Today, we are going to see one of many examples of the powerful combination of Neo4j & Hume. In this video we will demonstrate the value that graphs and graph databases can bring to a business by surfacing critical insights at scale and in ways not possible in more traditional database formats.

Neo4j is a graph database that is capable of storing many kinds of graph data across various business domains. 

Specific Neo4j Example - Geospatial Data

In this case, we are looking at geographic data in the Neo4j browser as a means to illustrate how leveraging graph with Neo4j and then extending Neo4j’s graph database capabilities with the companion graph product, Hume, would enable a company to unearth meaningful information about immensely valuable patterns locked away in their data. 

In this graph database demo, we will look at geospatial data, which Neo4j stores as points, each of which is made up of latitude and longitude information. 

Points can be stored as properties on a graph node and used in Neo4j’s native spatial and analytic capabilities. 

Nodes can store additional geographic information such as boundaries in the form of polygons commonly found in shapefiles. 

In this Neo4j example graph, two nodes are selected and their geographic information is used to calculate the distance between two points on a map’s model as nodes in a graph. 

Neo4j returns a floating point number representing the linear distance between the two points. 

The returned units will be the same as those of the point coordinates and it will work for both 2D and 3D Cartesian points. 

Hume & Neo4j Demo Combined Example

Hume as a graph insights engine capable of working with graph data stored in Neo4j and out of the box includes the ability to project standard, time series, and geospatial data from Neo4j for advanced analytics. 

Here, city bike station data in New Jersey is viewed in the Hume browser and their locations are projected on Hume’s map visualization. 

You may have seen this data earlier on Graphable’s YouTube channel where we explore bike share data using Hume. 

Graphable is occasionally asked by a client to extend Hume to meet their particular use cases. 

Next, continuing with our geospatial theme as an example, we are going to look at how Graphable’s development team integrated Uber’s H3 framework for a client using Hume’s extensibility. 

We created a dockerized container service that wrapped Uber’s H3 spatial index framework to enable geospatial analytics at different levels of resolution. 

Hume can be extended for many use cases beyond geospatial. 

We used Hume’s ability to easily store custom functions called actions, which any user of Hume’s interface can call without needing to know the underlying code. 

In this visualization, we start by displaying level four hexagons and utilized Hume’s internal context information to change visible hexagons and their underlying data at different resolutions based on the map’s zoom level. 

Using the city bike data to demonstrate the value of Hume’s extensibility, Graphable’s H3 integration enabled users to explore bike stations in New Jersey to display hexagons at a user’s desired level and to project hexagon areas which aggregate the bike stations contained by each hexagon. 

This H3 extension augments the ability to analyze and aggregate data efficiently and at scale and across different data resolution levels for performance analytics. 

Graphable’s H3 integration into Hume enhances Neo4j’s graph native algorithm analytics. 

In the city bike data, hexagons and Hume actions are used to calculate the shortest path between bike stations at various resolutions. 

Using the different levels of data resolution, it can result in the performance planning of new station routes. 

See additional Neo4j Use Cases

More Neo4j Examples - Augmented hume functionality

To provide additional Neo4j examples of how Hume can be extended to meet a business need, Graphable was asked to augment Hume with the ability to provide driving directions for the shortest route between two locations. 

We use the same Docker container that we built Uber’s H3 in. 

We leveraged OSRM, a uniquely powerful open source routing engine for the shortest route in road networks. 

Looking at the OSRM interface, it’s easy to see that if you were in Boston and wanted to get fresh Island Creek Oysters in Duxbury, that you can get both the shortest route and the turn by turn instructions. 

It was easy to take OSRM and integrate it into our containerized service and call it using Hume’s API capabilities. 

From Hume’s interface, a user can easily select two points on the map or even H3 hexagons, and using Hume’s multinode actions view OSRM generated routes and directions. 

Learn more about Uber's H3 Grid System

Final Hume & Neo4j Example

As a final example of how Graphable can enhance Neo4j’s data functionality and extend the Hume platform’s analytic capabilities, we will once again look through the lens of geospatial data, though this is just one example of many possible business use cases. 

The Graphable team responded to a client’s requested feature by incorporating the Python package, GeoPandas into our containerized service. 

GeoPandas is a powerful open source project to make working with geospatial data in Python easier. 

One of Hume’s features that we have discussed is actions. 

Hume actions can be used to call parameterize cypher queries. 

See article on Neo4j Cypher Tutorial

Actions in our GeoPandas use case were used to enable users to create new nodes in a Neo4j graph using the Hume interface without needing to know any underlying cypher code. 

Using actions, users created new vertices nodes to define the border of and create a polygon. 

This capability gave Hume users the ability to quickly partition areas of interest, in this use case, locations in France. 

The partition locations were efficiently filtered from a large data set using Hume functionality and cypher code as part of the supporting actions. 

The GeoPandas service that Graphable built was called to enhance location filtering through its functionality for identifying geographic locations contained within a polygon. 

The combination of new Neo4j data created by the user in the Hume interface and the GeoPandas’ service provides the users with the ability to dynamically create areas of interest in their geographic datasets. 

Hume can use these filtered results in downstream analysis to surface insights that would be difficult in other data platforms. 

Hopefully, this video has been helpful and demonstrated the value of graph databases Neo4j, Hume and Hume’s extensibility. 

For more detailed information, a demo or help, reach out below to connect with our graph engineering and data science experts.

See additional Graph Datatabase Use Cases

Curious about other Neo4j examples? Learn more about Text to Graph Machine Learning.