[Back]

S. Chawanji:
"Vector tile cache connecting effective spatial communication and geospatial AI";
Supervisor: M. Jobst, G. Gartner; Department für Geodäsie und Geoinformation, FB Kartographie, 2020; final examination: 11-02-2020.

The growth of the Web has been characterised by publication and establishment of linked open data on the Web powered by technologies such as RDF and SPARQL. Much of this data are embedded with geographic
content that links them to a position on Earth´s surface. This data can be extracted from sources such as social media, remote sensing and government portals. The abundance of the data on the Web create a huge pool of spatial big data. By employing geospatial artificial intelligence (geoAI), knowledge can be extracted from the spatial big data that can answer questions to real world phenomena. Since the data is geolinked it can be integrated with geospatially enabled Web services so that it can be visualised on Web GIS applications.
As most datasets on the Web are distributed on different computers connected by the internet, Web services are developed with the purpose of providing interoperable data access, data integration and data processing functionalities. One such Web service is the Open Geospatial Consortium (OGC) Table Joining Service (TJS). TJS provides an interface that can take a geospatial framework on one node, and attribute data on another node and merge them based on common geographic identifiers.
The goal of this thesis is to examine the feasibility of implementing a TJS that uses cached vector tiles as the geospatial framework and Comma Separated Value (CSV) format for attribute data. TJS specification requires that attribute data be formatted in an XML based structure called GDAS. However, most datasets published on the Web are in CSV format. Additionally, RDF data stores can be queried via SPARQL endpoints, the results of these queries are in tabular format and can be converted to CSV data format.
To achieve the thesis goal a prototype implementation of the TJS concept is developed which ingest attribute data in CSV format and cached vector tiles. The two datasets are to be merged based on common
geographic identifiers. Vector tiles are small pre-package containers of vector data. Vector tiles have several
advantages over other means of distributing geographic features through the internet. Vector tiles are small, they can be cached for later use, they are rendered by the client and users are able to interact and use the underlying geographic features for further geographical processing and spatial analysis. The results of this thesis will be displayed as a Web thematic map in a simple OpenLayers Web map application.

Table Joining Service, Vector tile cache, Linked Open Data, Geographic identifier, SPARQL