Key Stories: How does clustering work?

Key Stories is a feature of the Signal AI platform that identifies and clusters together similar articles (even when they’re not exact syndicates) in real time to help you quickly understand what storylines have been driving trends in data without needing to sift through thousands of individual articles. How does it work?

Step One: Entity & Topic Extraction

At its core, this capability relies on entity extraction – one of the fundamental features of Signal AI that powers much of the advanced analysis on the platform.

In this process, Signal AIQ, the AI-powered engine behind the Signal AI platform, comes in to perform Knowledge Extraction. Using AI, unstructured, messy content is transformed in real-time into thousands of searchable Concepts which are made up of Topics like Sustainability and Automation, and Entities such as Organisations, People and Locations.

Step Two: Co-mention monitoring & Similarity Detection

AIQ then counts the co-mentions between every entity and topic in the platform and monitors the co-mentions over time. Key stories identifies and focuses on spikes in co-mentions. By doing so, we cut through the noise, overlooking any trivial news such as daily stock ticker announcements. This process distills down about 5m news stories a day to about 150k spikes in co-mentions a day instead.

When AIQ identifies a spike in co-mentions it then analyzes the text itself of all of the articles in which an entity and topic have been mentioned together and begins to decipher the underlying story driving this coverage by understanding what other entities and topics are mentioned and how similar the text is for each story.

Step Three: Story Clustering

Once AIQ has clustered together all of the similar coverage and identified the topics and entities, this distilled information can be used to quickly identify the most important coverage and underlying narrative impacting a brand. This information is presented in a variety of ways throughout the web app.

1. Through a key stories search: Rather than sifting through nearly 5k articles about Tesla and cyber trucks, you can quickly understand, in the example below, the biggest story driving coverage on Oct 19 was Tesla cutting the price of its cyber truck.
   
2. Through dashboards and insights: Understand what storylines have been driving trends in data – identify instantly what’s driving the spike or drop in coverage without needing to sift through individual articles.
   
3. Through Topic Analysis: Topic Analysis allows you to Map your Topic Landscape and discover how closely associated your business or your competitors are to the topics that you care most about. With a single click, you can dive deeper into the key stories that are driving that association.