FarolApp is a mobile web application that aims to increase the awareness of light pollution by generating illustrative maps for cities and by encouraging citizens and public administrations to provide street light information in an ubiquitous and interactive way using online street views. In addition to the maps, FarolApp builds on existing sources to generate and provide up-to-date data by crowdsourced user annotations. Generated data is available as dereferenceable Linked Data resources in several RDF formats and via a queryable SPARQL endpoint. We propose Live Linked Data, a new approach to publish data about city infrastructures trying to keep them synchronized leveraging the collaboration of citizens. The demo presented in this paper illustrates how FarolApp maintains continuously evolving Linked Data that re ect the current status of city street light infrastructures and use that data to generate light pollution maps.
Light pollution is one of the most unknown and rapidly increasing environmental
problems nowadays[
A starting point to reduce light pollution is to increase public awareness about this type of pollution by providing the relevant data in a human consumable manner. Public institutions may implement wellness policies tailored to the current situation and provide open data that may help getting a deeper understanding of this problem. However, to be usable, open data does not only have to be accessible, but also be o ered in a comprehensible way. Linked Data allows to explicitly describe the meaning of data by using a vocabulary and linking them to other data sources.
Cities are living organisms in constant evolution. Live Linked Data, i.e. continuously maintained and updated data, o ers a more realistic representation of ? This research is partially supported by the STARS4ALL project (H2020 - 688135), the 4V project (TIN2013-46238-C4-2-R) and the FPI grant (BES-2014-068449.) a city rather than traditional static RDF dumps. However, in order to do this, someone has to be aware of the changes that occur in the city and re ect them in the data. Citizen science projects could bring some light into this situation.
FarolApp (http://farolapp.linkeddata.es/) is a Linked Data based application whose goal is to increase the awareness of light pollution among citizens and public administrations, presenting pollution maps and allowing citizens to contribute to that information. Citizens can participate both on-site or remotely using Google Street View. In this demo, we present the approach followed by FarolApp and its architecture. A video1 also is available on the project website. 2
Linked Data Generation FarolApp is seeded with existing heterogeneous datasets. On the one hand, if the existing data is already in RDF, its vocabulary is mapped to the FarolApp Vocabulary2 to create a new dataset by using SPARQL construct queries. On the other hand, if the existing data is not in RDF (e.g, CSV, JSON), it is transformed to RDF using external tools such as OpenRe ne. Initially, FarolApp integrates data from ten di erent cities and the seed data is available in datahub3.
Furthermore, during the generation process the data is linked to other relevant datasets. For instance, each street light is associated to the city and the country where it is located based on its GPS coordinates and linked to the corresponding entities in DBpedia. Provenance information are also provided to attribute the data with their original data sources.
Linked Data Publication The main interface of FarolApp is a web UI in which the street light data is overlaid on a map. Furthermore, the app provides pollution maps which are generated based on application data as well as an API where information about street lights can be extracted. Also, this information is published in Social Networks such as Twitter. All these representations use the Linked Data created and maintained by FarolApp as their data source.
In addition, FarolApp provides raw Linked Data as dereferenceable resources (supporting several RDF formats such as RDF/XML, Turtle, JSON-LD using HTTP content negotiation), RDF dumps, and via a SPARQL endpoint, enabling other third-party applications to reuse its up-to-date information. For instance, interlinking street light data with data about energy consumption, tra c accidents, or street crimes may reveal interesting ndings.
Linked Data Evolution With the aim of keeping the data inline with the current state of the street lights infrastructure, it is necessary to get live information from any kind of sensors that are able to communicate their observations. Humans are the most reliable sensors in this scenario, mainly because there is no
common data source, neither a sensor network (e.g. an IoT platform) to date
suitable to build on for: i) identifying street lights whose data is not yet openly
available, ii) updating the most interesting attributes of all of them. Therefore,
this approach leverages crowdsourcing techniques [
Citizens can contribute by giving multiple value annotations for attributes of street lights. The range and type of values depend on the nature of the attribute. In some cases, values expected to be discrete to simplify annotations. For instance, the height attribute, which is normally a real-value, is converted to a categorical value such as low, medium and high.
FarolApp follows a consensus-based approach to conclude a satisfactory level of agreement from all annotations. Two main basic rules support consensus detection: i) a minimum number of annotations is required, ii) the standard deviation must be lower than an attribute-speci c threshold. When both rules are satis ed, FarolApp transforms the agreed value into Linked Data format and enriches the dataset making it evolve. 3
The Farolapp architecture ( g. 1) follows a Staged Event-Driven Architecture
(SEDA) [
Web/Mobile UIs allow users to navigate and annotate the street lights information in an intuitive way using Google Maps. The REST API provides JSONbased descriptions about individual or clusters of street lights. The Loader bootstraps the application with the existing street light data. The Consensus Engine detects and noti es about consensus from annotations of street lights' attributes. The Publisher is in charge of publishing relevant events from FarolApp in social media platforms such as Twitter. The Transcriber persists the updates generated from the consensus engine based on the user annotations.
In addition, FarolApp makes use of a Virtuoso server as a triplestore for providing a SPARQL endpoint, RabbitMQ as a messaging broker, and Redis as an in-memory cache for annotations. 4
Inspired by the DBpedia Live4 approach, which transforms information from
Wikipedia into Linked Data, FarolApp collects data from citizens, transforming
them into Live Linked Data. Other approaches, such as Zooniverse5 (a citizen
science platform for crowdsourcing scienti c research) or OpenStreetMap6 (OSM,
a collaborative platform to create a free editable map of the world) are not
directly available as semantically enriched Linked Data. LOSM[
FarolApp leverages Live Linked Data and serves as a proof of concept of such approach to provide up-to-date street light information and light pollution maps.
As future works we plan to integrate with IoT platforms from smart cities initiatives (e.g. photometers), keep track and o er historical data to study its evolution, and use OSM both as a data source and publisher. Besides, the UI will follow a more educational approach and provide multilingual support.
5 https://www.zooniverse.org/ 6 https://www.openstreetmap.org/