How do the toolboxes work?
Toolboxes provide processing, analytical and visualization functionalities that can be used and composed in complex processes in an integrated way through the Track & Know platform. Each toolbox takes as input mobility data, in most cases trajectories of vehicles or information derived from them, and produce either enriched data or analytical models of various types: Machine Learning mobility prediction models, driving behaviors recognition, individual mobility models, events prediction, etc. Toolboxes read their inputs either in streaming (e.g. through Kafka connectors technology) or from Big Data storage servers (e.g. MongoDB), and provide their outputs through the same channels, making the exchange of data and models across different toolboxes easy and efficient.
The BDP Toolbox provides (a) an online pipeline that receives streaming GPS traces and performs the necessary pre-processing (cleansing, map-matching, enrichment) towards the generation of semantic trajectories and (b) a set of tools that big data developers can exploit in order to interact with different NoSQL storage solutions through a common API.
The BDA Toolbox combines advanced data analytics and machine learning techniques in order to provide invaluable services, including: driver behavior analysis, individual mobility modeling for crash prediction, and future location prediction.
The CER Toolbox targets complex online recognition of events, such as over-speeding or dangerous driving, based on hand-crafted event patterns, by exploiting enriched representations of streaming GPS traces. In addition, the CER Toolbox offers methods for online learning of complex event patterns, in case these patterns are unknown and need to be learnt from the data. The VA Toolbox integrates visual analysis methods that target mobility data and support various tasks including: assessing data quality, exploring the data, identifying patterns in the data, building models that explain the data, and monitoring current situation against the models.
How can I access this research?
How can I use the results in my business?
Get in touch with us for possible collaboration. >> More info
How is the Track and Know data stored?
Track and Know stores its experimental data in encrypted data stores on the platform. The platform is hosted by the Hertzner Cloud, an EU-only cloud provider, that provides for GDPR compliant service.
How does Track and Know address GDPR?
Where possible the Track and Know project adopts aggressive PII data minimisation approaches. Our algorithms and toolboxes are for analytical insights and are not designed to support automated decision pipelines. The data controllers meet the legal basis for processing this data on the basis of consent, contract (insurance and fleet pilots), public task (medical pilot), and legitimate interest. Those citizens who have/will directly participate with us, can withdraw their consent at any time.
How is people’s personal data being used in the project and can you trace/identify people from the data?
No data anonymization process is perfect. Where possible PII has been stripped out of the data and location information has been obfuscated. The medical pilot requires the Track and Know team to have patient Age, Sex, BMI, Residential Postcode, and Condition Severity to enable adequate insights. By Abstracting the Postcode to fewer digits (e.g. SW1A instead of SW1A 1AA) ensures that a journey record can not lead back to an individual citizen.
How can I run my data on the Track and Know platform? / toolboxes?
If the toolbox has been made public, you can clone the repositories and run locally. If you would like access to the platform please get in touch with the project coordinator and we can work out an arrangement. >> More info: LINK 1 – LINK 2
How do I get involved?
If you would like to learn more please check our events page and/or contact the project coordinator using the form on the contact page.