The e-FoodLab project is inspired by the fact that food quality, security and safety, are high priority issues in our society. The food supply chain is a primary economic driver. Innovative information technology is crucial for transparent food research, intelligent food production and well-informed consumers.
The ICT-challenge of e-FoodLab is to generate and disclose semantically enriched content in the area of food and food production for research, industry and consumers, while minimizing human effort for creating and accessing this information. The underlying data is highly distributed and heterogeneous. Given the complexity of the food sector, transparency can only be achieved with advanced ICT tools that incorporate expert knowledge.
This requires the creation of formal vocabularies called ontologies (terminological networks for organizing information) that are specific for the food sector, and a way to supply and retrieve meaningful information. We identify three questions: How can experts in the field of food and food production be supported in creating dedicated ontologies? How can food data be annotated such that it can be shared and reused? How can users create and retrieve information in their specific context?
We focus on minimizing the effort required to provide semantically rich data, currently a laborious and expensive activity. In this project we combine expert knowledge and automatic data processing to solve this problem.
The e-FoodLab project is inspired by the fact that food quality, security and safety are high priority issues in society. The food supply chain is a primary economic driver and has a major influence on society in terms of health, sustainability and wellbeing. Optimizing the food supply chain is one of the societal challenges mentioned in the European Horizon2020 research program. Knowledge on food and food production has become a primary export product of the Netherlands, most notably to China. Innovative information technology is crucial for transparent food research, intelligent food production and well-informed consumers.
The ICT challenge of e-FoodLab is to generate and disclose semantically enriched content while minimizing human effort for creating and accessing this information. The underlying data is highly distributed and heterogeneous. Given complex data, transparency can only be achieved through advanced ICT tools that incorporate expert knowledge.
Development of such tools requires the creation of formal vocabularies called ontologies (terminological networks for organizing information), specifically in this project for the food sector. These ontologies help to create and retrieve meaningful information. The eFoodLab project focusses on minimizing the effort required to provide semantically rich data, currently a laborious and expensive activity. In this project we combine expert knowledge and automatic data processing to solve this problem. We identify three research objectives: How can we support experts in the field of food and food in creating dedicated ontologies? How can food data be annotated such that it can be shared and reused? How can users create and retrieve information in their specific context?
Major results have been achieved on each of the COMMIT/dimensions. Scientifically, the project has resulted in an improved understanding of scientific data management. Furthermore, a semantic model for scientific observational data that is more flexible than the present standard is developed. Moreover, the eFood-project has devised an approach to collect specific documents from a large data set given a simple ontology. Sharing three PhD and Postdoc positions with other COMMIT/projects increased synergy. Synergy is further enhanced by teaming up with the Data2Semantics project (P23) to collaboratively work with COMMIT/partner Elsevier in the area of annotating and disclosing scientific data. In terms of valorization the request from four multinational food companies to apply e-FoodLab technology in their R&D processes confirms that our work addresses a real problem and provides a promising solution. One of the spin-off products from the eFood project P26 is the ‘Allergy Checker’ that is being developed in cooperation with the Dutch Voedingscentrum. It is based on the integration of multiple heterogeneous data sources. These valorization examples are a direct result of dissemination activities, which we have extended to reach other potential international users.
Biggest results so far
Smart search using common sense
Searching information on a specific topic can be time consuming. Instead of using brute force searching that does not use any knowledge about the world, we develop smart methods that use background knowledge on specific domains. This background knowledge is formalized in an ontology: a network of concepts within a domain. An ontology in the food-domain might for example formalize that ‘Jonagold’ is a type of ‘apple’, which is related to ‘orchard’. We have developed two tools that enable smart search: the ROC+ tool and the SIEVE tool. ROC+ helps a domain expert to create his own ontology by making associations, using publicly available vocabularies. The SIEVE tool in turn uses this ontology to refine a collection of documents for example in a library, a company-specific corpus or the internet. The ontology and the refined document set enable end users to efficiently find or organize specific and high quality information. More.
ICT science question: although domain-specific ontologies enable smart search, their construction is hard and time-consuming. Domain experts are generally required to distinguish relevant from irrelevant concepts. The scientific challenge is to minimize the effort by experts while maximising its usefulness.
Application: the combination of our two tools minimize the effort needed from experts to build a specific document set and a dedicated ontology: the time needed can be reduced from weeks to hours. On our FoodVoc-website we publish ontologies related to the food domain that can be used by these tools. One application is the Valerie project, which generates innovation advice for European 7. Smart search using common sense Volume & Velocity 20 21 farmers based on results from EU-projects. The method has also been applied to assist in identifying innovation opportunities for small and medium food enterprises.
Involved partners: TI Food Nutricion, Universiteit Wageningen.
How te get more out ouf food research (video at the end of this page)
Effective food research requires that data and methods are shared. At present, careful data management is considered as a burden rather than a tool for good science. As a result, data can no longer be found or interpreted once time has passed. Furthermore, potential synergies slip through the net and costly duplications and mistakes occur.We have developed two tools for the easy management of food research data. The first tool, Tiffany, helps researchers to document their data in such a way that others can easily trace, understand and reproduce the research process. The researchers can use a second tool, Rosanne, to annotate their data in order to further improve search and reuse. Together, Tiffany and Rosanne increase the chances of successful valorisation of food research. More.
ICT science question: computers excel in data processing but do not understand the data. For example, they can find a scientific paper containing some given keywords, but they cannot tell where the conclusions in the paper came from. They can calculate the formulas in a spreadsheet but they cannot find related datasets and combine them. To support the researcher in these tasks, the computer needs machine readable models of the real world. Challenges include: How can we develop a model that supports information exchange without restricting researchers? How can we embed this support in a user-friendly manner that fits in the everyday research practice?
Application: we work together with the TI Food and Nutrition, a public/private partnership for long-term strategic research to enhance innovation in the food industry. We also cooperate with the member food companies of the institute in order to understand their needs. These companies include Unilever, Danone, Nestlé and DSM. Data management is in its infancy in food research, limited to particular tasks, such as lab management systems, or to generic document systems. These systems have difficulty in exchanging data and don’t allow the entire research workflow to be traced. Support for integrating datasets is non-existent.
Involved partners: TI Food Nutricion, Universiteit Wageningen.