Francisco Webber, Co-Founder & CEO of Cortical.io – Interview Series

Francisco Webber is co-founder and CEO of Cortical.io and inventor of the company’s proprietary Retina technology. This technology applies the principles of cerebral processing to machine learning and natural language understanding (NLU) to solve real-world use cases related to big text data. Cortical.io solutions are based on the actual meaning of text, rather than on statistical occurrences.

Francisco’s interest in information technology developed during his medical studies, when he was involved in medical data processing. Over the course of two decades, he explored search engine technologies and documentation systems in various contexts but became increasingly frustrated with the limitations of state-of-the-art statistical methods.

In 2005 he founded Matrixware Information Services, which developed the first standardized global patent database. In 2007 he set up the Information Retrieval Facility with leading academic expert in information retrieval Professor Keith van Rijsbergen. The Facility aimed at bridging the gap between academia and industry in the field of information retrieval. Francisco recognized that the brain was the only existing reference system when it came to processing natural language. While closely following developments in neuroscience, he formulated his theory of Semantic Folding, which models how the brain represents language data.

What initially attracted you to become involved in Natural Language Understanding (NLU)?

My interest in information technology developed during my medical studies, when I was involved in medical data processing. Over the course of two decades, I explored search engine technologies and documentation systems in various contexts but became increasingly frustrated with the limitations of state-of-the-art statistical methods. I recognized that the brain was the only existing reference system when it came to processing natural language.

You are both the founder and CEO of Cortical.io. Could you share the genesis story behind this company?

We knew that we had a technology that would enable businesses to search more quickly, accurately, and effectively, to extract, annotate and analyze information from unstructured text. Our AI-based NLU technology could be a real differentiator in addressing the challenge of processing unstructured text data as it understands the meaning of sentences, paragraphs, and concepts, not just keywords and consequently delivers highly accurate results. As industry leaders learned about what we had, it opened the door for us to test our approach under real and complex requirements. This resulted in numerous Fortune 500 business successes. These have become the models for our contract and message intelligence products.

The Cortical.io technology is based on the Semantic Folding Theory (SFT) whitepaper, an innovative computational theory for the processing of language data. What are some of the current limitations with how most AI tackles NLU?

Standard AI and ML approaches involve building large-scale statistical models that are trained using massive amounts of training data. Oftentimes, sufficient amounts of training data are not readily available, nor have the subject matter experts the time to annotate such large quantities of documents. In addition, training large models (with billions of parameters) require huge computing resources, which is every expensive and consumes a lot of energy. A typical language model consumes more energy than five cars over their entire lifetime!

The Semantic Folding Theory is built on top of the Hierarchical Temporal Memory Theory (HTM). Could you briefly describe what this theory is?

Hierarchical Temporal Memory (HTM) is a biologically constrained theory of intelligence originally described in Jeff Hawkins’ book On Intelligence. It is a machine intelligence framework strictly based on neuroscience and the physiology and interaction of neurons in the brain. In this theory, the brain acts as a memory system that stores patterns and makes predictions out of current experiences based on memorized patterns. To be truly intelligent, machines must recreate this memory-prediction system. Hawkins describes sparse distributed representations as the sole data format used by the neocortex. The semantic fingerprints used in our technology are sparse distributed representations of text (of words, sentences, or paragraphs).

Could you share with us how the SFT method is superior to competing NLU strategies?

The Semantic Folding Theory backing Cortical.io technology has several big advantages. First, it does not require a huge repository of sample data do deliver accurate results. Secondly, it is fast, and it requires fewer computing resources than state-of-the-art machine learning algorithms. Finally, the training process is much simpler and can be managed by the subject matter experts who will be using the system, even if they have no AI experience.

One of Cortical.io’s products is Message Intelligence, what precisely is this application?

Message Intelligence fits into the category of known as Intelligent Document Processing. Corporations are overwhelmed with processing a wide range of messages and documents. Message Intelligence is an out-of-the-box solution that uses Cortical.io’s technology to accurately extract and classify key information in these messages and documents and appropriately process them at enterprise scale. Because of Cortical.io’s patented approach enables the development of custom classifiers in hours and extraction models in days.

Another product on offer is Contract Intelligence, could you share some details on this?

Cortical.io Contract Intelligence leverages a new approach to natural language understanding to analyze relevant information from large quantities of documents quickly and with an accuracy that is difficult to achieve at scale with manual labor or with other contract analysis tools. The solution offers a powerful, meaning-based tool to extract and classify key information, search within individual documents or over the entire database, as well as to compare documents against a template or previous versions.

Is there anything else that you would like to share about Cortical.io?

We are excited to see the maturation of Artificial Intelligence and Natural Language Understanding from an interesting technology to its integration into application that solve real world business problems.

Thank you for the great interview, readers who wish to learn more should visit Cortical.io.