Case Study

Surfacing and categorizing 3 million+ risks for Fortune 500 businesses

We developed a machine learning model that was able to surface, identify and classify 3 million+ risks pertinent to any specific company and industry.

Moon Graphic

The Client

A company developing integrated security, risk management and compliance software for mid to large-sized organizations. Our client’s customers include Fortune 500 businesses, hospitals, academic institutions, high tech, manufacturers, retail and critical infrastructure organizations such as airports and utilities.

The Challenge

The world’s leading companies are required to report new and emerging risks that can negatively impact their businesses. The cost of mitigating a risk increases exponentially over time. The earlier we can identify risk, the lower the cost associated with resolving it.

As an industry-leading compliance software provider, our client challenged us to accurately surface potential risks across millions of online articles, assessing and categorizing each in an efficient and affordable way. Surfacing, identifying and entering risks has traditionally been done manually, relying on hours of review, analysis and processing by internal staff.

The Solution

We analyzed our client’s process and automated it by developing a custom SBERT model that was able to surface, identify and classify 3 million+ risks based on the company and industry. We refined the model to achieve 88% accuracy, categorizing risks in real-time across millions of online articles, SEC filings and publications. Using industry-standard web services, we integrated the model within our client’s monitoring and production infrastructure, making the model available to their customers in a cost-efficient manner.

IP Generated

Crater Labs created a unique Siamese Sentence-BERT (Bidirectional Encoder Representations from Transformers) network to classify and group risk statements.

Benefits & ROI

The project saved millions of dollars in costs and created a new line of revenue for our client, with our model acting as a key point of differentiation.

Keep Reading

More Moonshots Worth Celebrating

Automating document review with 98% accuracy

We used Natural Language Processing (NLP) to automate the internal review of RFPs, RFIs and security questionnaires, reducing time and costs associated with the process.

Read More

Saving $80K+ with employee FAQ generation

We built an NLP-based neural network designed to automate question generation, saving $80K+ in the first year for one of Canada’s leading HR software businesses.

Read More

Energy demand prediction for generation management

We developed a deep reinforcement learning model to assist energy producers in scheduling maintenance and minimizing operating costs.

Read More