the cognistx blog

Artificial Intelligence for Question Answering

October 20, 2022
By
Sanjay Chopra | CEO, Cognistx

Introducing SQUARE

Sanjay Chopra, CEO of Cognistx

Overview & Background

Institutions and organizations are inundated with text files, tables and lots of unstructured data. Information exists within organizations in emails, PDF documents, word files, Google docs, excels, structured relational databases and non-relational databases. If only it was easy to organize all this information and find the right answer to your question or query.

Our goal with SQUARE - “Scalable Question Answering Reference Engine” is to do exactly that. Based on your question, find the best answers and point you to the right source document quickly.  

Our data scientists and product managers under the able leadership of my co-founder Dr. Eric Nyberg have been working on Natural Language Processing (NLP) techniques for a couple of decades to refine our modeling techniques and abilities to bring the best possible answer to you. We have utilized years of research and our experience with Kroll, Society of Automotive Engineers (SAE), Odds on Compliance, Iconekta, Schneider Downs and Solvaire to present an intuitive and very easy to use interface to answer your questions from your corpus of documents and your databases. In some cases we are the engine that powers their solutions via API calls and related interfaces.

Our SQUARE Team

Our SQUARE team comprises of world class researchers, data scientists and product managers. They have been looking at QA problems for over two decades for various organization types. Now we have a solution which we can deploy quickly with very high accuracy.

Our Solution: SQUARE: “Scalable Question Answering Reference Engine”

What is SQUARE?

SQUARE is an end-to-end platform that automates manually searching information from unstructured documents, URLs, and relational databases (structured data).

Why do we need SQUARE?

Digitalization advancements led to a rapid increase in information storage of structured data like relational databases and unstructured data like documents, images, videos, etc. However, manually searching for information in this data storage is tedious and time-consuming.

How does SQUARE help?

Within a few seconds, SQUARE offers you the answers to all your questions, be it from local files or a domain-specific repository. Let us help you find that “needle in the haystack.”

How does SQUARE work?

The diagram below shows the working of SQUARE. The user utilizes a simple interface to present their query. In the background SQUARE has already organized all the documents and databases to be able to answer your questions correctly and efficiently.

See SQUARE in action for the Legal Services industry:


Who can utilize SQUARE?

SQUARE can be utilized by the following entity types to find answers to your questions/search terms in large text documents, or any urls or databases on the fly.

  • Legal Tech / Law firms
  • Smart chatbots
  • Regulatory and compliance companies
  • Regulatory/compliance departments
  • Pharmaceutical companies
  • Medical Devices (FDA)
  • Banks (FDIC, SEC)
  • Insurance companies
  • Government regulatory bodies
  • Standards organizations
  • QA for pharmaceutical companies

Delivering ROI & Business Value

Key to success is to demonstrate business value and ROI. SQUARE has already proven itself to demonstrate value in the following areas.

  • Aerospace Standards Search . SQUARE powers OnQue™ to first parse complex aerospace standards including requirements, engineering drawings, specification tables and plain text to answer the query for an aerospace engineer as to what standards cover various parts or materials or search for certain key items. OnQue has shown to reduce the search time by 80% with 98.6% accuracy.
  • Needle in a haystack. Kroll needed to parse millions of  documents from the Deep Dark Web (DDW) in a few hours to find critical information or PII that may have been leaked. We were able to do just that and alter customers of data breaches and in some cases predict breaches will happen saving them millions of dollars.
  • Discovery. Solvaire utilizes SQUARE to analyze legal contracts and then answer due diligence questions posed by the lawyers. In a scalable deployment it can save hundreds of legal hours with very high accuracy for critical legal due diligence/discovery questions.
  • Compliance. Odds On Compliance presents online gamling requirements and compliance needs to their customers. SQUARE can help answer dynamic queries by their customers in near real time with 99.2% accuracy for FAQs. This provides significant value to their customers by presenting the right information quickly.
  • Chatbots. Telecommunications companies that have millions of clients asking questions can deliver much higher value and customer satisfaction by deploying SQUARE powered chatbots. SQUARE can also help summarize customer service calls and build knowledge graphs of customer issues as they interact with your organization.
  • Static QA - Information Security Questionnaires. Schneider Downs is planning to utilize SQUARE to help review answers provided by partner companies for banks and other organizations on information security questionnaires. It will again help save hundreds of hours of analysts time and provide the ability to review documents submitted quickly.  

Please note that it is important to build ROI measures into your SQUARE implementation and operations from the beginning. It is equally critical to be able to report on them and learn from them for ongoing improvements and continued executive sponsorship.

Easy to Deploy & Integrate

SQUARE is easy to deploy and integrate. In its simplest form it is hosted on AWS and you can drop all your documents on an S3 bucket. Utilizing DIP it sorts, organizes, clusters, annotates your documents and databases. All you need to do is ask it your question on a very simple interface like Google. It can also integrate with your existing systems via API calls.

To get started and learn more about SQUARE please reach out to Jagriti Pandey at Jagriti@Cognistx.com and we can get a conversation started!

Past Blog Posts