Previous year's Nominees
Register App Engine Graph API feature (aka Graph API, http://dev.ir.ee/) is a RESTful service providing access to continuously maintained linked open government and commercial data for business applications. More specifically, the data origins from a mix of public and private sources and includes tax debts and other debts, unsubmitted tax declarations and submitted annual reports, contact details, VAT registrations and board members of all companies in Estonia. Additionally some derived data is served, more specifically, company risk scores and predictions. The Graph API provides data is a number of formats like serialisation of RDF in JSON, Turtle and RDF/XML.
Data has been linked and encoded by using ontologies such as DCMI Metadata Terms, Schema.org, GoodRelations, W3C Time Ontology, W3C Organization Ontology, W3C Registered Organization Vocabulary, W3C vCard Ontology, FOAF Vocabulary, Named Graphs vocabulary RDFG, The CPA Ontology, Changeset Vocabulary, W3C PROV Ontology. Furthermore, some of the mentioned ontologies were extended and new ones designed to cover the vocabulary related to taxes, claims, documents and credit management.
The data provided via Graph API is continuously updated. Updates are harvested and encoded into RDF together with provenance data, changesets for the current and previous version of the objects are computed and finally the changesets are appplied to the main graph, which is served via the API. Via usage of provenance data and changesets it is possible to trace back specific source and raw input data of each served triple if this is needed, hence, traceability of data can be ensured and thereby reliability is established.
Graph API is currently used in company credit reports of a due diligence site http://www.inforegister.ee, which provides insight into Estonian usinesses and executives and has ca 40k registered users and 250k unique monthly visitors. Other Graph API users are accounting programs and CRM systems, where data about suppliers and customers is needed. Although the Graph API is a public feature and can be consumed from anywhere in the Internet, we have prepared dedicated runtime environments at Register App Engine cloud infrastructure where applications can access the service with minimal latency. There is also a runtime environment where advanced analytics can be performed by using already prepared monthly timeseries and social network data.
Register's approach is innovative from the perspective that it 1) deliberates the application developers from the data harvesting and maintenance issues for major data sources used in decision making, 2) it reduces the need for data pre-processing in common advanced analytics tasks, by providing a selection of pre-processed time series and graph datasets for maintained data, and 3) it eliminates the learning curve for advanced analytics by proving a selection of analytical models, such as for company risk assessment, for instant usage. Furthermore, since the Graph API feature can be consumed by applications via the same infrastructure as the feature, there is need for data-intensive applications to store and maintain the data outside the infrastructure. Finally, as more and more Register App Engine features are added, the easier it gets to build and deploy linked data applications for business.
Register App Engine is a PaaS offering, which makes data and advanced analytics as easily usable by software developers as spreadsheets made it for financial math and simple modeling for nonexperts in business. This is achieved by means of a collection of services, such as Graph API, Stream API, several runtime environments, running on the same infrastructure. Although so far the most of the effort was went into development of Graph API feature, we are in the phase we start implementing the Stream API feature. The feature exploits computed changesets to stream linked data changes, in parallel to applying them internally, instantly to subscribers. Moreover, the Stream API allow via stream reasoning derivation of business events.