Data2Neo - A Tool for Complex Neo4j Data Integration

TL;DR

This paper introduces Data2Neo, an open-source Python library for converting relational data into knowledge graphs stored in Neo4j databases, designed to be user-friendly, efficient, and scalable to large datasets.

Abstract

This paper introduces Data2Neo, an open-source Python library for converting relational data into knowledge graphs stored in Neo4j databases. With extensive customization options and support for continuous online data integration from various data sources, Data2Neo is designed to be user-friendly, efficient, and scalable to large datasets. The tool significantly lowers the barrier to entry for creating and using knowledge graphs, making this increasingly popular form of data representation accessible to a wider audience. The code is available at https://github.com/jkminder/data2neo .

Figures (2)

• Figure 1: A schematic overview of data integration with Data2Neo: We are given three input data sources, each represented by a color, with one row per table. The Figure displays three timesteps ($t=0, t=1, t=2$). The user defines a ResourceIterator that iterates over all the input data and specifies an abstract conversion recipe for each data source within a Conversion Schema. We visualize different node labels by colors. The converted data is pushed to the Knowledge Graph. You can also merge information from different sources – the $f$ and $x$ node from the blue and green data source are merged because they have the same value ($0$) and share the same label (as represented by the color). To customize the integration the user can define pre- and postprocessor functions. These functions support arbitrary Python code.
• Figure 2: Example Conversion Task: A simple product sales system is converted into a knowledge graph.