Periodic Reporting for period 2 - QROWD (QROWD - Because Big Data Integration is Humanly Possible)
Période du rapport: 2018-06-01 au 2019-11-30
Across Europe, we witness a rise in the availability of mobility, road information and urban transportation data, from real-time streams coming from sensors installed all across cities, to data coming from personal devices of citizens and workers. The exploitation of these data sources is expected to greatly enhance the quality of life and overall well-being of European citizens, and stimulate the economy by facilitating novel data-driven services and better informed infrastructure investments and policies. Citizens and workers would benefit from increased accuracy in the traffic information they need, and from efficient and flexible data-driven transport flows. Public administrations can take data-driven decisions for investment and policy changes, being able to effectively evaluate them.
QROWD's overall objective was the development of a socio-technical solution to cross-sectorial Big Data integration in a European urban Smart Transportation, supporting participation and feedback of various stakeholder groups to foster data-driven innovation and synchronized access to Big Data sources. To achieve it, QROWD will create the first hybrid architecture for Big Data integration and analytics that seamlessly combines the scalability and reliability of Big Data processing and social computation, across the entire Big Data Value Chain, by leveraging machine learning algorithms and crowdsourcing services, providing the ability of connecting to different crowds, and synchronizing the varying speed of human and machine computation.
QROWD delivered a platform that implemented the hybrid architecture, and successfully piloted it to (i) estimate modal split with a combination of machine learning analysis and feedback collected from participating citizens (ii) generate maps of bike racks and yellow parking spots with data collected by volunteers on the field, crowdworkers on top of street imagery, Volunteered Geographic Information databases, all combined using QROWD's own fusion and interlinking framework.
QROWD's overall objective was the development of a socio-technical solution to cross-sectorial Big Data integration in a European urban Smart Transportation, supporting participation and feedback of various stakeholder groups to foster data-driven innovation and synchronized access to Big Data sources. To achieve it, QROWD will create the first hybrid architecture for Big Data integration and analytics that seamlessly combines the scalability and reliability of Big Data processing and social computation, across the entire Big Data Value Chain, by leveraging machine learning algorithms and crowdsourcing services, providing the ability of connecting to different crowds, and synchronizing the varying speed of human and machine computation.
QROWD delivered a platform that implemented the hybrid architecture, and successfully piloted it to (i) estimate modal split with a combination of machine learning analysis and feedback collected from participating citizens (ii) generate maps of bike racks and yellow parking spots with data collected by volunteers on the field, crowdworkers on top of street imagery, Volunteered Geographic Information databases, all combined using QROWD's own fusion and interlinking framework.
"=== First period ====
- Elicitation of use case requirements of the Trento Smart City, and preparation of dataset catalog, leading to the precise definition of the two vertical big data value chains to be implemented by the project and informing the development of the rest of the components of the QROWD platform.
- Development of a ""Participatory Framework"" that describes the different types of crowds and ways of incentivisation. The framework serves as a guideline for effective design of tasks to be carried out by humans.
- Implementation of an ideas competition to foster collaboration
- Initial design of the hybrid human-machine architecture of the QROWD platform, as a part of the Big Data Europe Framework.
- Development of several hybrid human-machines tasks based on the elicited requirements, including
- Exploration of virtual spaces defined by street view imagery for aid at the completion of mobility infrastructure maps.
- A platform for gamifying crowdsourcing tasks
- A general system architecture for realizing distributed crowdsourced streaming data quality assessment workflows using a publish/subscribe infrastructure.
- Providing a unique URI for named entities by integrating the crowdsourcing functionalities for disambiguation.
- Development of a set of machine learning classifiers for segmentation of trips and transport mode classification from data streams collected from citizens
== Second period ==
- Piloting of the hybrid human machine workflows for estimating modal split and completing bike rack and yellow parking spot maps in the city of Trento. Results will be ultimately exploited in future app of the municipality.
- Pilot of road information services based on TomTom data, that will be part of the pipleine of TomTom product offers to cities.
- The QROWD platform, based in the architecture developed in the first period.
- Development of a vocabulary to describe datasets collected with crowdsourcing
- Completion of framework for interlinking and fusion datasets
- Development of framework for generating Linked Data, applied to TomTom data
- Development of data acquisition framework for datasets, to be integrated in to the offer of ATOS for services to municipalities
- Final prototype of i-Log, an application for collecting sensor data and ask questions about the collected data
- Research and development of workflow for hybrid multilingual processing"
- Elicitation of use case requirements of the Trento Smart City, and preparation of dataset catalog, leading to the precise definition of the two vertical big data value chains to be implemented by the project and informing the development of the rest of the components of the QROWD platform.
- Development of a ""Participatory Framework"" that describes the different types of crowds and ways of incentivisation. The framework serves as a guideline for effective design of tasks to be carried out by humans.
- Implementation of an ideas competition to foster collaboration
- Initial design of the hybrid human-machine architecture of the QROWD platform, as a part of the Big Data Europe Framework.
- Development of several hybrid human-machines tasks based on the elicited requirements, including
- Exploration of virtual spaces defined by street view imagery for aid at the completion of mobility infrastructure maps.
- A platform for gamifying crowdsourcing tasks
- A general system architecture for realizing distributed crowdsourced streaming data quality assessment workflows using a publish/subscribe infrastructure.
- Providing a unique URI for named entities by integrating the crowdsourcing functionalities for disambiguation.
- Development of a set of machine learning classifiers for segmentation of trips and transport mode classification from data streams collected from citizens
== Second period ==
- Piloting of the hybrid human machine workflows for estimating modal split and completing bike rack and yellow parking spot maps in the city of Trento. Results will be ultimately exploited in future app of the municipality.
- Pilot of road information services based on TomTom data, that will be part of the pipleine of TomTom product offers to cities.
- The QROWD platform, based in the architecture developed in the first period.
- Development of a vocabulary to describe datasets collected with crowdsourcing
- Completion of framework for interlinking and fusion datasets
- Development of framework for generating Linked Data, applied to TomTom data
- Development of data acquisition framework for datasets, to be integrated in to the offer of ATOS for services to municipalities
- Final prototype of i-Log, an application for collecting sensor data and ask questions about the collected data
- Research and development of workflow for hybrid multilingual processing"
The design and ongoing implementation of two hybrid workflows for two urban mobility use cases.
- The first workflow enables the collection of labelled data for segmentation of trips and transport mode detection from citizens that are engaged and rewarded
- From AI perspective, enabling the bootstrapping of machine learning processes from a situation where no training data is available
- From a socio-technical perspective, the workflow enables a virtuous engagement circle with contributing citizens, improving the quality of contributed data and potentially bringing applications closer to the human-centred ideal of Smart Cities.
- The second workflow enables the completion of maps of mobility infrastructure Points of Interests
- Improving on cost and speed over sending experts on the field
- Improving on accuracy and speed over Volunteered Geographical Systems
The workflows are supported by the QROWD architecture, that enables the leverage of the human factor as a part of data processing flows, overcoming the bottleneck of both manual and fully automatic approaches for cross-sectorial data integration and analytics. The integration of the human factor is also expected to increase the number and size of datasets available for analysis.
From a socio-technical point of view, QROWD enables more flexible and personalized interactions with public administrations through services co-created together by citizens (involved not only as end users, but also as collaborators), businesses and public administrations through practicing the quadruple helix approach. Services provided by QROWD provide the cornerstone for the development and assessment of mobility policies for lessening overall congestion in the cities, enhancing energy efficiency and environmental protection, e.g. with respect to the transportation-related energy use and greenhouse gas emissions.
- The first workflow enables the collection of labelled data for segmentation of trips and transport mode detection from citizens that are engaged and rewarded
- From AI perspective, enabling the bootstrapping of machine learning processes from a situation where no training data is available
- From a socio-technical perspective, the workflow enables a virtuous engagement circle with contributing citizens, improving the quality of contributed data and potentially bringing applications closer to the human-centred ideal of Smart Cities.
- The second workflow enables the completion of maps of mobility infrastructure Points of Interests
- Improving on cost and speed over sending experts on the field
- Improving on accuracy and speed over Volunteered Geographical Systems
The workflows are supported by the QROWD architecture, that enables the leverage of the human factor as a part of data processing flows, overcoming the bottleneck of both manual and fully automatic approaches for cross-sectorial data integration and analytics. The integration of the human factor is also expected to increase the number and size of datasets available for analysis.
From a socio-technical point of view, QROWD enables more flexible and personalized interactions with public administrations through services co-created together by citizens (involved not only as end users, but also as collaborators), businesses and public administrations through practicing the quadruple helix approach. Services provided by QROWD provide the cornerstone for the development and assessment of mobility policies for lessening overall congestion in the cities, enhancing energy efficiency and environmental protection, e.g. with respect to the transportation-related energy use and greenhouse gas emissions.