Improving Global Agricultural Data (IGAD)

Community of Practice

Second Annual Virtual Meeting

22 March to 8 April 2022

*last updated version on 16 March 2022

After the success of the first Research Data Alliance (RDA) IGAD annual meeting in 2020, the IGAD co-chairs are thrilled to host the second annual meeting between March 22 and April 8, 2022. The meeting will be held virtually to allow members and interested attendees from all over the world to participate. The 2022 virtual meeting will provide an online platform to continue to build collaborations, share knowledge,  and develop innovations with activities including panel sessions, group discussions and contributed presentations distributed throughout the three weeks. In place of an annual meeting in 2021, the IGAD Coffee Break webinar series worked to transition the Interest Group on Agricultural Data, to the new Improving Global Agricultural Data Community of Practice (CoP). The CoP will provide better alignment with global practice, better ability to impact stakeholders via improved data systems and practices, and mutual learning from exchanging experiences. It will also increase opportunities to form partnerships on specific projects.

Registration

There are  individual registration links for each virtual session. Interested participants are encouraged to register for multiple days based on interests and needs. The full schedule of presentations and registration links for the three week long virtual meeting are available below. If you have any comments or suggestions please contact us at [email protected].

About IGAD

Formed in 2013, since its inception the Interest Group on Agricultural Data (IGAD) has grown in community strength to over 200 members, becoming one of the RDA’s most prominent Thematic Groups. IGAD is a domain-oriented group working on all issues related to global agriculture data. It represents stakeholders in managing data for agricultural research and innovation, including producing, aggregating and consuming data.

Beyond this, IGAD promotes good practices in research with regard to data sharing policies, data management plans, and data interoperability, and it is a forum for sharing experience and providing visibility to research and work in agricultural data. One of IGAD’s main roles is to serve as a platform that leads to the creation of domain-specific Working Groups.

#RDA_IGAD2022

Participants in the 2022 RDA/IGAD annual meeting are encouraged to follow along and contribute to the conversation on social media using #RDA_IGAD2022. Follow @FAOAIMS, @resdatall, @RDA_Europe, and @RDA_us to stay informed throughout the annual meeting.

Co-chairs of the Improving Global Agricultural Data (IGAD) Community of Practice

Debora Ducker (EMBRAPA), Cynthia Parr (USDA Agricultural Research Service), Valeria Pesce (Global Forum on Agricultural Research and Innovation,  GFAR), Imma Subirats Coll (Food and Agriculture Organization of the United Nations, FAO)

Activities

Tuesday, March 22. 14:00 to 15:30 UTC

Sharing experiences and creating digital dialogues in Latin America and the Caribbean (in Spanish)

Organizers & moderators: Wouter Schallier (United Nations Economic Commission for Latin America and the Caribbean) and Imma Subirats (Food and Agriculture Organization of the United Nations, FAO)

Speakers and Information about the presentations:

---

Carolina Botero, Fundación Karisma. Colombia

Carolina Botero es directora de la Fundación Karisma. Es abogada, tiene maestrías por la Universidad Libre de Brucelas (VUB) y la Universidad Autónoma de Barcelona (UAB). Desde hace más de una década trabaja en la promoción y defensa de los derechos humanos en Internet. Forma parte de la Junta Directiva de Creative Commons y del Comité de Dirección de CSISAC en la OECD. Bajo su dirección, Fundación Karisma fue premiada con el premio Index Censorship Freedom of Expression 2019 en la categoría de activismo digital.

A Carolina le interesan los temas de libertad de expresión, inclusión social, acceso a la educación, a la cultura y a la información, la privacidad, la seguridad digital y los retos jurídicos que traen los entornos tecnológicos.

Ciencia participativa abierta en América Latina

La pandemia puso a prueba la capacidad de respuesta humana y para quienes trabajamos en la cultura abierta una gran oportunidad para la ciencia abierta y sobre todo para que la propia comunidad se articule. Revisemos diferentes proyectos de ciencia participativa que aparecieron en la región para responder a la pandemia y lo que  podemos aprender de ellos para la política pública de la región.

Oscar Luis Figueroa Rodríguez, Colegio de Postgraduados Campus Montecillo. México

Originario de Texcoco, Estado de México, graduado del departamento de Sociología Rural de la Universidad Autónoma Chapingo como Ingeniero Agrónomo especialista en Sociología Rural; del Posgrado en Desarrollo Rural del Colegio de Postgraduados Campus Montecillo como Maestro en Ciencias en Desarrollo Rural y del Departamento de Desarrollo Internacional de la Universidad de Reading (Reino Unido) como Doctor (PhD) en Desarrollo Rural e Internacional. Ha tomado diplomados en Procesos de Innovación Rural en ICRA, Wageningen Países Bajos y en Planeación Estratégica para la Administración Pública en el INAP, México. Su experiencia de trabajo ha girado alrededor del diseño, implementación y evaluación de programas y proyectos de desarrollo rural en regiones rurales de México, particularmente regiones de alta marginación y comunidades indígenas. Sus líneas de investigación son: desarrollo territorial, fortalecimiento organizacional, planeación y evaluación de programas y proyectos, participación y toma de decisiones colectivas, soberanía de datos indígenas y dirección de proyectos. Es Profesor Investigador Adjunto en el Posgrado en Desarrollo Rural del Colegio de Postgraduados Campus Montecillo, miembro fundador de la Alianza Global de Datos Indígenas (GIDA por sus siglas en inglés), miembro de la Academia Mexicana de Evaluación (ACEVAL) y de la American Evaluation Association, así como del Sistema Nacional de Investigadores, Nivel 1

La Soberanía de Datos Indígenas

La Soberanía de Datos Indígenas (SDI) se refiere a la posesión por parte de los pueblos indígenas del control y autoridad sobre la gestión de todos aquellos datos relacionados con sus comunidades, territorios y modos de vida (Kakutai, 2016) Es decir, la recolección, manejo y uso de estos datos por ellos mismos y otros actores tales como los gobiernos, empresas, organizaciones sociales y agencias de desarrollo. Este concepto de SDI se enmarca en un enfoque de derechos representado en la declaración de las Naciones Unidas sobre los Derechos de los Pueblos Indígenas (UNDRIP). La gobernanza de datos es el ejercicio del derecho en toda su amplitud de los pueblos originarios de controlar toda esta información. La soberanía de datos indígenas, la gobernanza de datos y la reconstrucción de los pueblos indígenas van de la mano. Los pueblos indígenas requieren de datos precisos, relevantes y en tiempo para la toma de decisiones y establecimiento de políticas públicas; de igual manera requieren de mecanismos de protección y control de su información. En ese sentido la Alianza Global para la Soberanía de Datos Indígenas presentó los principios CREA. La charla se centrará en presentar los fundamentos de la SDI y los principios CREA y su relación con los datos agrícolas de los pueblos indígenas.

Claudia Bauzer Medeiros, IC-UNICAMP. Brazil

Claudia Bauzer Medeiros (PhD Computer Science - University of Waterloo, Canada) is a full professor at the Institute of Computing, University of Campinas, UNICAMP. Conducts research in scientific data management, particularly with challenges associated with data heterogeneity, volume and complexity, for various types of real-world applications.   President of the Brazilian Computer Society from 2003 to 2007, she has been a member of many commissions and national evaluation committees in Brazil. She is currently a Coordination Member of the eScience and Data Science program at FAPESP, where she coordinates actions associated with open data policies, including the coordination of the construction of the state network of open research repositories. Professor Medeiros received several national and international awards for teaching, research, and initiatives to attract women to computing. Member of the boards of ACM, RDA and WDS. Member of the Brazilian Academy of Sciences and the World Academy of Sciences. To learn more about her, check out her Lattes and her website.

Implementation of the network of open research data repositories at the state of Sao Paulo, Brazil – challenges and opportunities.

Though there are many definitions of Open Science, all agree that it is based on collaboration and sharing of research results and processes. Funders, academic institutions, or even supra-national institutions are devising recommendations and policies to facilitate its implementation. However, there are many challenges to be faced in the path from recommendations to policies to specifications and implementations, including political, administrative, technical, scientific and cultural changes.

This talk will give a brief overview of the implementation of a network of open research data repositories at the state of Sao Paulo, Brazil, commenting  on some challenges and how they are being faced. Inaugurated in December 2019, actual data publication was hampered by COVID-19.

Nevertheless, the participating institutions continue to expand their resources, with open curated data from all disciplines.

Irene Ramos Pérez, Comisión Nacional para el Conocimiento y Uso de la Biodiversidad (CONABIO). México

Irene Ramos colabora en la gestión de datos del Sistema de Información de Agrobiodiversidad de CONABIO. Tiene una formación interdisciplinaria en matemáticas y ciencias de la sostenibilidad. Su investigación actual se enfoca en los retos técnicos, conceptuales y de colaboración para integrar datos socio-ecológicos. También le interesa fomentar la ciencia abierta.

Colaboración para la gestión de datos en el Sistema de Información de Agrobiodiversidad

Presentaremos las estrategias que hemos implementado para fortalecer la participación de la comunidad académica en el Sistema de Información de Agrobiodiversidad (SIAgroBD) de lCONABIO. El SIAgroBD forma parte del Proyecto GEF de Agrobiodiversidad Mexicana, que tiene a la FAO como agencia implementadora, y su objetivo es generar y sistematizar datos abiertos relacionados con cultivos nativos de importancia global.  Describiremos nuestra colaboración con los grupos de investigación que generan datos para el sistema, desde la colecta hasta la publicación. Estos grupos están conformados por estudiantes e investigadores con experiencias muy diversas en cuanto al manejo de datos. A través de talleres, encuestas y grupos de enfoque, brindamos capacitación sobre el uso de datos abiertos, evaluamos los alcances  de las herramientas digitales para  recolectar datos en campo, y promovemos la creación de bases de datos como un esfuerzo colaborativo. Construir un piso común en temas de ciencia abierta con la comunidad académica que colabora en el SIAgroBD ha facilitado poner en práctica los principios FAIR más allá de los aspectos técnicos. Esto es de particular importancia en un contexto donde hay incentivos limitados para mejorar la gestión de datos para la investigación. Las lecciones que hemos aprendido podrían ser valiosas para otros proyectos que busquen fortalecer el involucramiento de la comunidad académica en los sistemas de información.

Juan Carlos, Gobierno Regional de O'Higgins. Chile

Sociólogo, profesional de ciencias sociales en áreas como estadísticas aplicadas a las ciencias sociales, análisis de datos territoriales y población. Maestría en gestión pública, con experiencia profesional y laboral en áreas de planificación urbana, desarrollo rural, planificación territorial, análisis prospectivo y en la elaboración y actualización de políticas públicas, programas de gobiernos, programas gubernamentales, estrategias de desarrollo y planes de gestión y operativos ( POAS). Trabajó en instituciones públicas como la Subsecretaría de Desarrollo Regional y Administrativo, en el Programa de fomento productivo CHILE EMPRENDE de Sercotec, Asociación Municipalidades y FLACSO.

Experiencia de sistematización de datos del Programa de protección del bosque nativo en zonas de secano costero

Dada las condiciones de sequía crítica y crónica de los últimos 15 años en varias regiones de las 16 territorios regionales en Chile, es imperativo generar estrategias, programas y planes de acción que tiendan revertir y/o contener la escasez de recursos hídricos. En este sentido con el apoyo de FAO se creó el Sistema Integrado de Monitoreo de Ecosistemas Forestales SIMEF que recopila distintas capas de información territorial en territorios específicos que requieren especial conservación y seguimiento de calidad de suelo, de identificación de plantaciones forestales y reforestación de bosque nativo, para sustituir plantaciones más captadora de recursos hídricos. En este contexto y en base a un modelo de gestión y gobernanza participativa se crea la Plataforma digital Simef, con sistemas de información geográfica y otras herramientas de recopilación, sistematización y homologación de datos espaciales usando la experiencia de la Infraestructura de Datos Espaciales de IDE Chile.

Thursday, March 24. 14:00 to 15:00 UTC

Creating an RDA community in Latin America and the Caribbean (in Spanish)

Organizers & moderators: Wouter Schallier (United Nations Economic Commission for Latin America and the Caribbean) and Imma Subirats (Food and Agriculture Organization of the United Nations, FAO)

Speakers and Information about the presentations

---

Yolanda Meleco es el Director de Comunicaciones de RDA. En su función, sirve como enlace con los equipos de comunicación internacionales y los miembros para implementar un plan de comunicación integrado centrado en la marca, los mensajes y las actividades de divulgación eficaces, al tiempo que trabaja con la comunidad de RDA para establecer prioridades, desarrollar políticas y ejecutar las estrategias de la organización. Además de su función en RDA, Yolanda trabaja en Tetherless World Constellation en el Instituto Politécnico Rensselaer como directora de proyectos. Antes de estos puestos, Yolanda ocupó varios puestos de alto nivel en las industrias farmacéutica, de salud y financiera, incluidas AMRI (ahora Curia), Beacon Health Options, SUNY Research Foundation y Healthcare Association of New York State.

 

Patricia Rocha Bello Bertin, doctora en Gestión de la Información por la Universidad de Loughborough, Reino Unido (2014), con Maestría en Patología Molecular (2002) y Licenciada en Ciencias Biológicas por la Universidad de Brasilia (1999). Patricia es investigadora de la Empresa Brasileña de Investigación Agropecuaria (Embrapa) desde 2002, se ha dedicado al estudio y desarrollo de enfoques sistémicos para la gestión de datos, información y conocimiento en el contexto de la Ciencia, Tecnología e Innovación. Temas como Gestión de Datos de Investigación, Datos Abiertos y Ciencia Abierta son de especial interés. Fue copresidenta del Grupo de Interés de Datos Agrícolas de Research Data Alliance entre 2016 y 2021 y coordinadora del Brasileño compromiso con la Ciencia Abierta en el IV Plan de Acción Nacional de Gobierno Abierto - 2018-2020. Actualmente, Patrícia es Supervisora ​​de Gobernanza y Transparencia de la Información en el Departamento de Desarrollo Institucional de Embrapa.

Ambas presentarán actividades que RDA ha realizado o se están realizando en América latina y el Caribe y responderán a cualquier duda sobre cómo un investigador o organización pueden participar en RDA y cuáles serían sus beneficios.

Tuesday, March 29. 14:00 to 15:30 UTC

Plant-pollinator interactions data standards

Organizers & moderators: Debora Drucker (Brazilian Agricultural Research Corporation, Embrapa) and Abram Bicksler (Food and Agriculture Organization of the United Nations, FAO)

Speakers and Information about the presentations

---

Tereza Cristina Giannini. Instituto Tecnológico Vale. Brazil

Tereza works with ecological interactions, with an emphasis on species that provide ecosystem services, such as pollination, seed dispersal, as well as pollination for agriculture and food production. I also work with restoration of ecological interactions, valuation of ecosystem services and the potential impact of climate change on these services.

Interaction database on Brazilian agricultural pollinators and their multiple uses

The baseline to structure data about interactions between fauna and flora in Brazil was based on DarwinCore standards. The data standard used for this aim consisted of 219 fields divided into four main categories: [i] taxonomic information about the pollinator; [ii] taxonomic information about the agricultural crop with which the pollinator interacts, including its degree of dependence on pollinators and its annual production in the country (year 2012); [iii] location where occurred the interaction; and [iv] type of interaction (robber, floral visitor, potential and effective pollinator). After a systematic survey of the literature on agricultural pollination, 249 references were found, which identified 1470 interactions specifically referring to pollinators. Eighty-five crops were listed as well as 250 species of pollinators. From this same database, the dependency of agricultural crops on pollinators were specifically found. The data collection showed that within 141 crops analyzed, 85 have some degree of dependency on pollinators. The economic contribution of pollinators was equivalent to 30% of the total annual agricultural production of dependent crops (approximately US$ 12 billion; year 2012). Using species distribution modeling, we evaluated the effects of climate change on the geographic distribution of 95 species of pollinators from 13 Brazilian crops and estimated their relative impacts on agricultural production. Our models suggest that almost 90% of the analyzed municipalities will experience pollinator species loss. Municipalities in central and southern Brazil will potentially face relatively large impacts on agricultural production due to the loss of pollinators.

Muo Kasina. Kenya Agricultural and Livestock Research Organization. Kenya

Dr Kasina is currently the Acting Institute Director for the Apiculture Research Institute, Kenya Agricultural and Livestock Research Organization. He is an agricultural expert with lenience on pest and pollinator management practices. He has been involved in generation of plant-pollinator data of agricultural importance, advocated on the need to include pollination management as a factor of crop production, and has been in the forefront in advocating protection of pollinators. He has also supported sharing of pollinator data across the globe and contributed to global publications on pollinators and pollination management among others. He is the Chairman of the Entomological Society of Kenya

Plant-pollinator interaction data generation, use and sharing in Kenya

Agriculture is the main and important economic sector in Kenya. The country has seen upward trends in adopting modern agriculture for the past 50 years. The country was rich in life diversity, with high forest cover and high use of pollinator friendly practices in farmlands. However, the increasing reliance on modern farm management tools easily available in the local markets, has resulted in replacement of various ecological practices in the farms. This has had some negative impacts on the pollinator densities and diversities. Further, the complex environmental, economic and social realms have exacerbated the current status of pollinators across the country. Since 2000, there has been increased studies in the country to link pollinators with the country's agricultural and economic performance. Various aspects have been considered such as the community knowledge levels used in decision making; impact of pollination on specific crops; threats to pollinators; management of pollinators; policies and regulations governing various aspects of pollinators. Concurrently as data generation continued, its use became important for advocacy on regulations, policies and practice. This has seen slow but positive upward trends in the adoption of various pollinator management plans and inclusion of pollination management in productive agricultural systems. Further, various sectoral policies are utilizing the data in supporting guidance to their stakeholders with focus on enabling pollinator environments in the country. This paper will delve into specific details and experiences in Kenya across the data generation, use and sharing of plant-pollinator interaction data.

Laura Becker. ICARDA. Iceland

Laura Becker has been involved with several data-for-decision making initiatives in the nutrition and agriculture sectors. She currently coordinates the AI-Driven Climate-Smart Beekeeping for Women projects in Ethiopia, Uzbekistan, and Lebanon.

The beekeeper’s secret: sharing standardized beekeeping data while protecting beekeepers’ competitive advantages

Knowledge on good forage locations gives beekeepers a competitive advantage and a unique honey product to market. Yet if no information is exchanged on where a beekeeper’s apiary is placed, then other beekeepers may accidentally overpopulate a select area without even knowing it, resulting in reduced honeybee productivity. Additionally, data sharing on best practices and environmental factors with the wider population of beekeepers can support improved decision-making, thereby increasing the health of bees and reducing hive losses. Monitoring and sharing honeybee data is also extremely important to protect our biodiverse ecosystems, because it can give indications about broader environmental health or concerns. This session will focus on how we balance sharing beekeeping data to benefit a wide group of beekeepers and extension services, while still allowing beekeepers to keep their competitive advantage. To ensure FAIR data practices and follow international data standards, we abide by the BeeXML data standardization principles and handle data in five different ways: 1) Data that is private to the beekeeper only; 2) Data visible only to “admin” (extension services); 3) Data that is anonymized then shared to other beekeepers; 4) Data that is aggregated then shared to other beekeepers; and 5) Data that obfuscates location then shared to other beekeepers. Specifically, we will leverage examples and practices from our AI-Driven Climate-Smart Beekeeping (AID-CSB) for Women projects which localize a hive management app, the Beekeeper’s Companion, in Ethiopia, Uzbekistan, and Lebanon. Through this project, we aim to support women beekeepers to improve bee health and productivity, making local agriculture more resilient to climate change, collating vital indicators on food security, bee health, and environmental conditions.

Diana Cox-Foster. United States Department of Agriculture- USDA/ARS Pollinating Insect Research Unit. United States of America

Diana Cox-Foster is a Research Leader and Entomologist at USDA-ARS Pollinating Insects Research Unit (PIRU) in Utah.  PIRU focuses on the biology, management, and systematics of all bee species.  PIRU is home of the U.S. National Pollinating Insect Collection and associated database. Cox-Foster’s research focuses on the impact of biotic and abiotic stresses on bee health.  Cox-Foster received degrees in Entomology from Colorado State University (B.S.) and University of Illinois at Urbana-Champaign (M.S., Ph.D.). Cox-Foster served as a full professor at Penn State University in Entomology and transitioned to USDA-ARS in 2015.  At USDA, Cox-Foster is a member of the Pollinator Steering Committee.

Plant and pollinator Interactions- Understanding relationships via data at USDA

USDA is interlinking several data platforms with information on plant-pollinator interactions to provide insights on forage, habitat, and nutrition for pollinators. To interlink pollinator and plant data, USDA’s Agricultural Research Service (ARS) is coordinating with USDA’s Natural Resource Conservation Service (NRCS) and the Office of the Chief Scientist (OCS).  USDA-ARS Pollinating Insect Research Unit houses the National Pollinating Insect Collection and associated database, with plant-pollinator interactions.  This database and others are being pulled into the ARS Partnership for Data Innovation Pollinator Data Portal and the USDA NRCS PLANTS Database. The NRCS PLANTS database provides information about vascular plants, mosses, liverworts, hornworts, and lichens. NRCS is redesigning the PLANTS database to include tabs listing bee species known to visit and nutritional composition of nectar/pollen for each plant species. To coordinate efforts for national native bee monitoring via federal and non-federal entities, a US National Native Bee Monitoring Research Coordination Network has five priorities: (1) Defining scope, aims, and cost of a monitoring program; (2) Improving national capacity in bee taxonomy and systematics; (3) Gathering and cataloging data that are standardized, accessible, and sustainable; (4) Identifying core survey methods and prioritizing taxa to monitor; and (5) Prioritizing geographic areas to be monitored.

Ignasi Bartomeus. Estación Biológica de Doñana (EBD-CSIC). Spain

I am a community ecologist broadly interested in understanding how global change drivers impact community structure and composition, and how those impacts translate to the ecosystem functioning. I like to work with plant-pollinator communities because they show complex responses to land use change, climate warming or biological invasions, and encapsulate a critical ecosystem function, pollination.

CropPol: A dynamic, open and global database on crop pollination

Seventy five percent of the world's food crops benefit from insect pollination. Hence, there has been increased interest in how global change drivers impact this critical ecosystem service. Because standardized data on crop pollination are rarely available, we are limited in our capacity to understand the variation in pollination benefits to crop yield, as well as to anticipate changes in this service, develop predictions, and inform management actions. Here, I present CropPol, a dynamic, open, and global database on crop pollination. It contains measurements recorded from 202 crop studies, covering 3,394 field observations, 2,552 yield measurements (i.e., berry mass, number of fruits, and fruit density [kg/ha], among others), and 47,752 insect records from 48 commercial crops distributed around the globe. CropPol comprises 32 of the 87 leading global crops and commodities that are pollinator-dependent. The most abundant pollinator guilds recorded are honey bees, bumblebees, other wild bees, and flies. Locations comprise 34 countries, but are biased to Europe and North America. This is the most comprehensive open global data set on measurements of crop flower visitors, crop pollinators and pollination to date, and we encourage researchers to add more datasets to this database in the future.

Friday, April 1. 9:00 to 10:30 UTC

Ethical and legal issues around agricultural data

Organizer and moderator: Valeria Pesce (Global Forum on Agricultural Research and Innovation, GFAR)

Speakers and Information about the presentations

---

Foteini Zampati, Chapman Freeborn Air Marketing GmbH, Germany

Foteini Zampati is a legal professional with over 20 years of experience in various areas of private and business law. She holds an LLB in Law and a LLM in European Union and Business Law. She works at Chapman Freeborn Air Marketing GmbH as a Data Protection Advisor and was previously Data Rights Specialist at the Global Open Data for Agriculture and Nutrition (GODAN) initiative and the Kuratorium für Technik und Bauwesen in der Landwirtschaft (KTBL). Her main specialization areas are Open Data and Intellectual Property, ownership issues and data rights,compliance, best practices, and Codes of Conduct, as well as Data Protection Law and Regulations (i.e. GDPR).

Farmers´ethical and legal considerations in digital agriculture

Ethics is about choices, and agricultural ethics is about choices for people engaged in agriculture, from farmers, industries, researchers, governments, policymakers, technology developers to consumers. There is no doubt that data driven agriculture has a lot of potential and benefits. Nevertheless farmers raise concerns about data ownership, intellectual property, privacy and security. Mostly in developing countries smallholder farmers are not harnessing the power of data and many times must overcome challenges and risks to ensure that these investments benefit them. The lack of awareness about farmers ́ rights on how the data is used, will likely lead to the unfair distribution of wealth in the agricultural sector which will increase due to Data-driven knowledge.  Farmers need to feel and be engaged in their privacy and control, they seek trust and transparency in their interaction with providers and of course they would also like to receive benefits of their data, and to have access to all data. This topic is a good opportunity to start a conversation with different stakeholders by acknowledging farmers' importance in the data value chain and by exploring  how they could  actively contribute in the development of a fairer data governance framework.

Caroline Wanjiru Muchiri, CIPIT- Strathmore University, Kenya

Caroline is an intellectual property expert currently working as a Research Fellow at CIPIT- Strathmore University. She is in charge of IP and innovation research and Manager CIPIT IP Clinic. A Member, IP & Commercialisation Board Committee, Young Scientists Kenya (YSK); and Judge, Trainer, Coach and Mentor for entrepreneurs at @iBizAfrica, Kenya.Caroline is a peer reviewed writer; a World Intellectual Property (WIPO) contributor & presenter on IP, AI and data amongst others. Caroline’s research interests in Law include Innovation and Intellectual Property; Agriculture and the Law; African Feminism; Gender, Women and the Law. She is a Mentor. Mentee. A Rotarian. Amateur Farmer.

Nuanced Approach to Data Governance in Agriculture: What must be on the Agenda?

The question of agricultural data especially in Africa continues to be a subject of debate at different levels. This presentation addresses the subject using a two pronged approach: the nature of data in agriculture and gender considerations in handling such data. The presentation first categorizes data in agriculture into three classes capable of being expanded further. These categories are a) the farm data; b) farmer related data and c) the farming related data. This classification is important because it helps bring forth the nuances in the approaches to governing each data category. To govern the first category, the question on ownership of the farm or right to access the farm has to be addressed before addressing the question of data. It is in the second and the third categories the centrality of the natural person (s) undertaking the farming activities either physically or remotely becomes central. In cases of small scale farming, this person is more often than not a woman and the farming is at a subsistence and/or family level. In most cases, these farmers are not the legal owners of the farm where they are undertaking agricultural or farming activities.  

The relevance of small scale farmers and farming to the development of economies in developing countries is not debatable. They contribute directly and indirectly to the countries’ GDP, provide employment etc. In most developing countries, 70% agricultural production is at the small scale level, on food crops and is done by women. Use of technology at this level has many barriers including cultural, ownership and control, perceptions and assumptions perpetuated by industry, media and so on. In contrast, commercial agriculture is dominated by men producing horticultural crops and with easy access to technologies. Should the latter dominate the conversation on data in agriculture, the dialogue would organically revolve around ownership and control as dictated by commercial forces in the market. This single-party conversation would successfully lock out the majority of the farmers in small scale farming especially for subsistence.

As the world moves to develop and adopt forth revolution and emerging technologies, the place of data becomes an ever present and pressing issue to address. The place of data has been equated to oil in the industrial revolution. Against the backdrop of 4IR, this presenter considers agricultural data as a productive resource second to if not equal to soil where the actual productivity happens. Therefore the need to have continuous conversations on access, usefulness and equal distribution of agricultural data remains crucial. However, and noting the context raised above, this conversation would be one sided should it not include gender considerations of the natural person (s) generating, using, processing and storing the data. This is especially so for the third classification of data as provided above.

Against this context, this presentation proposes a nuanced approach to the conversation on agricultural data in Africa. The aim is not to provide solutions but to open a conversation on and highlight the existing gaps in agricultural data in Africa. The presenter hopes to propose standing agenda items that would assist in steering the dialogue to achieve inclusivity in governing data in agriculture.

Leanne Wiseman, Griffith University, Australia

Leanne Wiseman is an Australian Research Council Future Fellow and Professor in Law at Griffith University, Brisbane Australia. Leanne is an interdisciplinary scholar whose research lies at the intersections of law, science and digital technologies  She has most recently focused on the legal dimensions of the digitisation of agriculture  in national and international contexts, examining issues around ownership control and access of agricultural data. Leanne is also currently investigating the role that IP law can play in responding to the emerging International Right to Repair movement, with a particular focus on ownership of ag data and the repair of agricultural machinery.

The merits of a mandatory data sharing scheme for Agricultural Machinery

Much has been written about the ownership control and access of agricultural data. Data codes of practice have been established and implemented however it remains to be seen how much evidence is being gathered about how these data codes of practice are changing data collection and sharing practices on farm. In 2021, the Australian Consumer and Competition Commission identified significant competition concerns in the agricultural repairs aftermarket. To address these concerns, a mandatory data sharing scheme is being proposed that will mandate the sharing of data by agricultural machinery manufacturers to ensure that farmers are able to access the data that they need to conduct repairs on their agricultural machinery or to give third party repairers access to that data. This paper will discuss the aim and merits of a mandatory as opposed to a voluntary data sharing scheme and how lessons can be learned for the sharing of agricultural data more generally.

Ivo Hostens, CEMA, Belgium

Dr. Ivo Hostens is a bio-engineer holding a PhD in Applied Biological Sciences, by the University of Leuven, BE. After several years in technical consultancy, he joined AGORIA the Belgian Federation for the Technological Industry as senior expert in horizontal technical legislation. In this function as of 2008 he also coordinated the technical work in CEMA. Since 2015 he works full time for CEMA as technical director and defends the interests of its 7,000 European agricultural machinery manufacturers. Since 2019 the function of secretary general of EurAgEng, the European Society for Agricultural and Biosystems Engineering is also part of his duties.

Beyond the code of conduct – a tale of open collaboration

It all started with the Code of conduct on agricultural data sharing by contractual agreement. For many, including our industry it is the basis for further discussions. It puts assigning rights on data to farmers high on the agenda. Unfortunately it had no legal value.

Until now! From the legal side there is now a European commission proposal for a DATA ACT, which states to build on recent developments in specific sectors, such as the Code of Conduct on agricultural data sharing by contractual agreement. That act tries to  provide more legal certainty to farmers and others, in relation to the data generated on the farm.However, the implementation of the Code of conduct and the future legal provisions will be done through technical design.

There are many aspects to be taken into account for proper data sharing with issues of safety and security, interoperability and smart contracts which would lead within the proper framework and architecture to a common European agricultural data space, a federated data space that connects the existing data platform. An EU project is in development to prepare the way.

There are different initiatives that already try to facilitate data exchange and build trust by design like AgDatahub or in relation to future AI application Agri-GAIA, two use cases linked to GAIA-X. That organization aims to build the next generation of data infrastructure: an open, transparent and secure digital ecosystem, where data and services can be made available, collated and shared in an environment of trust. It aims to ensure that data of sectors like agriculture are connected not only internally but also over other sectors in the search for new functionalities, new services, to bring more added value to the individual sectors.

When it comes to safety and security, linked with connectivity and data exchange, this is most related to mobile agricultural machinery. Also CEMA-AEF made their case with  the industry initiative AgIN. With this initiative the agricultural machinery industry wants to contribute to structuring the many ongoing developments/ initiatives/ architectures, maintain legal compliance with connected products, and extend its build-up knowledge and expertise over a period of 15 years from proven concepts to the level of digital platforms.

The overall vision is to provide a sustainable and industry quality matching interoperability structure, to connect the multi coloured online platforms through harmonized technologies. More precisely  a coordinated and non discriminatory governed network would be developed ensuring reliability and trust of the services in the network between agricultural software providers. This would enable them to streamline peer–to-peer interfaces to other platforms and enable their customers to use their data in any ag platform of choice.

Besides that, the network ensures trust and reliability through well known AEF security and conformance measures, two other important core elements are considered, being the development of use case solutions and a contract framework offering simplified contracting for the participants in the network.

Such a network should allow in-field data exchange, can facilitate e.g. edge computing, will keep the freedom of innovation and could also link governments systems. Within the network farmers’ rights could be embedded in the specifications for a use case, and the services related to them. We are fully committed to support with this initiative the development of a common agricultural data space. Furthermore the use cases embedded in such trustworthy reliable architecture can facilitate an  ‘Agriculture of Data’ within the agricultural production processes allowing policy monitoring and evaluation in relation to the green deal and farm to fork strategies.

Mark Ryan, Wageningen University, The Netherlands

Mark is a Digital Ethics Researcher at Wageningen Economic Research, focusing on areas of robotics, AI, and digital developments and responsible innovation. He worked on H2020 projects, among which the SHERPA project, focused on the ethical, social and human rights implications of smart information systems (data analytics and artificial intelligence). He has published on a wide range of digital ethics topics, such as: smart cities, self-driving vehicles, agricultural data analytics, social robotics, and artificial intelligence.

Ethical and Societal Considerations for the Development and Use of Agricultural Robots

This presentation will examine the social and ethical impacts of using artificial intelligence (AI) in the agricultural sector. It will identify what are some of the most prevalent challenges and impacts identified in the literature, how this correlates with those discussed in the domain of AI ethics, and are being implemented into AI ethics guidelines. This will be achieved by examining published articles and conference proceedings that focus on societal or ethical impacts of AI in the agri-food sector, through a thematic analysis of the literature. The thematic analysis will be divided based on the classifications outlined through 11 overarching principles, from an established lexicon (transparency, justice and fairness, non-maleficence, responsibility, privacy, beneficence, freedom and autonomy, trust, dignity, sustainability, and solidarity). While research on AI agriculture is still relatively new, this presentation aims to map the debate and illustrate what the literature says in the context of social and ethical impacts. Its aim is to analyze these impacts, based on these 11 principles. This research will contrast which impacts are not being discussed in agricultural AI and which issues are not being discussed in AI ethics guidelines, but which are discussed in relation to agricultural AI. The aim of this is to identify gaps within the agricultural literature, and gaps in AI ethics guidelines, that may need to be addressed.

Monday, April 4. 14:00 to 15:30 UTC

Platforms for agricultural data: how to address the fragmented landscape? (Part 1)

Organizer and moderator: Lars Kahnert (Deutsche Gesellschaft für Internationale Zusammenarbeit, GIZ)

Speakers and Information about the presentations

---

Christopher Brewster. TN. The Netherlands

Professor Christopher Brewster is a Senior Scientist in the Data Science group at TNO, and Professor in the Application of Emerging Technologies at the Institute of Data Science, Maastricht University. His research has focussed on the application of Semantic Technologies, Open and Linked Data, interoperability architectures and Data Governance. His main concern has been the application of these technologies to the food and agriculture domain, and more recently concerning the environment and biodiversity.

The Ploutos Project: The Common Semantic Model and Data Sharing Architecture

In this presentation, we present the approach taken in the H2020 funded project Ploutos (https://ploutos-h2020.eu/) to data sharing across the agrifood system using semantic principles. The Ploutos project is a three year project begun in 2020, involving 32 partners participating in 11 pilots covering a variety of agri food sectors. The ambition of the project is to “re-balance the value chain” through the integration of business model innovation, behavioral changes and data driven technological innovation, underpinned by a broader approach for assessing the sustainability of innovations from an economic, social and environmental perspective. Here we will briefly report on the technological developments of the project.

We have adopted three key principles in view of past experience and extensive interviews with the 11 pilots, viz.

• a) Stakeholders should retain control of their data;
• b) Interoperability must be maintained with conventional software, whether legacy or developed in the future;
• c) Vendor lock-in must be avoided ensuring both flexibility and future proofing of any given part of the overall system.

Our approach has been based firstly on a semantic data model (an ontology)and second on a distributed architecture based on “interoperability enablers” to enable querying across a distributed network of stakeholder systems. The Ploutos Common Semantic Model (PCSM) follows standard good practice in ontology design, building on a set of competence questions identified for each of the 11 pilots. It reuses a number of widely used ontologies (SAREF, ENVO, SSN, SOSA, OM) and aims to cover the main farm activities in its initial release i.e. concepts related to farm, parcel, crop, soil, fertilizer and pesticide applications, and product and parcel operations. Extensions are under development for specific sub-domains such as the data requirements for farm certification (e.g. GlobalGAP, organic). The Ploutos architecture is designed to enable the controlled flow of data among the various information providers and consumers without disturbing the current operations of the underlying systems, enabling a distributed set of actors to share data without centralisation. We have designed a “Ploutos Interoperability Enabler” (PIE) as middleware which sits on top of or as an extension of existing software systems including FMISs or ERPs. Each PIE is registered with a registration and discovery directory, enabling orchestration of knowledge discovery. A PIE allows knowledge bases to exchange data in an interoperable manner with other participants in the Ploutos data sharing network. It advertises its existence using metadata descriptions of the respective type of services that it offers (e.g. available information types, data utilization policies, location, time span of data sets). The core capability of a PIE is a mapping or knowledge translator service which translates data from the common format to the stakeholder specific format and vice versa. Data never leaves a stakeholder apart from in response to specific query and only if the stakeholder has indicated agreement for data sharing with that other stakeholder. A query issued by one PIE (e.g. “When was the last time this lot of peaches was sprayed with pesticides?” is forwarded to the next PIE up the supply chain, each PIE supplying whatever data it can provide and sending the query further up the chain if needed. Each stakeholder has complete control of what data they share and with whom. The integrated system has been implemented for the frozen peach supply chain in a Greek pilot and is currently being adapted to a number of other pilots/use cases within the project including a wine supply chain Cyprus.

Michael E. Ikehi. University of Nigeria, Nsukka. Nigeria

Ikehi Michael holds a PhD in agricultural education from the University of Nigeria, Nsukka. He is an academic staff of the same institution. His research interests include quality and sharing of agricultural data, climate change, and vocational agriculture.

Agricultural research output and the issues of Data sharing with the extension service in Nigeria

Agricultural data provide essential guidance for informed decision making in food production and other aspects of agriculture including policy and programme development and modification. Research institutions including universities (of agriculture) publish research findings regularly. Data from these research outputs often present information/strategies for improving food production and general farming practices. However, the level of access to the data and availability of the data in usable form by end users such as the extension agents and the farmers remains a concern. The agricultural extension services are by design, bridges between data sources and users for improving agricultural productions. The study adopted a mix method approach in generating data to analyze the major means of dissemination of research findings by scholars, major sources of obtaining updated/recent information/findings for the improvement of agricultural practices by extension agents, and problems of data sharing between research centers and extension services. The qualitative aspect of the study involved the analysis of published agricultural researches with regards to their focus and accessibility of the data produced. The quantitative aspect of the study involved a survey of the opinions of extension agents and scholars of universities of agriculture in Nigeria, using a questionnaire. The questionnaire was face-validated by three experts and had a reliability index of 0.87 using Cronbach Alpha test for internal consistency. The study analyzed the responses from 150 lecturers and 43 extension officers. Mean was used to analyze the nominally distributed data. Findings revealed most scholars prefer to publish their research findings in international journals for recognition and promotion benefits at the place of work. The decision to make the published data open access depends on availability of funding for the research work and support for the payment of publication charge. Extension officers revealed that their sources of updated data for improving farming include training center update programmes and manuals, newspaper, television, radio and internet. The problems of data sharing between research centers and extension services include poor synergy between institutions, lacking access to research data, poor policy focus on data sharing, bureaucratic blockages in communication between agencies, and high cost of storage and sharing. Based on the findings, recommendations made include increased funding for agricultural research, and funding agricultural data curation and sharing activities. The study adds to the discussion of agricultural data availability and dissemination in Africa, particularly in agrarian countries like Nigeria.

Jan Willem Van Casteren. eProd Solutions. Kenya

A result oriented project manager with over 20 years of international work experience covering policy research, private sector development, inclusive business in agriculture and food security, renewable energy and public private partnerships, with an extensive record of securing funds from private sector and bi- and multilateral donors.

Introduction of supply chain management platform in Uganda: From Data Warehouse to Big Data

It is a big challenge for the tens of thousands of East African agricultural commodity traders, food processors and farmer cooperatives to source from the millions of small scale farmers in Africa. Tools to digitalise supply chain management processes are not available in Africa, inappropriate or too expensive.

Localized weather and agronomic advisory services are not available or unaffordable for small scale farmers. Buyers that can offer guaranteed markets would be in a unique position to offer these services bundled with other services, such as soil testing, input supply and credit, as they lack ICT tools to manage these as part of other commercial services to their small scale suppliers. However, suitable digital platforms are again unavailable or unaffordable. Therefore, the technological advancements that are made don’t turn into practical solutions as they don’t find applications in commercial markets.

In addition, agribusiness in Africa are often informal and lack the senior management level and appropriate and affordable tools to manage transparent sourcing from large numbers of small-scale farmers. Typically, field operations are therefore unstructured and incompatible with the rigid financial accounting systems required by financiers. As a result, these SMEs cannot access debt or equity finance and struggle to grow their business. eProd Solutions offers a platform to assist these agribusinesses.

This presentation covers the journey to introduce in Uganda an agricultural ERP for the private sector and its path to Big Data

Carsten Hoffmann. Leibniz-Zentrum für Agrarlandschaftsforschung (ZALF). Germany

Carsten Hoffmann is soil scientist and data curator in the BonaRes Repository. He is responsible for standards, certification and data quality and one of the German Agrovoc editors.   

BonaRes Repository – FAIR Research Data Management for agriculture

The BonaRes Centre as part of the German research initiative BonaRes (“Soil as a sustainable resource for the bioeconomy”) curates research data from the agricultural and soil science domains. Research data is managed according to the FAIR data principles, and provided via the BonaRes Repository, a geoportal which is also qualified for the provision of non-geographical data. We enhance data visibility and enable data accessibility, reusability, and interoperability with international data infrastructures (e.g., Google dataset search) using open and widely used standards, e.g. access points for OGC services, and rich metadata description, including all mandatory elements of DataCite and INSPIRE, and DOI allocation. Metadata are entered by an online metadata editor that supports the AGROVOC thesaurus for standardized keyword annotation. As a general rule, licensing is provided by CC-BY for research data and CC-0 for metadata.

The BonaRes Repository is today an important national research data infrastructure for soil and agricultural research data in Germany. Within the last years, several tools have been developed to facilitate research data management for scientists. Examples are the “Soilfiles” tool to upload standardized soil profile data and the data upload tool. With this paper we offer to to present the latest application developments in our Repository and discuss further improvements and the role in international agricultural research data management within the IGAD community.

Tuesday, April 5. 14:00 to 15:30 UTC

FAIRifying data and code for Artificial Intelligence in agricultural production

Organizers & moderators:  Erik Van Ingen (Food and Agriculture Organization of the United Nations, FAO) and Ioannis N. Athanasiadis (Wageningen University and Research)

Speakers and Information about the presentations

---

The FAIR principals (Findable, Accessible, Interoperable and Reusable) originally focused on scholarly data, see also this original publication. AI brought the concept of data to another level, where the AI models became as important as the input data used (machine learning practitioners refer to this as training data). There are great large data initiatives which have in common that they publish a wealth of data, but not the underlying models used. At the same time, AI struggles with identifying and treating biases in AI models. Google makes efforts to make the underlying models more FAIR through Colab and the Jupyter Notebook Manifesto, but there is a long way to go. The good news is that the digital public goods standard identifies open AI Models as one of the four pillars of digital public goods in the context of its work on the sustainable development goals. 193 countries adopted the first-ever global agreement on the Ethics of Artificial Intelligence on 25 November 2022, so the momentum is there to apply the FAIR principles to AI code and data.

How can we contribute to this momentum? To what extent agricultural AI models, code and data are distinct, in contrast to other application domains? Erik and Iohannis will have a panel session with Filipi and Karel. In a podcast format, Filipi, Karel and Lan will present shortly what they do, in relation to Fairifying data and code. Erik and Iohanis will then guide a discussion to get more insight on how to use the FAIR principles to make high quality training data available for data scientists and to avoid bias in machine learning.

Filipi Miranda Soares. Center for Artificial Intelligence (C4AI). Universidade de São Paulo. Brazil

He holds a degree in Library Science, a master's degree in Information Science, and is currently a Ph.D. student in Computer Engineering at the University of São Paulo. He is a member of the Biodiversity Information Standards (TDWG) Biological Data Interest Group, Safeguarding Pollinators And Pollination Services (SURPASS2), and the Brazilian Association of Agro-Informatics (board member).

Karel Charvat. WirelessInfo. Czech Republic

Karel Charvat graduated in theoretical cybernetics. He is a member of the International Society for Precision Agriculture, Research Data Alliance, Club of Ossiach, CAGI, and CSITA. He was in the period 2005 - 2007 President of European Federation for Information  Technology in Agriculture Food and  Environment (EFITA), now is chairman of OGC Agriculture DWG and member of the Program board of GEO.

Dr. L (Lan) van Wassenaer MSc.

Lan van Wassenaer is Strategic Senior Scientist at Wageningen Economic Research, part of Wageningen University & Research (WUR). Lan holds a PhD in Business Economics and MSc in Agricultural Economics from Wageningen University. Her research is focused on economic analysis and modelling of key issues in agri-food chains, in particular, risk and uncertainty, resilience, transparency and IT-mediated governance and business models and lately the impact of data economy.

Wednesday, April 6. 15:00-16:30 UTC

Governance of farm-level digital technologies and agricultural data

Organizers & moderators: Sarah-Louise Ruder (Canada, University of British Columbia), Dr. Hannah Wittman (Canada, University of British Columbia)

Speakers and Information about the presentations

---

Sarah-Louise Ruder. University of British Columbia. Canada.

PhD candidate at the University of British Columbia’s Institute for Resources, Environment, and Sustainability. Her interdisciplinary social science research explores transitions to more sustainable, food secure, and just food systems and the politics of novel agri-food technologies.

Challenges of Data Sovereignty and Data Justice in Practice

LiteFarm (www.litefarm.org) is an open-access, open-source, and free tool (web-based application) that enables farmers to collect and manage their own data on farm management practices. In October 2021, Sarah-Louise Ruder, Dr. Hannah Wittman, and Kevin Cussen hosted an IGAD “Coffee Social” introducing theory on data sovereignty and justice with LiteFarm as a case study (available here: https://youtu.be/b8veOOboQZM).

This presentation builds on the earlier webinar to focus on the practical challenges of enacting data sovereignty and justice for on-farm digital technologies and agricultural data. We will speak openly about the bottom-up process of designing the data and privacy policy for the LiteFarm and the ongoing process of revising the framework. We will report on what farmers express as most important for their relationship to data. Then, we call attention to trade-offs and challenges around aggregate data, open access, data commons, and open databases, as well as the responsibilities and permissions afforded to different actors with interest in the data (e.g., academic researchers, industry, non-governmental organizations, and government). For example, how can a data policy protect the sovereignty of farmers to make their own decisions on just relations to data? (e.g., some farmers want to be able to sell their data; others do not want data to be sold under any circumstances).

Marcus Schmidt. Leibniz Centre for Agricultural Landscape Research (ZALF). Germany

Coordinator of the BonaRes Repository at ZALF (since 2021), Project coordinator and researcher in agroecology at Göttingen (2015-2021), Doctorate in Forest Ecology at Göttingen (2015), Diploma in Physical Geography at Leipzig (2011)

Publishing sensitive agricultural data

We curate research data from the domains of agricultural and soil science and support data publication via the in-house BonaRes Repository - as part of the German research initiative BonaRes ("Soil as a Sustainable Resource for the Bioeconomy"). A majority of geospatial data we publish in the BonaRes Repository originates from field sites with high privacy requirements since many of these sites are owned by active agricultural businesses. In publishing such research data, repositories face several challenges. Firstly, it is often required that backtracking to the original area (spatial reference) is no longer possible and we will discuss three ways in which this can be achieved. Secondly, certain data such as plant disease data needs to be anonymized in ways that allows comparison between research treatments but does not allow to deduct specific values. In both of these cases, it is our aim to achieve a high level of meaningful reusability while addressing privacy concerns. The mentioned use cases require experienced data stewards and highlight the advantages of data curation and publication assistance in a field-specific repository compared to more general and less monitored data storage infrastructures.

Anton Eitzinger. Alliance of Bioversity International and CIAT. Colombia

Anton Eitzinger is a geospatial analyst and climate change scientist. He holds a doctor philosophiae in GeoSciences from the University of Munich (LMU) in Germany. In his research, he is using analytical-spatial approaches to assess the impact/magnitude and vulnerability of different agricultural and landscape systems to climate change and climate extremes. As researcher of the Alliance Bioversity International and CIAT, he is doing research in the field of Climate Resilience and Data Science.

The Metrics Coffee Risk Surveillance Network

Currently, collected farm data is fragmented between various actors and often goes underutilized because it is not shared or analyzed. In any case, farmers do not have access to farm data that were collected about them, and since most smallholders do not keep records, they are often making uninformed farming decisions. Through combining multiple data sources, a data-driven agronomy could be enabled and recommendations for farmers improved. Digital data can make farming more productive and agri-food chains more efficient and transparent. Shared access to data and interoperable information across the value chain could sustainably contribute to risk mitigation, improved service delivery and inform due diligence requirements, hence benefiting multiple stakeholders of the sector. The goal of developing the Metrics Coffee Risk Surveillance Network is to unlock the hidden farm risks by creating trust and collaboration among stakeholders. Stakeholders of the Metrix share data in a safe, secure, and decentralized network where trustworthy and incorruptible transactions of data-assets can be carried out efficiently without the need to expose the source data on the network. The Metrics network is a collaboration between science, governmental institutions, non-for-profit organizations, and the private sector and brings together multi-data sources with analytical expertise to create new insights from combining different type of data, i.e., production data collected at farm level, socio-economic data, data sourced from sensors, and among others. The newly generated insights created by the Metrix engine can be made available to all actors participating in the distributed data-network. The proposed approach seeks to fill the gap of missing coffee sector risk assessment in the economic domain of the evidence and gap map by exploring the hidden farm risks that have the largest impact on small-scale coffee farmers. The tool algorithms calculate risk indicators as smart metrics that combine multiple datasets and remotely sensed environmental data on the fly and store the calculated risk indicators in a spatial-tiling system. Combination of multiple data contributes to the robustness and credibility of the data itself, through algorithmic plausibility checks and contextualization. The spatial aggregated risk indicators are shared on the data network to all members involved in the coffee value chain without ever exposing the original data set from each provider. While the lack of understanding of the risks in coffee production especially affects the farmers, it also impacts the supply-chain members as it hinders planning of downstream activities and development and delivery of new services for farmers. A technical concept has been developed and discussed with several actors along the coffee supply chain who are interested in working on a first prototype of the Metrix network. In the next step of this initiative, we will be developing a prototype to prove the developed concept.

Alesha Miller, Digital Green, India

At Digital Green, Alesha Miller supports partnership, program development, and thought-leadership in key cross-organizational priorities, like gender, strategy, and learning. Prior to this role, Alesha was the Managing Director of the Global Food and Agriculture Program where she led research looking at key challenges and opportunities affecting the food system, including urbanization, water scarcity, and digital technology. Alesha began her career at the Bill and Melinda Gates Foundation, where she held several roles over eight years, including managing strategic partnerships with governments and grant-making to improve smallholder farmer access to markets in sub-Saharan Africa.

Farmer data sovereignty

While there has been a boom in data-powered agricultural solutions over the past several years, many of these new tools maintain ownership and control of farmers’ data for the benefit of the companies that provide those tools. At Digital Green, we start with the premise that farmers should own and manage the data they create, being in control of who and how they share that data with the broader ecosystem of service providers. Over the past decade of working on improving access and effectiveness of public advisory systems through video-enabled extension, we’ve also expanded how we think about data – beyond just strings of numbers and text – to content including video, audio, and photo files.

What does data sovereignty mean when considering data in this way? We believe that continuing to focus on core principles of control, security, and consent are key so farmers have the agency. In practice, the models for adapting these principles to useful technologies for farmers can vary widely. While there is not a perfect structure, we don’t see it as an all or nothing approach, and that creating tailored data usage and sharing policies embedded in intuitive technologies can help guide the development of solutions that benefit farmers and maintain their data sovereignty. We and other NGOs have developed examples of technologies using this approach and continue to evolve our methods and models as we learn from implementation and consider each step in the data architecture, from generation, to storage, transfer, and analysis.

Across all of these structures, there are two areas we’ve seen as critical to the development and promotion of farmer data sovereignty:

• 1) A government (or other public sector actor) that serves as a steward for secure data exchange, by investing in core infrastructure, enacting individual data protection rules, and building the human capacity to moderate and maintain a healthy and secure network.
• 2) A design approach that perpetually puts farmers first, that listens directly to and responds to their needs as individuals and as collectives.

As we go forward, we must consider not only how current technology can improve and adapt to serve individuals and maintain their rights, but also emerging capabilities. One significant transformation that is underway is the transition from Web 2.0 to Web 3.0 where greater decentralization will take place. As innovations in DAOs, NFTs, and cryptocurrency are emerging, we have an opportunity to see how these tools can apply to and help solve the perennial problems farmers and the food ecosystem face. With any potential application of these new technologies, we must still continuously look for ways to bring farmers to the table to design tools collectively and collaboratively so that the solutions are valuable to them (incentivizing adoption and use) and position them to be powerful stakeholders in the data economy going forward.

Thursday, April 7. 15:00-16:30 UTC

Platforms for agricultural data: how to address the fragmented landscape? (Part 2)

Organizer and moderator: Cynthia Sims Parr (National Agricultural Library, Agricultural Research Service, US Department of Agriculture)

Speakers and Information about the presentations

---

Nikolai Svoboda. Leibniz Centre for Agricultural Landscape Research (ZALF). Germany

Dr. Svoboda is the Head of Research Data Management at ZALF (since 2021), and he was a scientist and project coordinator at ZALF (2011-2021). He earned a doctorate in agriculture at Kiel (2011), and a diploma in physical geography at Kiel (2006).

An Overview Map of Long Term Experiments (LTE)

The BonaRes Centre as part of the German research initiative BonaRes (“Soil as a sustainable resource for the bioeconomy”) curates research data from the agricultural and soil science domains. Agricultural long term experiments (LTEs) are valuable research infrastructures to reveal the effects of agricultural measures in the long run. Worldwide, there are hundreds of such agricultural experiments but even general information about location, operation and research questions are scattered. The accessibility and reusability of research results, such as yield or soil quality changes with climate or management strategies, is even more limited. However, there is a large demand from agricultural stakeholders and the research community to get access to those data, e.g. to validate crop models under climate change but also for the improvement of decision support in agricultural practice. To overcome these limitations, the BonaRes Centre together with the EJP Soil project brought together information about more than 600 LTE (duration >20 years) in Europe. The LTE information was clustered into 49 different categories, and these clusters were statistically analyzed in various research themes, including fertilization, crop rotation and tillage treatments. Metadata is published in an online overview map (https://lte.bonares.de/experiments). In this map, users can narrow down selections by different research themes (e.g. fertilization or tillage), duration times, land use, farming category as well as status, and identify the location of trials. Users can easily download metadata and points of contact to the LTE operator.

Peter Kleinman. U.S. Dept Agriculture, Agricultural Research Service. USA

Dr. Peter Kleinman is Research Leader of the Soil Management and Sugarbeet Research Unit in Fort Collins, CO. His research in areas of agricultural sustainability and the environment has involved extensive collaborations, from focused research teams to large networks. He is a founding member of USDA's Long Term Agroecosystem Research Network and current lead of USDA's Legacy Phosphorus Research Project.

 

In pursuit of better automation, standardization and integration of agricultural field research data in USDA's network science

USDA's Partnerships for Data Innovation is the product of a grass roots initiative to improve data stewardship within the department's agricultural research community. Field research activities are diverse, such that coordinating data management offers both the reward of achieving collaborative potential as well as the challenge of confronting individual data management requirements and capacities. This presentation reviews some of the tools that PDI has developed to standardize, automate and integrate field research data. PDI's field management applications include semi-customizable surveys that support the planning and implementation of field management and field sampling efforts. Adapting these tools to the diverse needs and capacities of initiatives such as the Long Term Agroecosystem Research Network has required flexibility, compromise and vision by participants. Indeed, understanding the diversity of workflows within USDA's agricultural research community is necessary to focusing data management initiatives and avoiding unneeded friction. However, as demonstrated by USDA's Legacy Phosphorus Research Project, when PDI is engaged early in a project's life cycle, opportunity exists to integrate data streams from field research with laboratory analysis, data exploration and publication, streamlining workflows, improving coordination across sites and supporting network outcomes.

Louise Donnison. SEBI-Livestock, University of Edinburgh. UK

Livestock evidence synthesis in a fragmented data landscape

Good data on livestock disease and mortality is essential for making informed investments in animal health, and to improve productivity and incomes in low and middle-income countries. But decision-makers face enormous data challenges when it comes to grasping and making sense of the existing body of literature and evidence, which is scattered across multiple databases or even unpublished. Our research objective is to make it easier for decision makers to quickly take stock of the available evidence, by bringing together data from a fragmented landscape.

In the first phase of the project SEBI-Livestock collated evidence from currently available data on ruminant disease in Ethiopia to highlight areas of research in which there are evidence gaps and gluts. This systematic map uncovered 716 articles relevant to our scope in ruminant populations. The titles, abstracts and, subsequently, full texts were screened for inclusion based on predefined eligibility criteria detailed in a protocol published in Animal Health Research reviews. Data was extracted from the articles and used to produce an interactive dashboard, where users can explore the evidence for a given ruminant disease, region and species. In the second phase of this work, machine learning methods are being developed to automate future reviews to extend this work to other countries and ensure the evidence base stays up to date and the work can be scalable. The methodology was presented at the EMNLP Scholarly Document processing Workshop in 2020.

An important part of the success of this work is engaging with the livestock data community. LD4D runs regular meetings with members of the livestock community, and this work has been presented in those meetings. One particular area that is of active interest is standardization of terminology in describing the evidence base e.g. diseases, production systems. There is an informal working group forming around this area which is key to addressing the fragmented livestock data landscape and ensuring data can be FAIR.

Further reading:

• Scientists deploy machine learning to close data gaps on Ethiopian livestock health

• Vets get data savvy to map livestock health

Hercules Panoutsopoulos. Maastricht University and Agricultural University of Athens. Greece

Hercules Panoutsopoulos works as a Research Associate at the Agricultural University of Athens and is a PhD student at the Institute of Data Science at Maastricht University.  He holds a BSc degree in Mathematics from the National and Kapodistrian University of Athens and an MSc degree in e-Learning from the University of Piraeus. His PhD research revolves around the development of technology-supported methods for the automated update of agriculture-related, graph-based knowledge representation structures (e.g. AGROVOC) capitalizing upon the power of state-of-the-art deep neural network models and NLP techniques.

PhD research in the automated update of the AGROVOC thesaurus based on agriculture-related text corpora

Agriculture is a complex, knowledge-centric domain encompassing several fields of increased vertical specialization as well as horizontal, cross-sectoral concepts. The integration of data and information, becoming available from various sources, at the semantic level has the potential to facilitate knowledge exchange and interoperability, and is recognised as a critical factor for the advancement of sustainable agricultural practices. Formal knowledge representation structures allow for both human- and machine-readable descriptions of the domain knowledge and provide affordances for standardizing knowledge exchange. The AGROVOC multilingual thesaurus is a widely-used knowledge representation structure providing unambiguously defined agricultural concepts and the relationships between them. Given the increasing pace at which new data and information become available, using efficient, technology-based methods of keeping AGROVOC up to date is important for an effective domain-wide communication and knowledge evolution. Current advancements in machine learning, based on the proliferation of deep neural networks, and Natural Language Processing hold the promise of providing technological support to the manual labor involved in AGROVOC’s maintenance and curation. The aim of this presentation is to describe the PhD research work currently in progress regarding the utilization of state-of-the-art technologies for the automated update of AGROVOC based on available text corpora. Specifically, given an instance of the AGROVOC thesaurus released at a time point t, the focus of the research is on the development and evaluation of an automated AGROVOC update method based on a corpus of texts (publication abstracts available in the AGRIS database) spanning across a t+Δt time window. Using, a tool for the automatic recognition and extraction of agricultural terms, in-text mentions of concepts not seen in the existing AGROVOC instance are intended to be identified and added to AGROVOC. An initial version of an agriculture-related term extraction tool has been created based on the Python spaCy NLP library’s Tok2Vec and NER components, using their default architectures (i.e. spacy.Tok2Vec.v2 and spacy.TransitionBasedParser.v2 respectively). The term extraction tool has been trained on a set of 617 text documents (publication abstracts extracted from the AGRIS database) manually annotated by a group of human annotators. The results obtained from the tool evaluation have revealed the challenges inherent to approaching the agricultural term extraction as a binary classification problem. Specifically, the manual classification of a text string as an agricultural term has a great degree of vagueness, and consequently subjectivity, leaving room for different interpretations by humans and thus, impacting the tool’s performance. To this end, further research is proposed based on the utilization of the automatic text annotation affordances provided by AgroPortal. In addition, future research work will involve the development of a method for adding new relationships in AGROVOC linking the newly introduced concepts to those already in it. The development of the method will be based on the context capturing capacities of word vectorisation algorithms (e.g. bi-directional Gated Recurrent Units) and Relational Graph Convolutional Networks enhanced with graph attention mechanisms, bringing the potential to vectorise the context of a graph node with the assignment of different weights to its neighboring nodes depending on their importance.

Daniel Martini. Kuratorium für Technik und Bauwesen in der Landwirtschaft e.V. (KTBL). Germany

Daniel Martini has studied agricultural engineering with a focus on soil science at the University of Hohenheim. He is working at KTBL, a non-profit association in Germany mainly concerned with knowledge transfer in agriculture. His main area of interest is information management and processing in agriculture. He has been doing systems analysis and design as well as drafting and implementing protocols and services, mainly based on semantic technologies. He is involved in a number of research and development projects on that topic and is supporting the core team at FAO on conceptual, editorial and technical work on the AGROVOC thesaurus together with some of his colleagues.

Data vs. knowledge: a closer look at FAIR principle I1 and its implications on practical implementation

A lot of the current activities around improving data management in accord with the FAIR principles are cycling around data, services and (meta)data formats. The fact that FAIR principle I1 is however referring to knowledge representation explicitly instead of just demanding standardized data formats often goes amiss. The contribution outlines how knowledge is different from data and what properties and characteristics knowledge representation languages need to exhibit. It will then reason on why moving towards knowledge representation is important if we want to achieve interoperable services and machine-actionable data and illustrate some implementation aspects by examples.