IBM Watson: 10 New Jobs For Cognitive Computing
IBM Watson is adding language-processing, image-recognition, and reasoning services to power these 10 breakthrough applications that scale up human-like analysis.
![](https://eu-images.contentstack.com/v3/assets/blt69509c9116440be8/blt1de1de846bb828c7/64cb5715b3e24505d53f0502/IBM-Watson-Intro.jpg?width=700&auto=webp&quality=80&disable=upscale)
Computing power meets the human-like capacity to speak, seeing, reason, and learn. That's what cognitive computing is all about, as exemplified by the category leader, IBM Watson. But is this practical technology that can affordably handle important tasks?
Even before it acquired the AlchemyAPI platform last week, IBM's Watson business unit was busy adding language, speech, machine vision, and decision services aimed at powering breakthrough applications. AlchemyAPI expands and accelerates those efforts, bringing Watson a portfolio of language- and image-processing services, machine understanding of eight human languages, and, most particularly, a following of more than 40,000 developers who call on its application programming interfaces.
But a bundle of services won't necessarily add up to a useful cognitive computing app. The big idea with cognitive computing -- computing that can learn and improve, not just follow instructions -- is scaling up and accelerating human expertise. For example, our times have brought a deluge of information, so one big play for cognitive computing is quickly combing through troves of timely and potentially relevant information that even armies of humans couldn't possibly sort through in a matter of seconds.
[Read about the hottest skills IT pros need.]
For example, Watson powers medical diagnostic apps that "read" through the millions of research papers and clinical trials published each year to surface relevant insights on patient-specific treatments. Watson-based financial services apps introduced at ANZ Bank in Australia and CaixaBank of Spain offer investment advice, quickly combing through tens of thousands of potential investments and suggesting best-fit options based on customer-specific profiles detailing their life stage, financial position, and risk tolerance. Insurer USAA has adapted the IBM Watson Engagement Advisor, a learning app for complex service-and-support roles, to help veterans answer complex questions and find appropriate resources when they're considering leaving the military.
These are just a few examples of cognitive computing apps that are emerging, but read on for a peek at new types of applications and evolved applications that IBM expects Watson to power in 2015 and beyond.
Today it's easy for natural language processing technologies to classify the topics and assign sentiment to the individual posts within a stream of social content. It's much more difficult to get beyond binary, positive-or-negative understanding of human intent -- in social comments or in interactive commerce or support interactions. These are among the more subtle capabilities that AlchemyAPI and IBM have both been working on. IBM says AlchemyAPI's deep-learning technology and experience will broaden and advance Watson's capabilities.
"The performance of classification and ranking processes will get better, but they are just the first components of future solutions that seek to predict and individual's purchasing behavior," said Elliot Turner, founder and CEO of AlchemyAPI, in this 2014 interview by natural language processing expert Seth Grimes. "Success in this task will combine other elements such as a person's interests, relationships, geography -- and ultimately their identity, purchase history, and privacy preferences -- so that applications can plot where a person is in their 'buyer's journey' and provide the best offers at the best times."
Cognitive apps won't get there overnight, but AlchemyAPI's vast training corpora and language-processing capacity should improve Watson's learning capacity and, thus, accuracy of understanding over time.
Great retailers have smart salespeople who know products, ask questions, and quickly help customers find the products that are right for them. Digital stores, in contrast, tend to offer great prices, convenience, and selection, but the customer has to do a lot of research, particularly when the products are complex or technical, like high-end ski gear.
Ecommerce specialist Fluid developed Fluid XPS, powered by IBM Watson, to offer an interactive Expert Personal Shopper that interacts with would-be customers, posing simple questions to understand their interests and needs. Even before that interaction begins, Watson's learning capabilities comb through structured and unstructured data such as product catalogs, social media posts, product reviews, and call-center logs to develop a deep understanding of the products being sold.
Fluid started work on combining e-commerce interfaces with IBM Watson Developer Cloud services in 2013. Fluid XPS pilot projects started with retailers including The North Face in 2014. IBM says AlchemyAPI natural language processing and behavioral understanding capabilities will broaden and accelerate Watson Developer Cloud services supporting e-commerce scenarios.
Retailers need help to become great retailers, particularly when they sell lots of sophisticated products. Red Ant developed the Watson-powered Sell Smart application to help salespeople better understand both products and customers so they can sell, cross-sell, and up-sell more effectively.
The tablet-based retail-sales training app helps store employees identify customer buying preferences by analyzing demographics, purchase history, and wish lists, as well as product information, local pricing, customer reviews, and tech specs. Voice or text input and question-and-answer interaction helps guide salespeople to provide better, more personalized customer service.
It's not just retailers who sell and support complex products. Insurance companies, financial services, colleges and universities, and technology companies make digital interactions easier. What's more, customers often prefer online self-service, but not if sites are hard to navigate and make it tough to find what they're after. IBM Watson Engagement Advisor is aimed at challenging support scenarios where simplistic phone trees or mobile experiences might be maddening.
Engagement Advisor was the starting point for insurer USAA, which serves more than 10 million US military personnel and their families and advises them on a range of financial and life decisions when they decide to transition to civilian life. In 2014 USAA chose the Military Separation sections of its Web and mobile sites for a Watson pilot project because it involves an array of complex topics including education, health insurance, housing, and retirement decisions.
Watson interacted with registered users so it could combine its cognitive capabilities with some scripting based on what's known about each customer. Watson was initially trained on 3,000 documents, including USAA documents and content from external sources including the Veterans Administration and the Department of Defense.
Online booking engines and travel sites have undermined travel agents, but they really can't match the knowledge and guidance provided by an experienced professional. Travelocity founder and former chairman of Kayak.com Terry Jones developed the WayBlazer online service to provide concise, personalized travel advice based on traveler preferences and behaviors expressed in natural-language interactions and social profiles.
Currently in beta on a Austin Convention and Visitor's Bureau travel-advisory site, WayBlazer uses a Watson-powered customer modeling application to profile the style and interests of the would-be traveler. It then delivers targeted recommendations by combing through structured and unstructured sources about destinations, restaurants, hotels, events, and local attractions.
Doctors have trouble keeping up on the latest viruses, ailments, and diseases, not to mention the latest drugs and medical research uncovering changing thinking on treatments and best practices. Now imagine the complexity faced by veterinarians, who don't know whether their next patient is the kind that walks, crawls, swims, or flies.
LifeLearn developed the Watson-powered Sofie app to advise veterinarians, who often face time-critical diagnostic situations treating animals ranging from dogs and cats to lizards, snakes, birds, horses, and livestock. Watson has been trained on vast troves of diagnostic information, best-practice treatment options, and specialist insights on a range of animals.
The massive volume of medical research and clinical trials published each year is not only a challenge for doctors, it's hard for medical schools and full-time medical students to take in. IBM Watson is interacting with medical students and clinicians at the Cleveland Clinic Lerner College of Medicine at Case Western Reserve University to gain an understanding of medical terms, concepts, and reasoning so it can be used as a tutoring tool.
The idea is for Watson to be able to teach medical students by taking them through diagnostic scenarios in which anonymous patient profiles and symptoms are matched with a prioritized list of appropriate treatments. This ability to wade through troves of information is built into Watson Discovery Advisor, a starting point for cognitive computing applications tuned specifically for life sciences and medicine. Natural language processing and deep learning finds connections among words and related concepts to generate hypotheses about treatment options or new uses for drugs.
(Graphic: IBM)
IBM acquired AlchemyAPI in large part because it stands apart in machine vision (image processing) capabilities. This will accelerate Watson Watson's ability to handle images, according to IBM, which has mobile, social, and e-commerce applications in mind.
IBM and AlchemyAPI were partners even before the acquisition, and we suspect the AlchemyVision API is behind the recently released IBM Watson Cloud Visual Recognition service. The service analyzes images or video frames and interprets what's happening in the scene. Prebuilt classifiers offer more than 2,000 trained image tags and taxonomies for different domains. A sports taxonomy, for example, recognizes more than 150 sports and can tag images or footage with a confidence level as to whether it's an example of soccer or baseball. Use-cases include organizing large collections of imagery and understanding consumers' shopping preferences based on the images they're viewing.
Leveraging its machine vision capabilities, AlchemyAPI last May rolled out a Face Detection/Recognition API that can identify the gender and approximate age of people in pictures. Age ranges supported are <18, 18-24, 25-34, 35-44, 45-54, 55-64,>64. The service can also identify well-known people, pulling from a corpus of more than 60,000 celebrities and business and political leaders. The API returns both the person's name and a confidence score.
Custom training would enable Watson to recognize any collection of faces you would like, so border security, police, and intelligence work is as possible as celebrity sighting. A more commercial possibility is bringing demographic insight, sentiment analysis, and understanding of intent into e-commerce or customer-support scenarios based on social profile images or mobile app interactions.
Imagery is all-important in medical diagnostics, whether involving X-rays, MRIs, ultrasound, or high-res photographs. IBM Research is collaborating with Memorial Sloan Kettering Cancer Center on a way to use cognitive computing and visual analytics to diagnose skin cancer more quickly and accurately. IBM says that in a scan of 3,000 images, IBM technology was able to spot melanoma with 95% accuracy, much higher than largely manual methods that average 75%-84% accuracy.
This application is in its infancy, but it's easy to imagine Watson combing through thousands of images as an extra layer of scrutiny, backing up human experts by flagging images that might have been missed when looking for lung cancer, breast cancer, or other life-threatening ailments. Skin cancer alone afflicts nearly 5 million people per year.
-
About the Author(s)
You May Also Like