Crowd4SDG is a three-year Horizon 2020 Research and Innovation Action supported by the European Commission’s Science with and for Society (SwafS) programme.
Through an innovation cycle called GEAR (Gather, Evaluate, Accelerate, Refine), the transdisciplinary Crowd4SDG consortium of six partners will promote the development of citizen science projects aimed at tackling the SDG’s, with a focus on climate action. Its goal is to assess the usefulness of practical innovations developed by the teams and to research how AI applications can enhance and provide effective monitoring of SDG targets and indicators by citizens.
Learn more here: https://crowd4sdg.eu/
Crowd4SDG proposes three one-year cycles following a GEAR Methodology to iteratively develop and test new citizen science projects. Each GEAR cycle includes online coaching and in-person challenge-based innovation and is composed of 4 different phases: GATHER, EVALUATE, ACCELERATE AND REFINE.
Each phase of the GEAR methodology filters participants’ fields by a factor of about five while helping the projects advance towards practical deployment. Recognizing that some deserving projects may not pass through the filter, Crowd4SDG will develop and provide a series of guidelines for running local Challenge Based Innovation Training events. It will also actively reach out to partners in maker spaces in Europe and around the world that can host these events, with technical support and guidance from the Crowd4SDG partners.
The CS projects developed in the three GEAR cycles of Crowd4SDG will aim to address Climate Action (SDG 13). However, each GEAR cycle will explore a specific sustainability dimension of climate preparedness in connection with other SDG: sustainable cities (SDG 11), women empowerment (SDG 5), and human rights (SDG 16).
In this phase of GEAR, a call for CS projects through the O17 Challenge, on a specific SDG theme is launched and widely publicized, notably through related EU support actions such as EU-Citizen Science. This phase lasts ten weeks, including a two-week period where a committee selects a set of 50 participants from a pool of 250 applicants based on a series of objective criteria.
In the second phase of GEAR, the selected participants take part in the O17 Challenge 5-week coaching programme, to develop their CS ideas in virtual teams towards compelling pitches. This phase aims to challenge participants with real-world constraints that their CS projects would face if deployed. A panel judges the pitches in the final week of the coaching programme.
In the third phase, between 10 and 20 participants, corresponding to 2-4 projects selected from the O17 Challenge, are chosen based on both the quality of their projects and specific soft skills demonstrated during the coaching sessions. They are invited to participate in a two-week intensive Challenge Based Innovationworkshop at CERN. Other participants are encouraged to develop their projects locally in satellite events held in parallel with the Challenge Based Innovation workshop at CERN, using similar methods.
In the final phase, two participants representing the most promising projects from the Accelerate phase are invited to present themselves during a two-day international event on SDGs held in Geneva or Paris. Representatives of various stakeholders (UN agencies, National Statistical Offices, academic CS experts, private sector, and NGO representatives) provide the projects with concrete feedback. In this phase, Crowd4SDG partners work with regional incubators for technology and social innovation to deliver the projects with substantial opportunities for subsequent development.
The Citizen Science Solution Kit (CSSK) is a set of tools for developing and running Citizen Science (CS) projects, maintained by the Crowd4SDG partners. The tools enable anyone to design and launch their own CS project, and support teams that are developing innovative CS projects. Some of the tools are being enhanced with AI features by the Crowd4SDG partners. The list below is updated regularly, as new tools are added or enhanced. All the tools are open source projects.
GEAR Cycle II participants can use the CSSK tools for a variety of purposes, such as self-organization within their teams, extracting relevant social media content for an ongoing event, utilizing crowdsourcing to annotate gathered images, and interpreting annotations by AI. Before going into the detail of the tools, here is an example case story:
Jane Doe wants to start a Citizen Science project to help her community in Ruritania be better prepared in case a flooding occurs as a result of climate change. In order to understand what the needs of the community are, she uses Decidim4CS, where she can easily set up a homepage and blog for the project, receive the proposals from the community, organize meetings and request citizens to vote. After three months of debating and self-organizing through Decidim4CS, it has been agreed with the community that better data is needed to make adequate decisions. In particular, they have co-decided that a map of the places which are more likely to get damaged when floods occur would be a very valuable asset for the project, and they are willing to work together to make it. They agree to build the map based on information of the last two floods that took place in Ruritania. By means of VisualCit they crawl the tweets containing images from those two floods, filter those images which do not contain images of the floods, and geolocate the tweets. After that, they are left with several thousand images. They create a project in Project Builder, upload the images and request a set of volunteers to connect to Project Builder and help them classify the images by labeling them with one out of five different levels of damage. They decide that, to improve the quality of the data, each image will be labeled by five different volunteers. Once the volunteers have labeled all the images, they use Crowdnalysis to provide them with advanced AI models to go from labels provided by each of the volunteers to a consensus opinion which takes into account the accuracy of the different volunteers, or specific characteristics of the images. After a consensus labeling is established for each image, VisualCit helps visualize a map coloring the different regions of Ruritania with different intensities based on their likelihood to get damage in a flood and visualizing the associated images from the last two floods. Based on the map, Jane decides….
The Citizen Science Project Builder (CSPB) is a web-based tool that allows volunteers to participate in complex data classification tasks that automatic tools cannot handle. It supports projects where citizens can analyze or enrich existing data, typically large sets of images or texts, such as satellite pictures or social media posts, as well as other media formats such as videos and scanned documents.
CSPB also enables the development of CS projects that involve data classification, using a project-building interface that does not require any coding skills. The web interface is based on Crowdcrafting, a project launched in 2011 by Citizen Cyberlab, which with its underlying PyBossa open source software framework was spun out as part of the European SME SciFabric in 2015. The CSPB software is publicly available under the ‘CitizenScienceCenter’ organisation on Github.
Getting first-hand information about an emergency situation while it is happening is important, but is often difficult to obtain in a timely fashion. The image-based social sensing tool VisualCit allows extraction of visual evidence about a situation from Twitter by searching for images posted and geolocating them. Using AI methods, VisualCit enables the user to crawl Twitter with user-defined keywords to search for posts with images.
VisualCit can apply selected filters (e.g. contains photo, occurs outdoors etc.). It can associate locations to posts, even if tweets are not natively geolocated. Posts can be evaluated by crowdsourcing initiatives using the PyBossa platform (same technology as Citizen Science Project Builder above). A collection of images for a location or thematic maps can be created to support interested users.
Co-creation of citizen science projects requires citizens and scientists to self-organize, propose and discuss ideas, schedule meetings, conduct surveys, and much more. decidim4CS is a digital platform for participatory citizen science. It allows citizen scientists to organize themselves democratically by making proposals, attending online meetings, making decisions through different forms of digital voting, and monitoring the implementations of these decisions.
Decidim4cs is based on decidim, a free open-source software originally created by the Barcelona City Hall as a participatory democracy platform for cities and organizations. Open-source code of the tool and further information can be found on GitHub.
CS Logger is an open source data collection platform that makes it easy for anyone (without prior programming or design experience) to build and configure customized mobile apps (iOS and Android) for their CS projects. The Apps can feature the most common functionalities for “data collection” CS projects, for example taking geo-located images and adding additional information based on survey questions.
As for the CS Project Builder, a simple web interface offers a menu of several ready-to-use components for the App, including geolocated image/video/audio and several kinds of text based questions/entries.
The implementation is based on the integration of an existing open source solution, Mind-Logger, developed by the ChildMind Institute (New York, US). The development is done in partnership with the ETH Library Lab (at ETHZ). The plan is to integrate the PB and Smartphone App backends to provide users with a unique infrastructure where data collected with CS Logger can be seamlessly analysed in the CS Project Builder.
CoSo (Collaborative Sonar) is a smartphone application aimed at understanding how team interactions impact team performance and learning. How do team members collaborate? How are subgroups formed? How do these interactions lead to better learning, productivity, creativity, and success? CoSo allows team members to journal the tasks they work on during the course of their project, along with the collaborators involved. In addition, CoSo allows to send longer-form surveys to collect answers about qualitative team features such as diversity (demographic, skills) or organization (roles, relationships). CoSo is combined with a web dashboard for teams to visualize their own data, empowering them with a meta-cognition of their collaborative processes. Developped in the context of the iGEM student competition, CoSo is of general use for any team study geared at understanding how dynamic task allocation and team organization underlies team performance.
The SDG in Progress platform allows project developers to document ongoing projects, or to get inspired by other people’s projects, re-use them or re-purpose them. The platform is based on Build in Progress, originally developed as an open source tool by MIT Media Lab. Compared to other documentation platforms (wikis, Github, etc.), SDG in Progress provides a highly visual overview of how a project is conceived and iteratively improved. It allows for easy visual documentation of the sort of branching that naturally occurs in projects, where different options are explored.
The goal of SDG in Progress is to provide an open repository that charts the step-by-step development of SDG projects, many involving citizen science tools and methodologies, as well as more general crowdsourcing and open science techniques. The idea of SDG in Progress is to document creativity, and support sustainable innovation.
According to the European Commission Horizon 2020 Manual, “Deliverables are additional outputs (e.g. information, special report, a technical diagram brochure, list, a software milestone or other building block of the action) that must be produced at a given moment during the action”.
Crowd4SDG is going to produce 26 public deliverables, from May 2020 to April 2023, distributed among the project’s 6 Work Packages. They will be made available hereunder as the project progresses.
All deliverables can be found here: https://crowd4sdg.eu/about-2/deliverables/
The following scores were calculated using a statistically-driven machine-learning approach, a type of AI that learns to perform a task by analysing patterns in data. This is an experimental approach to citizen-science impact assessment, and the exact reasoning behind the scores is not explainable. The scores represent a best guess of the impact the project is having in each domain. Scores are recalculated and updated when “View impact report” is clicked.
Proportion of questions answered in each domain.