•  

 


The aim of the research project is the development of software solutions for data-supported decision-making in Montenegrin food companies and government agencies, which on the one hand increase food safety and on the other hand accelerate investigations in the event of a food-borne disease outbreak. The new software tools to be developed should also be available as open source software for German or European interest groups. To this end, the project builds on research work by the BfR and the KLU, which have already developed the first open source software solutions as well as algorithms and models of food supply chains. In the FoodDecide project, the German project partners BfR and KLU, together with the Montenegrin partner University of Donja Gorica (UDG), can further develop the existing software technologies, algorithms and information exchange concepts in such a way that they are adapted to Montenegrin needs and can be tested in Montenegrin practice . The FoodDecide project thus also supports the existing open-source software projects such as PMM-Lab and FSK-Lab. In addition, if required, efficient data exchange protocols are established that allow a connection between existing IT or laboratory systems and the new user-friendly software tools to be developed for decision support.

 

Keywords: Capacity building, Virtual research environment, Modeling, Food Safety, Outbreak investigations

Work plan

The project work is divided into three work packages (WP).

  • WP1: Coordination, dissemination and sustainability.
This work package includes the establishment of the necessary data-related project infrastructure and the general coordination and management activities. It will also coordinate communication activities within the project (regular web meetings, physical project meetings) and with the advisory board (annual web meetings). In Task 1.2, common resources (knowledge base) are set up to enable the exchange of data as well as other project-related information and distributed collaboration, e.g. by setting up a project-specific online platform such as Confluence (https://de.wikipedia.org/wiki/Confluence_(software)). Another central focus of work is the joint establishment of standardised information exchange formats and interfaces for IT-based data exchange. Finally, Task 1.3 bundles all activities dedicated to the long-term perpetuation and commercialisation of the project results. This includes activities that build and strengthen links with stakeholders in the food production and food control sector as well as with other sectors. In addition, activities will be carried out to identify suitable business models and conduct market analyses.
Task 1.1: Communication (internal & external) (BfR)
Task 1.2: Data standards and knowledge database (BfR, KLU, UDG)
Task 1.3: Business models and dissemination (UDG, KLU, BfR)
  • WP2: Algorithmic developments.
Work package 2 concerns methodological developments for the elucidation of foodborne outbreaks. This includes both the further development and validation of existing methods (Task 2.1 & 2.2) and the development of new approaches (Task 2.3 & 2.4). The timetable foresees that existing methods will first be tested and validated in order to incorporate these findings into the development of new algorithms.
Task 2.1: Algorithms for the detection of distribution patterns (KLU)
Task 2.2: Algorithms for the identification of food (KLU, BfR)
Task 2.3: Algorithms for the inclusion of time-dependent information (KLU, BfR)
Task 2.4: Algorithms for the inclusion of analytical data (UDG, BfR, KLU)
  • WP3: Application development and testing.
Work package 3 focuses on software development tasks and follows an agile software development approach. In close coordination with WP2, this work package will conduct a requirements analysis with Montenegrin end-users, develop customised software prototypes for decision support and continuously optimise the generated demonstrator solutions based on feedback from experts and end-users.
Task 3.1: Software applications for food business operators (UDG, BfR)
Task 3.2: Software applications for food safety authorities (BfR, UDG)
Task 3.3: Software applications for SMEs in the hospitality industry (gastronomy, tourism) (UDG, KLU, BfR)

 

Events

  • Business ideas exchange market 2023.
UDG organizes Business ideas exchange market. Business ideas exchange market is place where the “Supply” and “Demand” encounter. On the „Supply” side can be your ideas, which can be licitated by the “demand”, represented by companies and institutions willing to invest in new business ideas. Depending on the quality of ideas, which should present synergy of your courage and creativity, the best 11 will have the opportunity to be presented to the "demand". Traditionally, after the presentation comes quotation for the ideas. The best 30 ideas will also be presented in the eight edition of the book "Book of ideas." It is important to emphasize that the book carries the message that the burden of success lies not only within the best 11 ideas, but in many other great ideas, in which the representatives of the "demand" will also be able to invest even though the ideas itself will not be presented. UDG, as the initiator of the competitions based on the rules of fair, open-market competitions, once again opens the call for high school and university students from both, Montenegro and Europe. You can apply via website. Location: UDG
 

The RAKIP Model Repository is a web-based repository for data and models. It currently contains microbiological risk assessment models and model modules in the FSKX format provided by RAKIP partners. All users are able to search, filter or download models from the RAKIP portal. Members of the RAKIP Initiative and registered users from other EU risk assessment agencies will also be allowed to execute simulations online with user-defined parameters for the models which are available within the repository.

 

The FoodDecide project was presented at the first international virtual fair in the WB region in the field of agriculture and information technology SmAgTech EXPO, that was held on November 17, 2021 through a specialized online platform. During the conference part of the fair, 24 lecturers from 12 countries presented theit research. Director of the UDG Centre of Excellence (FoodHub), prof. Aleksandra Martinovic, PhD and young researchers Andrea Milacic and Amil Orahovac, took part in this event as panelists within the session on the food industry.

 

The FoodDecide project was showcased in a workshop focusing on the Digitalization of Agriculture and Food Supply Chain. This event, organized by UDG is a part of the DIPOL project. The workshop featured presentations and lectures from prominent European experts in the field. In light of the COVID-19 pandemic, the workshop was conducted as a virtual online event. For those interested in accessing the materials presented during this workshop, a registration form is available at the bottom of the event page, allowing participants to gain access to these valuable resources.

 

The second virtual online fair "SmAgTech EXPO 2022" on the application of information and communication technologies in agriculture, was held on 23-24 November 2022. The Center of Excellence (FoodHub) and the FoodDecide project, which FoodHub implements in cooperation with German partners (BfR and KLU), were presented on the platform, which was created only for the needs of this project.

The fair was organized within the ERASMUS+ project VIRAL. More than 60 institutions, companies, projects and programs in the fields of agriculture and ICT were presented at the fair. During the fair, 20 plenary and sectional lectures were held.

 

  • FoodDecide project presented at IEEE IT2022 (UDG)

Young Researcher of the Centre of Excellence (FoodHub), MSc Andrea Milacic, presented FoodDecide project at the 26th International Conference on Information Technology, IEEE IT2022. The conference is traditionally taking place in Žabljak, Montenegro, but this year the event was organized in a virtual setting.

 

The advancements of the FoodDecide project in Task 2.1 were presented at the 2023 Production and Operations Management Society (POMS) Conference in Orlando. An algorithm was showcased, applying spatial statistical analysis to trace the origins of foodborne disease outbreaks by comparing outbreak case locations with retail chain store locations. This presentation, featured in both the Mini-Conference of the College of Healthcare Operations Management and the Healthcare Analytics track, facilitated the dissemination of findings and attracted constructive feedback from the global healthcare research community.

 

The poster titled "Identifying The Source: Harnessing Commonly Available Data For Spatial Modelling Of Foodborne Disease Outbreaks" was presented at the 2023 The Institute for Operations Research and the Management Sciences (INFORMS) Conference in Phoenix. This work showcased the project's ongoing development and the improvements. The session facilitated valuable conversations and exchanges, drawing on the wider community's insights and feedback to refine and advance the research further.

 

The conference focus will be on innovative techniques and digital solutions to integrate data and knowledge about hazards, exposure and technologies for improved risk assessment along feed and food chains. It will be held on 27-29 May 2024.

 

This UDG event,will be hosted on 9 May 2024 at UDG (Podgorica - Montenegro). It annually attracts over 100 student companies from Montenegro and beyond, fostering a vibrant atmosphere of idea exchange, spirited competition, creativity and global collaboration.

 

  • Final FoodDecide meeting 

This meeting is scheduled on 14-16 May 2024 at UDG (Podgorica - Montenegro).

 

 

Publications

 

 

Tools and other resources

 

The Food Security Resource Repository (FSRR) is an inventory for models related to food security. These models are used in major foresight initiatives and studies.

The webservice BfR-WarenstromInfo allows the user to easily extract data of trade in goods between countries from two different databases: The EUROSTAT database "EU trade since 2002 by HS2-4-6 and CN8 (new content) (ds-059322)" and BACI database (based on modified data from UN-Comtrade). This tool has been developed through funding from the ZooGlow, DEMETER and FoodDecide projects. It is possible to access this tool by requesting access via email to the German Federal Institute for Risk Assessment (BfR). Department Biology Safety. Group Study Centre Supply Chain Modelling and Artificial Intelligence (4SZ@bfr.bund.de).

FoodDecide has developed a new KNIME node that enables the visualization of international supply chains. This node allows visualizing the trade flows in a global map.

This software for lab data management has been successfully implemented in Montenegro after on-site training and visits from BfR expert to UDG Laboratory for Food Quality and Safety. This management tool can streamline the data transfer between laboratories, Food Business Operators and Food Safety authorities.

This is a python package that helps with creating your own supply chain model. You dont need intricate knowledge about python, you might need a bit knowledge about graphs (essential part is provided in video tutorials) and a bit about supply chain models in general. This package also gives you the opportunity to store the data you may need to make this supply chain model run in an efficient manner. A full documentation is provided inside the python package. Videotutorials will be linked here later, as well as a documentation to read.

  • Several algorithms were developed to enhance the detection of epidemiological outbreaks:
    • Gravity Model: This model simulates the flow of goods between shops and the population in different cells.
    • Outbreak Model: The outbreak model generates artificial outbreak scenarios in a given urban area based on a gravity model of the flow of goods between stores and population cells.
    • Visualization Script: A script has been created to visualize the generated outbreaks.

The Gravity Model and the Outbreak Model along with the visualization  have been transformed into FSKx files. The FSKX format aims to improve the exchange of food safety knowledge, including predictive models and associated metadata. This allows annotation with sufficient detail for any food safety model, enabling reusability and understanding of the models.

This is a proof-of-concept tool for generating population distributions based on different types, including random, uniform, linear, radial clusters, and main-and-small clusters. It provides a visual representation of the generated populations to aid analysis and exploration. The use of this tool was restricted to the BfR.

This is a proof-of-concept tool for generating and visulizing the allocation of stores for two retail chains in a hypothetical scenario based on user-defined parameters such as spatial distribution, sales distribution, and population data. The use of this tool was restricted to the BfR.

This tool is open and enables the identification of all restaurants and/or supermarkets within one or multiple areas with a specified radius (e.g., 1km). The application proves valuable in tracing the origins of food-borne outbreaks by pinpointing potential sources of contagion, such as supermarkets and/or restaurants, located at the same distance from individuals affected by the outbreak.

BfR conducted a training on the CAPRI model, a food security model, with the aim of assessing its potential in the field of risk assessment. Regrettably, no practical application was identified in this domain.

The web application "FoodProcess Web App" was improved by implementing features such as drop-down menus for ingredients, upload of custom ingredients, development of a product code, development of a metanode feature (i.e. a node containing further nodes) and a timetable for temperature imported via CSV file. It was used to create a database of traditional Montenegrin  food recipes.

 

  •  Jupyter notebooks
  1. Semantic comparison (Annex 1. Semantic-comparison-faiss.pdf / Intermediate report 2023): This script runs an AI model that enables the mapping of different terms and sentences. This can reduce the workload of recurrent tasks.
  2. Web scrapping (Annex 2. Web-scrapping.pdf / Intermediate report 2023): This script was generated to improve the efficiency of collecting relevant food data.
  3. Multilingual Text Translation (Annex 3. Translation.pdf / Intermediate report 2023): This script demonstrates multilingual text translation using the Hugging Face Transformers library and the M2M100 model. Additionally, this code supports transliterating text between cyrillic and latin text for those languages that allows both methods.
  4. Image Captioning (Annex 4. Food-image-recognition.pdf / Intermediate report 2023): The code snippet in this section illustrates the utilization of BLIP models for image captioning tasks. The resulting captions provide insights into the models’ ability to generate textual descriptions corresponding to visual content.
  5. Restaurant Location Search (Annex 5. Geolocation.pdf / Intermediate report 2023):  This script allows searching and viewing characteristics of culinary dishes such as location, ingredientes, vegetarian dishes among others in a database.
  6. Interactive Map Creation with OpenStreetMap Data (Annex 5. Geolocation.pdf / Intermediate report 2023): This script creates an interactive map and incorporates markers for specified addresses, blue circles indicating a radius to be specified and additional markers for nearby food-related establishments.
  7. Local Establishments Query (Annex 5. Geolocation.pdf / Intermediate report 2023): Utilizing the Google Custom Search API, this script queries local establishments like bars, restaurants, and supermarkets in a region. The API key and custom search engine ID are specified for authentication, and a search query is defined.
  8. Edamam Recipe Search API Integration (Annex 6. Recipes.pdf / Intermediate report 2023): This script enables to extract food data from Edamam recipe search API.
  9. Recipe Information Retrieval and Display (Annex 6. Recipes.pdf / Intermediate report 2023): These scripts illustrate how to perform recipe procedure retrieval, information retrieval and display. Functionality can be further enhanced to include additional recipe-related features, improving user experience.
  10. Recipe Chatbot (Annex 6. Recipes.pdf / Intermediate report 2023): This script creates a chatbot capable of answering questions about recipes. The structured approach involves interacting with the model using a predefined set of questions and a dictionary containing recipe-related content.