We are pleased to announce our latest addition to TRAC - Segments. 

This is a new method of clustering we've developed that groups any data on Pulsar into various clusters (or segments), giving you smarter tools to help you with your analysis. 

With Segments you can easily identify those specific data points that tend to be associated or occur together. These can be groups of authors in a search, based on their bios; groups of keywords or topics; and groups of images, based on the concepts we've extracted for each image. 

By understanding the co-occurence of these variables, you can use Segments to help inform your creative and content planning. For example you can use Segments to plan your content around the specific topic areas we've identified, or in the case of bios, you can tailor your content towards specific groups of people. 

A Brief Summary of our Clustering

Each node (circle) in the Segments graph represents either the following:

The size of each node is a combination of the number of connections plus the volume of occurrences for that node. 

The nodes are then coloured and clustered into distinct clusters (up to 8). And for each cluster, we use Betweeness Centrality, which is a measure of how often a node acts as a bridge between other nodes. In network graph analysis, nodes that have a high betweeness are often important controllers of power or information, and can connect disparate parts of a network. As such we name each cluster based on the node that we’ve identified as having a high betweeness score, as shown below:

You will find Segments in the Audience (Bios), Keywords, Topics and Content sections on TRAC.

You will find Bio Segments in the Audience section of your TRAC search. Use this graph to help identify the main groups of people involved in the conversation, based on the keywords in their bios. Below is an example:

You will find the Keywords or Topics Segments in the Keywords or Topics section of your TRAC search (see below). Use this graph to identify groups of terms, or themes that tend to be associated together in the conversation. This can be useful to help direct your content strategy and adapt similar themes and terms that resonate with your audience. 

You will find Image Segments in the Content tab under Images. NB To view the graph you need to have the Image Tags AI module enabled for that search. 

NB : Since this article was published, we have since introduced Segments 2.0 and you can find out more about our updated segments by clicking on the button below.