5 successful cases exploring the business value of natural language processing

Data is now one of the most valuable enterprise commodities. According to CIO.com's "State of the CIO 2022" report, 35% of IT leaders said that data and business analytics will account for the largest share of their organization's IT investments this year, and 58% of respondents said that in the next year They will increase their investment in data analysis.

While data comes in many forms, perhaps the largest, untapped data pool is text, whether it’s patents, product specifications, academic publications, market research, news, or social feeds. Text-based, and the number of texts is constantly growing. According to Foundry's 2022 Data and Analytics Study, 36% of IT leaders believe that managing this unstructured data is one of the biggest challenges they face. That’s why research firm Lux Research points out that natural language processing (NLP) technology—especially topic modeling—is becoming a key tool for unlocking the value of data.

Natural language processing is a branch of artificial intelligence (AI) used to train computers to understand, process and generate language. Search engines, machine translation services, and voice assistants are all powered by natural language processing. Topic modeling is a natural language processing technique that breaks down an idea into subcategories of common concepts defined by phrases. According to Lux Research, topic modeling allows organizations to associate documents with specific topics and then extract data, such as growth trends in a topic over time. Topic modeling can also be used to establish a "fingerprint" for a given document and then discover other documents with similar fingerprints.

As enterprises become increasingly interested in AI, they are turning to natural language processing to unlock the value of unstructured data contained in text documents. Research firm MarketsandMarkets predicts that the natural language processing market will grow from US$15.7 billion in 2022 to US$49.4 billion in 2027, with a compound annual growth rate (CAGR) of 25.7% during this period.

Let’s take a look at five examples of how organizations are using natural language processing to create business results.

Eli Lilly: Doing business globally through natural language processing

Multinational pharmaceutical company Eli Lilly is using natural language processing to help more than 30,000 employees around the world communicate within the company and Share accurate, timely information externally. Lilly has developed a homegrown IT solution called Lilly Translate that uses natural language processing and deep learning to generate content translations through a proven API layer.

For years, Eli Lilly relied on third-party human translation vendors to translate everything from internal training materials to formal technical exchanges with regulators. Now, Lilly Translate service provides users and systems with real-time translation of Word, Excel, PowerPoint and text, while keeping the document format unchanged. Eli Lilly uses deep learning language models trained on life sciences and Lilly content to help improve translation accuracy, creating refined language models that recognize Lilly-specific terminology and industry-specific technical language while maintaining the format of regulated documents.

Timothy F. Coleman, vice president, information and digital solutions information officer at Lilly, said: "Lilly Translate touches every area of ​​the company, from human resources to corporate audit services to the ethics and compliance hotline, Finance, Sales and Marketing, Regulatory Affairs, and many other areas. This saves a huge amount of time, as translation now takes seconds instead of weeks, freeing up key resources to focus on other important tasks. Business activities.”

Coleman’s advice: Support projects that are driven by passion. Lilly Translate began as a passion project by a curious software engineer whose idea was to solve a pain point in the Lilly Regulatory Affairs system portfolio: business partners were constantly experiencing delays and friction in their translation services. Coleman shared the idea and technical vision with other executives and managers and immediately gained project support from Eli Lilly's global regulatory affairs international leadership, who advocated for investment in the tool.

“[The idea] was a great combination of the opportunity to explore and learn about emerging technologies, and what started out as a great learning opportunity has now become one that Lilly software engineers grab and run.” Great project opportunity.”

Accenture: Using natural language processing to analyze contracts

Accenture is using natural language processing for legal analysis. Accenture's Legal Intelligent Contract Exploration (ALICE) project helps this global services company with 2,800 professionals conduct text searches in its millions of contracts, including searching for contract terms.

ALICE uses "word embedding", a natural language processing method, which can help compare words based on semantic similarity. The model examines contract documents paragraph by paragraph, looking for keywords to determine whether the paragraph is relevant to a specific contract clause type. For example, words like "flood," "earthquake," or "disaster" often appear with a "force majeure" clause.

Mike Maresca, global managing director for digital business transformation, operations and enterprise analytics at Accenture, said: “As we continue to leverage this capability and continue to enhance it, its use continues to expand and we see additional value. opportunities, and we're looking for new ways to get value from existing data."

Accenture said the project significantly reduces the time lawyers spend manually reading documents to obtain specific information.

Maresca’s advice: Don’t be afraid to delve deeper into natural language processing. “If innovation is part of the culture, you can’t be afraid of failure and let’s experiment and iterate.”

Verizon: Using natural language processing to respond to customer requests

Verizon’s business Service assurance departments are using natural language processing and deep learning to automatically process customer request reviews. The department receives more than 100,000 inbound requests each month, and previously they had to read and take action until Verizon's IT arm—Global Technology Solutions (GTS)—built the AI-Enabled Digital Worker for Service Assurance.

This Digital Worker combines web-based deep learning technology with natural language processing to read repair orders sent primarily via email and the Verizon portal. It automatically responds to the most common requests, such as reports. The current work order status or repair progress is updated, and more complex issues are submitted to human engineers.

"By automating responses to these requests, we can respond within minutes instead of hours after the email is sent," said Stefan Toth, executive director of systems engineering, Global Technology Solutions (GTS), Verizon Business Group explain.

In February 2020, Verizon stated that Digital Worker had saved nearly 10,000 man-hours per month since the second quarter of last year.

Toth’s advice: seek open source. "Look around, network with your business partners, and I'm sure you'll find opportunities. Think about open source and experiment before making a large financial commitment. We found that there is a lot of open source software available now."

Great Wolf Lodge: Using natural language processing-driven AI to track guest sentiment

Artificial Intelligence Lexicographer (GAIL) developed by hospital and entertainment chain Great Wolf Lodge sifts through comments in monthly surveys to determine Whether the author may be a troll, a critic, or a neutral party.

This AI tool utilizes natural language processing and was trained on more than 67,000 reviews specifically for the service industry. GAIL runs in the cloud and uses an in-house developed algorithm to discover the key factors that indicate how respondents feel about Great Wolf Lodge. Great Wolf Lodge stated that as of September 2019, GAIL's accuracy can reach 95%. For a small part of the information that GAIL cannot understand, Great Wolf Lodge will use traditional text analysis to process it.

Great Wolf Lodge Chief Information Officer Edward Malinowski said: "We want to be better able to interact with guests in every aspect."

Great Wolf Lodge's business operations team uses GAIL-generated Insights to adjust their service, the company is currently developing a chatbot to answer guests' frequently asked questions about Great Wolf Lodge service.

Malinowski’s advice: Avoid technology for technology’s sake. Choose tools that strike the right balance between technology and practicality and are aligned with business goals. "You have to be careful about what's a gimmick and what's a real solution to a problem." Provider Contracts app that automatically reads notes on each contract regarding payment, deductibles, and unrelated expense instructions, then calculates pricing and updates claims.

The application blends natural language processing and special database software to identify payment attributes and build additional data that can be automatically read by the system. As a result, many claims are settled overnight.

The app allows Aetna's more than 50 claims adjudicators to refocus on contracts and claims that require higher-level thinking, as well as coordination among different health insurance companies.

"It comes down to providing a better experience for the end user," Aetna Chief Technology Officer Claus Jensen said. The software will help Aetna become a better collaboration between providers and patients in the healthcare ecosystem. partner. "We do more than just pay bills and answer questions on the phone."

Aetna estimates that as of July 2019, the app has helped them save $6 million annually in processing and rework costs .

Jensen’s advice: Narrow your focus and take your time. In an ideal world, companies would implement AI that can solve very niche problems. Jensen said broad-based solutions are vague and ultimately fail, and if Aetna applies general-purpose AI to their business, it certainly won't work. In addition, Aetna spent several months instrumenting the process, writing rules, and testing the application. Jensen said many people don't have the patience to slow down and do things the right way.

The above is the detailed content of 5 successful cases exploring the business value of natural language processing. For more information, please follow other related articles on the PHP Chinese website!