Latest News

Five Things to Know About Key Point Analysis: The AI Innovation Powering “That's Debatable”

October 09, 2020

Tonight, audiences around the world will get a first look at the next generation of Natural Language Processing (NLP) capabilities developed for IBM Watson. The technology, called...

October 9, 2020 

Tonight, audiences around the world will get a first look at the next generation of Natural Language Processing (NLP) capabilities developed for IBM Watson. The technology, called Key Point Analysis, will be used on “That’s Debatable,” an interactive limited-run Bloomberg Television series presented in partnership with Intelligence Squared U.S. and sponsored by IBM. The episode will feature a lively debate, as former Secretary of Labor Robert Reich and former Greece Finance Minister Yanis Varoufakis square off against former Treasury Secretary Larry Summers and Manhattan Institute Senior Fellow Allison Schrager on the question of wealth redistribution. It will also showcase an analysis of viewer-submitted arguments using Watson, providing insight into global public opinion on the debate topic.

Here are five things to know about the latest NLP advancement and how it could be used to benefit citizens and businesses alike.

What is Key Point Analysis?

Key Point Analysis is the next generation of “extractive summarization,” which processes comments, opinions and statements in a set of text documents to select and summarize the most significant points. The technology is designed to examine large bodies of complex documents and produce a ranked list of summarized key points—a concise, data-driven list of information users can quickly act on.

Why is this technology so important now?

In a time of great polarization, there is a need to find more effective ways to communicate, whether among citizens and government, employees and managers or businesses and customers. Organizations are awash in a flood of data, and they need clear views of relevant insights and considerations to make data-driven decisions. Key Point Analysis can help enable that.

How does it work?

Key Point Analysis evolved from extractive summarization capabilities first used with Project Debater, the first AI system capable of debating humans on complex topics. Project Debater included advanced NLP capabilities—like understanding English idioms and summarizing and clustering topics—that IBM is integrating into Watson NLP products. Created by the IBM Research Project Debater team, Key Point Analysis takes summarization capabilities a step further. It examines documents and selects and grades high-quality passages to create a list of relevant points. It then filters out redundancies and incoherent or highly emotional statements and identifies how many sentences in the document set support each key point. After filtering the results again to cover the relevant information, it creates a ranked list of the summarized points, each with a prevalence score and examples of illustrative statements.

Why is IBM Watson being used in “That’s Debatable”?

Hosted by journalist John Donvan, the show blends the traditional Oxford-style debate model with today’s cutting-edge AI innovations. In “That’s Debatable,” IBM is showcasing one of the many ways AI can be used for good by helping people understand nuanced viewpoints, bringing out critical insights and informing high-stakes decision-making. For the debut episode, the technology examined more than 3,500 crowdsourced submissions for and against the motion “it’s time to redistribute the wealth,” showing how AI can help people think about competing opinions and complex topics.

How can Key Point Analysis help companies and citizens?

For businesses, there are several potential use cases. Take customer feedback. Businesses can use surveys, reviews, social media, call center logs and field reports to gain insight, but that’s tens of thousands of documents. And the top few concerns voiced by the largest number of customers can drown out other significant emerging issues. By offering a ranked, quantitative result, Key Point Analysis allows enterprises to see a comprehensive range of concerns, along with their prevalence, and get a more data-driven picture to guide their actions. Similarly, employers can gain insight into employee engagement, and governments can learn more about constituent needs and views.

“That’s Debatable” airs on Oct. 9 at 7:00 pm ET on Bloomberg TV, with a replay available Oct. 10 at Join the debate for Episode 2, on the motion “a U.S.-China space race is good for humanity,” by submitting an argument by Oct. 17 at

To learn more about “That’s Debatable” and the AI advancements powering the debate, visit


Statements regarding IBM’s future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only.

Article Categories