access to the underlying thesis and dissertation data, but does not itself explain the SDG classification methodology. I registered with OSF and searched the repository, but was not able to identify readily the algorithms or methodological documentation required to reconstruct the classification from the information provided in the manuscript.
Further investigation led me to the CAPES publication Impact of Brazilian Graduate Education on the 2030 Agenda (Sampaio et al., 2025), which is cited elsewhere in the manuscript but is not clearly identified as containing additional methodological information necessary to understand the quantitative analysis. This publication provides important additional detail. In particular, it explains that the English-language SDG descriptors from STRINGS were translated into Portuguese, and that both the original and translated descriptors were used to construct SQL search algorithms applied to titles, keywords and abstracts in the CAPES data. However, even this source does not appear to provide the SQL queries or sufficient operational detail to reproduce the classification.
I also followed the manuscript’s citation to Ciarli et al. (2022), Changing Directions. This highlights an important distinction that should be made much clearer in the manuscript. The original STRINGS methodology did not simply classify individual publications by applying the descriptors directly to their metadata. STRINGS constructed SDG queries using terms derived from multiple sources, searched 4,013 citation-based clusters of Web of Science publications, and associated research areas with SDGs according to the proportion of publications within the clusters containing the relevant query terms. It subsequently applied strict and loose interpretations of SDG-related research. The method applied to the Brazilian thesis data therefore appears to be an adaptation using the STRINGS descriptors, rather than an implementation of the complete STRINGS methodology.
I recommend that the authors make the quantitative methodology substantially more self-contained. The manuscript should distinguish clearly between (i) the original STRINGS methodology; (ii) methodological work undertaken previously in the CAPES project or other prior studies; and (iii) the methodology actually implemented for the analyses reported in the present manuscript. Where methods or results derive from previous work, these should be explicitly cited. Where they are new to the present study, sufficient methodological information and supporting evidence should be provided here or in clearly identified supplementary materials.
Comment 2: The Aurora-versus-STRINGS comparison needs evidence
A related concern is the manuscript’s account of the comparison between the Aurora Universities Network and STRINGS approaches.
The authors state first that algorithms were developed to operationalise both approaches, then that Aurora retrieved a larger volume of records, and that manual inspection showed a significant proportion of these to be false positives or only tangentially related to the SDGs, whereas STRINGS demonstrated greater specificity. This appears to provide the empirical justification for selecting STRINGS for the subsequent analysis. However, I was unable to locate the evidence supporting this comparison either in the manuscript or in the additional methodological sources I was able to trace.
This comparison needs to be documented. At minimum, the authors should report the numbers retrieved by each approach, how the records subjected to manual inspection were sampled, the sample size, the criteria used to determine whether a record was SDG-related, who performed the assessment, whether more than one assessor was involved, how they were selected, how disagreements were resolved, and the resulting estimates of false-positive rate/precision for each method. If this validation exercise has already been reported elsewhere, the relevant publication or repository materials should be explicitly cited and linked. If it was undertaken specifically for this study, the results should be reported here.
Comment 3: The integration of the quantitative and qualitative components should be clarified
The manuscript presents itself as a mixed-methods framework and describes three qualitative components; surveys of disciplinary coordinators, programme-level case studies, and consultations with state departments of education as complementing the quantitative mapping. Each of these is an interesting component in its own right, and the breadth of the underlying project is impressive. However, I found it difficult to determine how the quantitative and qualitative evidence are actually integrated within the analytical framework.
At present, the components appear largely to provide different perspectives on the broader question of graduate education and sustainable development. For example, the survey provides disciplinary interpretations of SDG-related research, the case studies illustrate particular research and outreach activities, and the state consultations examine relationships between graduate education and educational policy. The manuscript presents this combination as ‘triangulation’ and an integrated account, but it is not clear whether or how the qualitative evidence was used to validate, refine, challenge or interpret the results of the quantitative SDG classification.
I suggest that the authors explain more explicitly the mixed methods design and the points at which integration occurs. For example, were the disciplinary experts presented with results from the quantitative mapping? Was qualitative input used to identify limitations or omissions in the algorithmic classifications? Did the quantitative results inform the sampling or selection of the case studies? Were any disagreements between the different forms of evidence examined? If the components were conducted mainly independently and subsequently brought together at the level of interpretation, this should also be stated clearly.
Comment 4: The OpenAlex comparison and claims of under-representation need clarification
The manuscript also uses OpenAlex as an international benchmark for the Brazilian mapping exercise. I agree that comparison with an international bibliographic database is of considerable interest, particularly given the manuscript’s emphasis on research that may be poorly represented in conventional international databases. However, I found the interpretation of this comparison difficult to assess.
Sucupira and OpenAlex represent substantially different source universes. Sucupira contains a national collection of Brazilian postgraduate theses and dissertations, while OpenAlex is an international scholarly database covering a broad range of publication types. Differences between the two therefore cannot necessarily be interpreted as evidence of database under-representation without taking account of differences in document type, scope, inclusion criteria and corpus construction.
Nevertheless, the manuscript concludes that benchmarking against OpenAlex highlights the continued under-representation of research from the Global South on global platforms. Elsewhere, the comparison is described as showing areas in which Brazilian contributions are uniquely positioned or under-represented in the international literature. I would encourage the authors to explain in more detail how they reached this conclusion.
In particular, it would be useful to clarify what was actually compared between Sucupira and OpenAlex, whether equivalent document types, periods and search procedures were used, what quantitative evidence demonstrates under-representation, and how alternative explanations for differences between the two datasets were considered. If the OpenAlex comparison is intended primarily to provide context rather than to estimate coverage, the claims should be framed accordingly.
Comment 5: Distinguish SDG-related research from demonstrated contribution or impact
The manuscript would also benefit from greater consistency in the terminology used to describe what the quantitative mapping establishes. The algorithm identifies theses and dissertations whose textual metadata match descriptors associated with particular SDGs. This provides evidence that research is related to, or thematically aligned with, an SDG. It does not in itself demonstrate that the research has contributed to achievement of that SDG or produced societal impact.
At several points, however, the manuscript moves between the concepts of research being “related to’, ‘aligned with’, ‘contributing to’, and having ‘impact’. This distinction is important because these terms imply progressively stronger forms of evidence. The qualitative case studies may provide evidence of particular pathways of contribution or impact such as capacity building, local empowerment, knowledge co-production, or products reaching the market. But such evidence is different from the descriptor-based classification of the approximately 700,000 theses and dissertations.
I therefore recommend that the authors distinguish consistently between thematic relationship to an SDG, which is what the quantitative classification establishes, and contribution to or impact on sustainable development, which requires additional evidence. This distinction should be reflected particularly in the interpretation of the quantitative results and in the broader claims made in the title, abstract and conclusions.
