At the partner meeting in Aarhus, we (Ghent, Amsterdam, Amsterdam Data Science, Bradford University, Gothenburg) gathered in a working group to discuss ‘Open311’ and ‘Environmental multimedia citizen reporting system’.
Examples
- Amsterdam is handling around 300.000 reports by citizens a year (a range of possible topics, and through a range of channels) and - in close cooperation with Amsterdam Data Science - actively explores the benefits of machine learning (both text based classification and submitted image recognition), mainly to assist the operators that relay, follow up and communicate on such reports. Feedback is provided to the reporting citizen within 3 days. Smart priority/urgency tagging and rules help to manage the stream of incoming reports and complaints in the promised timeframe.
- Ghent has a reporting app for street litter operational, Open311 based, allowing citizens to submit pictures, exact location, define categories of litter (to properly link reports to the responsible city service). Thanks to technical workflows, reports are mapped, integrated and attached to traditional CRM systems, mainly the one of the trash company, efforts were put in smooth “case management”.
- Gothenburg also offers a citizen reporting system.
Decisions
We did not come to one particular system to build, therefore we decided to fall back to:
a) continue inspiring each other with developments in our cities and relevant insights, and
b) formalize the ideation that we started (as a SCORE collaboration the citizen reporting topic remains promising and we hope to reveal the common ground)
We will continue this discussion in the ‘Environment’ challenge working group, based on how the ‘Water’ group progresses, a challenge working group might enable us to continue exploring possibilities more formally and methodically. We hope that the process of facilitating, common solution ideas, validation, prototyping, A/B tests etc. will help us come up with an actual common solution plan driven by multiple SCORE co-developing partners. We count on the D3.1 guidelines to guide our enthousiasm into a tangible and mutual beneficial new product, initiative or proof of concept 
Other notes
As thematic scope we seem to agree on garbage reports and the many ‘shades of grey’ within ‘Trash’ as a reporting cluster. But other citizen report topics (abandonded bikes for instance) were not excluded at this point.
- Amsterdam Data Science is interested in collecting specific image datasets to improve accuracy of visual classification models.
- Both communication (answering, at least giving information on which service the report was relayed to, giving information if it takes some time to process, etc.) and action (what is done about the report visually on the reported location, eg. removing the object that was reported in public domain) are to be considered in a succesful case management workflow.
- As is the case in many technologic applications, for language processing based machine learning, languages and even local nuances in wordings typically used by citizens form a barrier to replicate the reached accuracy of classification models
To be continued.
