Sept. 17, 2021

Multi-Resolution Representation Using Graph Database

Researcher: Yizhi Huang (MSc) | Supervisor: Dr. E. Stefanakis

A novel approach for storing, querying, and managing multi-resolution representation data using graph database has been developed. Existing methods are: 1) time-consuming in resolution-related indexing (for example, in tile maps sets, retrieving representations for one spatial object may lead to duplicated search in every tile), 2) inflexible in structure, and 3) costly in executing queries. To provide direct indexing and straightforward navigation on resolution, Neo4j, a graph database that is popular for its flexible structure and powerful query execution, has been deployed.

Structure within Neo4j is the semantic translation of traditional map structure, in which nodes are labeled according to the information stored in. Data management could be performed on targeting nodes and relationships individually, without touching others. Information for maps and layers are existing as individual nodes rather than properties within nodes for representations, to avoid information redundancy. Resolution indexing could be directly performed on scale values defined in nodes for both layer and representation, which allows fast retrieving on representations under the same resolution. Navigation between adjacent resolutions is now available through zooming relationships that are explicitly defined between nodes. In addition, acquiring representations for one spatial object could be done within one query. Moreover, combined with a newly developed web-based interface, queries are simplified into keyword-search queries with more intuitive visualization on results, which lowers the barrier between novice users and complex queries.