This proof of concept was built around a simple question: Which contractors are we working with the most and who is communicating the most between these companies?

Based on email communication logs for 2018, email metadata was extracted from the Microsoft Graph API and stored in a custom database for analysis. The data model for the graphing tool was built by subsetting the email domains for top contractors to isolate emails from/to those organizations and our internal addresses. We started with a list of 10 contractors, and added several more based on global analysis of the entire dataset.

Metadata Data about data; in this case for each email we isolated the timestamp and sender/receivers by domain

Domain The host element in an email address (after the @) - some companies have more than one domain based on global / subsidiary relationships

Connection For each 2 email addresses, a connection is a sent/received relationship. A single email will generate a connection for each recipient. Once a connection is established, its weight increases for every communication logged.

1000000

Total Emails for 2018

=

10000000

Connections

CIRCLES (NODES)

Each circle depicts a user. The color depicts the company the user belongs within with grey for internal nodes. The size of the circle depicts the total number of connections by weight.

LINES (LINKS)

The weight of the line between two nodes indicates the volume of connection between any pair of nodes. If the line is dashed, it means that communication is largely one-sided.

FILTERS

CHART

The chart from hovering over a node will show the direction and volume of communications logged for each connection. The left side is received and right side is sent.

Single Company View

The top 5 internal nodes and top 5 external nodes are displayed in the inner ring. Curved links denote a connection to a top connection for this company.