Social Recommendations — Slack, Neo4j and NeoDash

David Stevens
6 min readJan 15, 2022


Small social network graph to identify common groups between people, this builds upon of Employee Fingerprint concept I was working on in my last job and triggered by a post by Michael Hunger on our internal “All nodes” Slack channel; where he shared how to import and work with a subset of our HR dataset (person, department & location).

Could I extend this information with the Slack channels we belong too and make recommendations for teams, new hires and generally mess about with Neo4j and some data?

The starting datamodel from Michael’s post

The simple Slack model

My combined working model

Models drawn using Arrows

Setting up a Slack application/API

Before getting into Neo4j, I need to set-up an API within Slack to provide the information needed to complete the dataset. Here I created a very simple API application with the following User Token Scopes

Once set-up, this is deployed into selected Slack Workspace, where my User tokens can be used to access the Workspace under my registered account — giving me access to the same information I have available within the Slack desktop application.

Importing the Slack data into Neo4j

In total 3 calls are required against the Slack API, the first is to retrieve the list of channels within the Slack workspace, the API is limited to only return public channels within the workspace, so any private channels are not included. The second call takes the channelID and retrieves the list of users who are members of the channel; here Slack provides just the users MemberID, so a 3rd call is required to get the actual name and email addresses of the channel members.

To help with data integrity I set-up a number of Constraints and Indexes.




  • CREATE INDEX ChannelName FOR (n:Channel) ON (
  • CREATE INDEX PersonName FOR (n:Person) ON (

Importing the data from Slack

As the API calls into Slack require an authorization header to be provided, all the dataloads are completed with the APOC procedure apoc.load.jsonParams, which allows for this information to be passed correctly to the API.

List of channels

First collect all the details of the public channels within the Slack Workspace

**Note : URI and Tokens set are parameters to help with the main queries.

:param uri => ""
:param token => "Bearer xoxp-12345567889-123455678" (not a real token ;) )


This will create this simple Graph model for the returned data.

Retrieving the list of members

As mentioned above, I now need to retrieve the list of members of each Slack channel. Here I loop against each channelId and make a series of API calls for each channel.

The Slack API required is

Again, I setup some parameters to help with building out the required API calls.

:Warning: This one can take a little while as it iterates over every channel, pulling in the list of members for each.

:param cMembers => ""
:param cMembersEnd => "&limit=1000&pretty=1"

Member details

The final call to make is to get the name and email of each slack user

New parameters

not sure if the cProfileEnd parameter is actually needed.

Our Slack import is now complete

:param cProfile => ""
:param cProfileEnd => "&pretty=1"

The final step is to import the internal HR dataset, to connect departments, locations and countries to each user. It’s imported at this stage, as the email address is required to ensure the correct information is stored for each person. Whilst I could have imported the Slack data into an existing Graph with the HR data already loading, the series of API’s call required against Slack would have been slightly different.

Gaining value from the collected data.

Whilst importing the data and playing around with the Slack API’s and Neo4j is kind of fun; how can I use this information to bring some value to a team or an individual ?

Neodash is an excellent tool to quickly create PoC’s or prototyping dashboards with no real coding required; just the same Cypher query knowledge used to import the data.

Install NeoDash in your desktop environment via the GraphApp library

The organisation view

A very simple dashboard, but one to just return the total number of people and the most popular channels.

Option to select based on a given country — this is where NeoDash is really powerful, as you can set one of your dashboard views to be a parameter selection for the other charts and views you wish to create.

Team view

Again using the parameter selector, I am able to show the most popular channels for a selected team and the percentage of team members within each channel.

Individual view

This is the opportunity to provide some personal recommendations and not just recommendations based on the team, department or location you reside in; but recommendations based on membership of the groups you belong too.

By selecting the name of current channel the person belongs too, a set of recommendations can be made of other channels the person may wish to join. This is achieved by simplify following the path of “People in this channel are also members of these channels”.

Storing the dashboard

This is very easy, as I can just store the dashboard configuration directly within my Neo4j database, allowing others to both view the results and also review/improve the queries used.


I have to say this was incredibly easy to complete, I started the project thinking I would need some Java scripts to help manage the data loading and sequencing of the API calls against Slack, but the bundled APOC procedure to support API calls with authorization headers made this a very simple and pleasant project to complete. The presentation of the data within NeoDash was also very easy, making this a very low code project.

Originally published at