Natural language processing (NLP) is among the hottest topics in data and AI today. Rapid innovation and NLP adoption stem from a pressing need to streamline business operations for better customer experience. Deeper penetration of technologies — such as cloud computing, artificial intelligence and machine learning — is paving the way for subdomains like NLP.

Jon Campbell, Director of Library Science for the Media Solutions Group at Moody’s Analytics, has been actively engaged in NLP since long before it became a hot topic in the data and analytics space. In an interview with CDO Magazine, he highlights the concept of NLP, the NewsEdge NLP Engine from Moody’s, its key benefits and diverse business use cases.

What is Natural Language Processing?

As Campbell defines it, natural language processing uses computers to analyze text. The NewsEdge team’s goal is to enrich news and textual data with additional signals such as industries, subjects, locations, important people, sentiment, event classification and news clustering.

“Our core engine processes news through a high-speed pipeline, so we can enable critical decision-making using news,” Campbell explains. Over the years, NewsEdge has had customers seeking solutions for specific use cases. The platform user could be a news publisher wanting to categorize and tag its content or an algorithmic trading firm with high confidentiality standards that wants to run the software in-house.

Steady demand, a series of custom projects, technology improvements, and the advent of the cloud led to the creation of the NewsEdge NLP Engine. It is used by Moody’s Analytics customers and partners — and Moody’s Analytics itself — to process textual content.

According to Campbell, benefits that set NewsEdge NLP Engine apart from other NLP solutions are:

• Active feedback loops with customers, QA staff, and editorial teams to constantly tune the models to make them better, analyze the outputs, and further improve the product.

• Complete transparency and explainability showcase why classifications or metadata tags were applied to the data, driving fair business practices and governance.

• Subscription contracts with a flat fee allow users to use the tool as much as they require instead of paying per transaction.

• Maintenance-free model. Customers don’t need to deploy additional personnel, such as software developers, data scientists, and editorial and library science staff. They receive regular updates without maintenance.

• It can run entirely in-house without any external communication, if necessary.

Elaborating further, Campbell says that the solution matches the high confidentiality standards certain businesses and government organizations set. Companies can analyze their internal news feeds, emails, and research reports, and government security organizations can analyze large amounts of external documents for security risks. 

Either way, the created metadata stays private without any external communication from the software. Additionally, customers can construct APIs to access JSON files and draw insights via a news or intelligence application, dashboard, widget, or alerts.

Next, Campbell explains how this can work for a global construction company, for example, by combining two separate information sources — a news story in the public domain and an internal email — to create relevant business.

The entity extraction and categorization from the text of a news story highlight intelligence such as competitor information, risk, locations, relevant topics, and geopolitical events.

The story in the public domain is followed by an internal email from someone who has insight into the topic. Processing the news and the email in a combined manner leads to exciting and actionable results.

“We process close to a million news stories a day,” Campbell says. “A story like this will categorize in less than a second, so decision-makers can use the created insight immediately.”

To learn more about the NewsEdge NLP Engine, visit  www.newsedge.com.

About Jon Campbell

Jon Campbell is the Head of Library Science for the Media Solutions team within Moody’s Analytics. He leads news data enrichment efforts, centered around using natural language processing, (NLP), artificial intelligence (AI) and machine learning (ML) to generate news-derived insights. Throughout his career, Jon has focused using news and other text-based information to generate meaningful intelligence for business decision-making.

Jon has been involved in news aggregation and value-added metadata markup since 2000, where he was an early contributor to dot-com pioneer ScreamingMedia, as well as holding information management positions in Liz Claiborne Corporation, Williams-Sonoma and Restoration Hardware. He joined Moody’s Analytics as part of its acquisition of Acquire Media, a leader in real-time news solutions.