Object Type-based Data Description by Aggregation for Graph Databases
Adam Kleinedler, Adam Dudáš · 2025
Aggregation of attribute values into a set of computed measures is one of the most widely used methods for a brief description of data in the initial phases of data analysis. This description by aggregation poses no problems in conventionally structured relational data, but issues arise as soon as aggregation functions run on non-relational semi-structured or unstructured data, such as the data stored in document or graph databases. Since in this work the latter of the two database types is considered, this study proposes design and implementation of two-phase aggregation based on object type and its structural sub-types in graph databases based on the Neo4j system. For the evaluational purposes, a synthetic graph database of specific structure is prepared, and the proposed aggregation model is examined and compared to the human-based aggregation of fields of the database objects.