FoodChain-Lab Introduction

FoodChain-Lab Concepts 1

Delivery: Something sent from A to B at a certain date. A delivery can have preceding and subsequent deliveries (e.g. strawberry-delivery -> strawberry-cake-delivery).
Station: Any food business operator, that sends and/or receives deliveries.
Trace: The path a contamination can take. A station/delivery “B” is on the forward trace of a station/delivery “A”, if a contamination at “A” can spread to “B” via the food chain network. If “B” is on the forward trace of “A”, then “A” is on the backward trace of “B”.

Weight: Weights are assigned to stations/deliveries, that are involved in an outbreak (e.g. a restaurant where customers got sick). Different weights can be used to model differences between involved stations/deliveries (e.g. higher weight = higher likelihood that station is involved)..
Cross Contamination: When it is applied at a station, its incoming deliveries contaminate its outgoing deliveries. When applied on delivery level, the selected incoming deliveries of station contaminate each others subsequent deliveries.
Kill Contamination: When it is applied at a station/delivery, the contamination is killed there. That means it does not spread to subsequent stations/deliveries.
Score: Is computed based on given weights and cross contamination. Should help to estimate the likelihood that a certain station is the origin of the outbreak (higher score = more/higher weighted stations on forward trace).

s_i is the i-th station or delivery
w_j is the weight of the j-th station or delivery
t_ij has a value of 1, if there is a trace from s_i to s_j and a value of 0 otherwise
n is the total number of stations and deliveries

FoodChain-Lab also allows to assign negative weights to stations/deliveries.
A negative weight should indicate, that a station/delivery is not involved in the outbreak.
When negative weights are used, the score computation changes to this formula.

KNIME is an open source data analytics platform, that allows users to assemble a data pipeline called “workflow”.
A workflow is built by dragging nodes from the Node Repository onto the Workflow Editor and connecting them (https://tech.knime.org/workbench).
Nodes are processing units with input- and/or output ports.
Data is transferred over a connection from an out-port to the in-port of another node.
A comprehensive KNIME quickstart guide can be found at https://tech.knime.org/files/KNIME_q….
An introduction video is available at https://www.youtube.com/watch?v=ft7K….

Detailed descriptions of all nodes are available in the Node Description view of the KNIME workbench (https://tech.knime.org/workbench).
All inputs and outputs are either data tables (triangles) or images (green square). Therefore standard KNIME nodes (Row Filter, Image Port Writer, …) can be used in FoodChain-Lab workflows.

Supply chain data is read from the internal database via the Supply Chain Reader.
This data can be visualized with the Tracing View. The Tracing View also allows to perform a tracing on the data.
The Tracing node performs tracing without visualization. Its output can be used in the Tracing View (e.g. to perform some tracings as a preprocessing step)

The Geocoding node allows to acquire latitude/longitude data from addresses.
This data can be geographically clustered with the GIS Cluster node.
The Tracing View allows geographical visualization, if GIS data is provided from the Shapefile Reader.