prFrame: A framework for risk analysis based on food provenance
1. prFrame: A framework for risk
analysis based on food
provenance
University of Southampton
Belfrit Victor Batlajery, Mark Weal, Adriane Chapman
{ b.v.batlajery, m.weal, adriane.chapman }@soton.ac.uk
King’s College London
Luc Moreau
luc.moreau@kcl.ac.uk LPDP
5. Problem 3: Optimize the place to
sample for contamination
5
primary
processing
transporting
retailing
transporting
storing
preparing
cooking
6. Required properties
A way to track and trace food, and assess risk
up and down the food supply chain
automatically
6
7. Required properties
A way to track and trace food, and assess risk
up and down the food supply chain automatically
8. Track and trace - Provenance
8[1] https://www.w3.org/TR/prov-dm/
Provenance is a record that describes the people, institutions, entities, and activities involved in producing,
influencing, or delivering a piece of data or a thing in the world.
10. Track and trace – with prFood
3
10[3] Batlajery, B.V., Weal, M., Chapman, A., Moreau, L. prFood: Ontology principles for provenance and risk in the food domain. IEEE ICSC.
2018
12. Required properties
12
A way to track and trace food, and assess risk
up and down the food supply chain automatically
1212
13. Assess Risk - Risk Model
13[4] Nauta, Maarten J. "The modular process risk model (MPRM): a structured approach to food chain exposure assessment." Microbial risk
analysis of foods. American Society of Microbiology, 2008.
Modular Process Risk Model (MPRM) is a risk model that uses risk distribution
in order to understand the likelihood of the presence of bacteria as food travels
through any process in food supply chain.
4
C = level of bacteria
14. Track and trace – with prFood
3
14[3] Batlajery, B.V., Weal, M., Chapman, A., Moreau, L. prFood: Ontology principles for provenance and risk in the food domain. IEEE
ICSC. 2018
15. Required properties
A way to track and trace food, and assess risk
up and down the food supply chain automatically
16. Up and down risk assessment - Belief
propagation
16
?
?
? ?
?
?
?
?
?
!
! !
!
!
!
!
!
Belief Propagation (BP) is an inference technique that relies on an iterative updated belief/message
over the graph representation.
17. Up and down risk assessment - Belief
propagation
17
Obs Inf
Level [X or Y or Z] Level [X] = 0% ≤ K% ≤ 100%
Level [Y] = 0% ≤ K% ≤ 100%
Level [Z] = 0% ≤ K% ≤ 100%
...
Prob (Obs | Inf )
21. Evaluation
21
Pr R T S P C
[5] World Health Organization. (2002). Risk assessments of Salmonella in eggs and broiler chickens (Vol. 2). Food & Agriculture Org.
5
22. Pr R T S P C
prFrame predicts the correct risk for
unobserved food processes.
Sample Result
Pr
23. Conclusion
Provenance is important to capture the history
of food and can be a basis of risk analysis;
hence can support due diligence.
prFrame successfully integrate the food risk
model and the food provenance, and can
potentially be a scientific framework to achieve
due diligence in the food domain.
23
24. Future work
Evaluate prFrame for non-linear food supply chain
(ongoing).
Facilitate moving all food operators in food supply
chain to a standardized digital format.
Develop more enhanced food risk models.
24
25. Interest in collaboration
Apply the model to real data
Aspiration to work with multiple stakeholders
(regulators, food producers, etc.)
Contact us!
Luc Moreau
luc.moreau@kcl.ac.uk
25
26. Acknowledgement
1. The Hampshire County Council Catering Services (HC3S) for a
description of their processes and illustrative data about food
supply chain in Hampshire County, United Kingdom.
2. Indonesia Endowment Fund For Education (LPDP) for supporting
author’s research at University of Southampton, United Kingdom.
26
LPDP
27. Questions and Comments?
For code or models please
contact:
Belfrit Victor Batlajery
b.v.batlajery@soton.ac.uk
For collaboration opportunities
please contact:
Luc Moreau
luc.moreau@kcl.ac.uk
27
