How do European officials working on agriculture and food topics deal with the emerging data revolution? How can they be aware of the recent developments in a way that may inform their thinking? And how can they respond to the community needs by appropriately shaping the policy programmes and support instruments?
This was the topic of a meeting in Brussels last September, organised by the Big Data Europe project, in collaboration with the Global Open Data for Agriculture and Nutrition (GODAN) initiative. The Food and Agriculture Organization (FAO) of the United Nations and Agroknow, the BDE partners responsible for the food and agriculture sector, worked with the Institut National de la Recherche Agronomique (INRA) and Wageningen Environmental Research (Alterra) of the Wageningen University to invite representatives from the European Commission (EC) DGs (from DG AGRI, DG CONNECT, DG RTD as well as the EU Publications Office) and a select group of community stakeholders. The meeting concluded with “six main challenges in introducing big data in agriculture and food.”
1 Mechanisms for Strategic Agenda Setting on data Driven Agriculture and Nutrition.
Scope: In realising impact for end-users (farmers, policy-makers, private sector, civil society) many different developments in open data, data interoperability, analytics and visualisation need to happen simultaneously and in parallel. This should ultimately translate digital agriculture and food systems from a utopic state into an operational service. However, there can be many obstacles and blocks along the way, as many different players are active and bottom up developments are required. Therefore, it is required to elaborate the use cases towards impact for end-users in a more meaningful and deeper way, identify potential obstacles and pathways to impact. Finally, these also need to be prioritised to enable more programmatic efforts along the impact chains and cross-use case synergies.
Expected impact: The strategic agenda enables prioritisation by national and international stakeholders and end-users and will lead to the efficient & synergistic use of resources.
2. Impact of interoperability, big data analytics for beneficiaries in agri & nutrition
Scope: With more and more data available, the importance of interoperability and appropriate big data analytics techniques was stressed repeatedly, as an enabler for achieving value from data that is often held by many different and distributed organisations. Investments in interoperability, standards and analytics have been lacking as the models for collaboration and the impact of adopting standards is often poorly understood. It should thus be elaborated what the added value of adopting such techniques is, especially for non technical audiences who need to come on board to develop the next generation of digital services in agriculture and nutrition. Also, there might be unintended and potentially negative consequences (privacy, liability, loss-of-control) of the use of more digital technologies in agriculture and nutrition that need to be better understood.
Expected impact: more organisations with different backgrounds adopt standards and big data analytics techniques, and a new generation of digital services becomes available.
3. Coherent evolutionary development of standards-as-as-service.
Scope: There is a proliferation of standards and interoperability mechanisms available, leading to organisations being overwhelmed and not adopting them. This is due to the highly fragmented development of standards and ontologies so far in different communities (agro, genetics, supply chains, livestock, marine), with different purposes (search, layered applications, data representation, data exchange) and on a different aspects (syntactic, technical, semantic, engineering, governance). There is strong need to bridge the different communities, and bring them together to connect the standards, identify white-spots, best practices, and develop a next generation of standard deployment for end users, such that these can be rapidly deployed.
Expected impact: More organisations deploy standards, costs of development decrease in digital applications and there is a stronger agenda setting in developments in the sector across stakeholders.
4. Big data technologies and machine learning with open data in agriculture & nutrition.
Scope: More and more data is becoming publicly available through government open data policies and the open publishing of research data, while simultaneously satellite programmes like Copernicus are producing a mass of data. This data serves as a great resource for developing algorithms in machine learning, big data technologies, and for developing new types of algorithms (e.g. learning algorithms that use distributed data instead of compiling it all together). For agriculture, food and environment the challenge is to build such data collections as a resource (to overcome the current lack), that can serve as reference datasets for evaluation of those algorithms. This allows for benchmarking of algorithms, and dedicated improvements in performance, and thus advancements for new applications. A community is required to share such algorithms openly for open innovation.
Expected impact: An open community for development of analytic methods in agriculture, food & environment should be expected to form that is open to new entrants and provides common pool resources to SMEs and research institutes around the world.
5. FAIR Data ecosystem to support open science in Food 2030
Scope: The European Open Science Cloud aims to unlock science as an open activity in which data and publications are openly accessible. As part of the FOOD 2030 strategy, DG-RTD aims to overcome the fragmentation in research in food, agriculture and environment, and open research data and data infrastructures in relationship to EOSC are integral part of this. For food, agriculture and environment there are specific requirements for such data infrastructures due to the large heterogeneity and diversity of data, with many different languages and different organisations holding or researching on such data, in the developed and developing world. These requirements need to be scoped out, organisational commitments need to be made, cultural aspects (carrot and sticks, incentives), and technological aspects need to be addressed to arrive at consistent frameworks and architectures for open science in relation to Food 2030 and the EOSC.
b>Expected impact: many data sets are published and FAIR in food, agriculture and environment, with consistent descriptions and meta-data standards. Organisations are committed to open science.
6. Farmer participation in data value chain.
Scope: Big data is often equated to big brother and there is considerable anxiety with farmers across Europe that data on their farming operations is illegally captured from them and put to use in, for example, policy monitoring and price negotiations with suppliers. Farming industry at the same time is manoeuvring to get access to data, and concerned for competitive advantages if data is shared with competitors. An additional problem for farmers is lock-in effects and the difficulties in moving from one supplier of machinery or inputs to another, if data cannot be easily migrated. To avoid a gridlock situation, sharing models for farmer-owned data need to be developed in which farmers can opt-in for services with suppliers, can actively decide on the level of sharing with others, and have a personal data wallet that can be linked to different services, and potentially be used to generate reports on compliance, and quality of operation.
Expected impact: Models for farmer-driven participation in the data value chain, and operational examples that serves as examples throughout Europe for best practices, and allow farmers, agri-industry, and agro-ICT companies to collaborate on data rich services inclusive of farmer interests.