Time is running out! Submit now
by Mathews Thomas, Janki Vora, Christine Dee, Utpal Mangla, Neena Sathi, Swami Chandrasekaran, Arvind Sathi | Published September 28, 2016
Cognitive computing is forecast to be a $2 trillion market over the next decade, and companies are looking at how they can incorporate cognitive computing to improve their bottom line. Developing an enterprise cognitive solution often includes implementing emerging technology with multiple services that use various data, so implementation can be challenging.
In this series, we explain how to design and implement a complex cognitive solution. We discuss the need for a cognitive platform that can form the foundation for the solution, and we provide examples of applications that can run on the platform. Our examples focus on the Telecommunication and Media & Entertainment industries, but can be applied to other industries. We’ll delve into some use cases to illustrate key cognitive services and software components that are needed for implementation. A key to successful implementation of a cognitive solution is integration with existing systems, so we’ll discuss integration alternatives. The series will conclude with more important factors such as governance, methodology, and challenges encountered. This tutorial, the first in the series, provides an overview of cognitive computing and gives examples in the Telecommunication and Media & Entertainment industries.
Today, we have systems that can provide accurate, in-the-moment knowledge that lets you make strategic decisions across multiple fields such as science, legal, tax, and finance. These examples include:
These examples are just the tip of the iceberg in terms of what is possible with cognitive computing. Cognitive computing represents a new generation of computing systems that are built to augment, accelerate, and scale human expertise, enabling a new era of genuine human-machine collaboration.
We define cognitive computing systems as systems that are able to understand, reason, and learn. In other words, systems that can determine meaning from data inputs — structured and unstructured, text-based, or sensory — by interpreting context and classifying the data as information, or knowledge. To reason, these systems use data inputs and established understanding to form hypotheses, consider arguments, and prioritize suggestions. Lastly, these systems are capable of continuous learning, accumulating data and insight through human interactions. Cognitive systems are not programmed, but are instead trained, acquiring knowledge through experience and improving with time.
Human interaction and partnership are central to both the technology and philosophy of cognitive computing. Earlier expert systems required hardcoded rules and decision trees that could supply only the answers to previously identified questions. Cognitive systems build on the knowledge, and learn a domain’s terminology, processes, and patterns of interaction through communication with human experts. These systems are not programmed, but pose hypotheses based on data patterns and probability. In turn, cognitive systems can then make recommendations, provide insights, and suggest discoveries.
For example, the IBM Watson platform can process millions of documents, reading 800 million pages per second. Although we can learn, discover and make decisions, the digital explosion of content makes it impossible for us to make sense of it all.When massive computing power is applied to vast and diverse products of expertise, cognitive systems help us find unanticipated correlations, draw data-driven conclusions, and observe critical details that were once impossible to locate or interpret successfully. In this way, cognitive systems act as a decision and discovery support system.
To enable natural human interaction with these systems, the IBM Watson platform has numerous capabilities including:
These capabilities help with the understanding of actions between entities and the understanding of individuals and their specific communication needs. For instance, CogniToys Dino is a children’s educational toy that is powered by Watson that encourages learning and play through interactive dialog. By listening and observing the children that it interacts with, the toy can adapt to a child’s age and educational abilities, even adapting its personality to children with special needs.
Cognitive systems also use visual analytics and data visualization techniques. These techniques display data in visually compelling ways and do not rely on language only to be understood, but also graphic communication. Color, tone, language, and mathematical figures are all ways in which Watson can understand and make itself understood. These systems fundamentally understand, reason, learn, and interact with humans naturally.
In short, cognitive technologies amplify human cognition: assembling massive scales of disparate knowledge sources, reasoning through domain-specific inquiries, and acquiring new learning through experience. The system can then draw upon its knowledge stores, identify connections that inform discoveries, and interactively guide decision making — all in a continuous feedback loop of human interaction and partnership.
It’s obvious that the world will change with the adoption and advancement of cognitive technologies. All industries and organizations will be able to use this technology with regard to:
In the midst of these shifts, cognitive systems are crucial to a successful enterprise. These capabilities are the key to enabling new kinds of customer engagement, building better products and services, improving processes and operations, making data-driven decisions, and leveraging expertise.
To drive cognitive computing adoption, IBM Watson is available as an open software development platform. Its APIs are the building blocks for cognitive technology, and IBM Cloud and the Watson Developer Cloud support rapid development implementations for you to experiment, prototype, gain user feedback, publish, refine, and iterate. These sites provide SDKs, demos, code samples, an app gallery, a developer discussion forum, learning resources, and links to live events to help you learn and use this emerging technology. Today, there are more than 10,000 apps in test, experimentation, or production that support 3 billion API calls per month. This means that the transformative power of cognitive technologies is not only accessible, but already increasingly ubiquitous.
There are many examples that show how these capabilities can change our lives — some more whimsical than others. For example, Edge Up Sports helps players in a fantasy football league reduce their research time and take advantage of evidence-based recommendations when making decisions about their teams. Orrecco is using Watson to coach elite athletes, by combining physiological test data, biomarker data, and data on an athlete’s nutrition and sleep. The application produces an individualized training program that can predictinjuries, recommend the intensity level of a workout, recommend rest and recovery, or even optimize sleep schedules around travel.
In healthcare, Watson advises physicians and supports scientific discovery. In research, cognitive technologies have made it possible to use genomic mapping data to match individuals to specific clinical trials. Watson Discovery Advisor read hundreds of journals and found connections in the literature that was previously overlooked by researchers, and guided scientists to new discoveries. In the clinical setting, Watson for Oncology can analyze a patient’s medical information and identify evidence-based treatment recommendations suitable to the individual patient. For example, doctors recently turned to Watson to look into the case of a patient with leukemia.All treatment had proven ineffective, and teams of doctors had struggled for months to effectively treat this patient as her condition became critical. Within 10 minutes, Watson cross-referenced her condition against 20 million oncological records and identified the problem: the patient had a different form of leukemia than was diagnosed, and a different treatment was required.
IBM Watson has brought cognitive technologies together to form expert advisors, not just for physicians and researchers in healthcare, but in other industries as well. Softbank and IBM partnered to create Pepper, an emotionally astute robot that can serve as a retail sales assistant or a home personal assistant. Local Motors debuted Olli, a self-driving mini-bus that operates through a Watson powered interface: “Olli, can you take me downtown?” or “what nearby restaurant do you recommend?” In one development environment, Watson is built as an immersive experience — as an interactive wall, or cognitive space. During meetings, Watson can track who is speaking, what they’re saying, and even how they feel, so that it can intelligently contribute facts and suggestions that will assist the discussion and answer inquires. It can also display graphs and charts that can help participants understand the data that is being presented.
Cognitive solutions are built to augment, accelerate, and scale expertise within a specific domain or industry. These characteristics are what allows a cognitive system to learn through experience, and over time to develop knowledge specific to a particular field and role.
In the past few years, the lines dividing the once-separate domains of Telecommunication and Media & Entertainment industries have blurred.As with healthcare, much of the language and mission in the these industries is consistent in the broader industry perspective, while certain roles and areas of expertise remain very specific. Generally speaking, we can identify three imperatives cognitive solutions address in the overall domain: digital transformation, network infrastructure agility, and enterprise/operational excellence. To address these needs, work in various stages of development is being done to target specific roles and functions while working to attain industry experience and knowledge.
Consider the wealth of data available to the Telecommunication and Media & Entertainment industries: network transmissions and performance, devices, services, apps, retail, social media, published and broadcast content, product usage, audience and subscriber demographics, and much more. The possibilities forare boundless. When these measures are combined with, the solutions that are built can derive meaning, apply context, and coordinate expertly advised actions at a massive scale.
For digital transformation, these kinds of solutions can take many forms, but the primary concern is enriched, deeply personalized customer engagement — truly listening to and understanding the customer and effectively communicating with them in the ways they like best about topics important to them. To accomplish this, we focus on two key areas: systems of record and systems of engagement.
From systems of record, and with external data sources such as social media, weather, and local event data, organizations can gather metrics that enable well-defined audience or customer insight, also referred to as audience or customer segmentation. Cognitive systems elevate this understanding with IBM Personality Insights, Tone Analyzer, and other tools that guide the interactions in an organization’s systems of engagement. For example, by adding this capability, IBM is able to introduce Watson for Advertising Intelligence as an action recommendation platform, Watson Personality Insights and Recommendations for delivery of personalized content, and Watson TV, a television interface that interacts in natural language and offers highly optimized content recommendations.
Customer insight and expertise can also be combined to enable call center agents. In some cases, an agent in a contact center might be advised, in real time, as Watson listens in and makes recommendations about how to answer a customer’s concerns or their intents. The knowledge base in this case can include millions of data points related to devices and connectivity, known outages and their expected resolution times, rate plans, payment options, auxiliary services, or accessories that might be appropriate for upsell, contract status and propensity to churn, and even the customer’s history of calls. In other cases, a Watson powered virtual agent might act as a phone or chat representative and perform the role of the customer care agent, advising and helping the customer directly, and “soft” transferring to a live agent if needed. Perhaps most impressively, customer conditions can be understood through predictive means in such a way that needs are anticipated ahead of the customer contact and a resolution is offered to the customer through the organization’s app, a message, a video, a digital agent chat, or other assistance that gives the customer exceptional service and prevents costly direct contact with a live agent.
Network infrastructure is also an essential part of the industry, as it is the delivery mechanism for all services: calls, data, streaming media, games, and so on. The demand for performance and faster speeds is extremely high, even as the competitive pressure to lower costs intensifies. All the while, the systems have become far more complex, requiring advanced and diversified expertise to successfully operate, plan, and maintain. Watson for Network Operations and Watson Field Tech/Equipment Repair are designed to work with technicians who are on the front line of resolving and preventing network issues by providing essential guidance to newer professionals who are on a learning track, helping to coordinate work and resources across teams, and providing up-to-the-minute information about conditions that might affect their services.
In the future, cognitive solutions will assist network planning and optimization, service assurance, and capital acquisition initiatives by pulling together network analytics, topology and hierarchy schematics, entity relationships, and regulatory compliance standards alongside financial metrics to determine opportunity cost and prioritization of network investment. Today, these functions are fulfilled by large teams of experts who assemble these types of planning and prediction recommendations. This is an ideal fit for human-machine collaboration, as advanced computing power could vastly improve the process and results, but human expertise is essential to build the knowledge foundation and ultimately to make the decisions for the business.
Other internal operations can benefit from cognitive solutions as well. Watson Company Analyzer performs company research by collecting data from news articles, financial performance sources, and other publicly available data. It enables peer benchmarking and real-time ability to predict risk of supplier failures or locate opportunities for potential suppliers, and it presents this data in an easy-to-use graphical representation. In another solution that is built for internal operations, Watson Procurement Intelligence supports data-based decision making, negotiating, and planning related to suppliers, products, and pricing.
So, how can you incorporate cognitive capabilities into your respective applications and products? Watson Developer Cloud can help by providing a secure, open cognitive platform that gives you easy access to cognitive “building blocks” made available as REST APIs. Using these building blocks, you can build cognition into your digital applications, products, and operations. These cognitive services can be fueled by publically available web and social data, your own private information, or data you acquire from data partners or others. The APIs on this platform can be grouped into four categories:
Each of these APIs can perform a different task, and in combination they can be adapted to solve numerous business problems or create deeply engaging experiences. When you combine these cognitive services and overlay with (traditional) data analytics capabilities, it facilitates for complex discoveries, predictive insights, and engines to carry the decisions that are driven by the insights.
The Watson services form the core of the platform. These services need to be integrated with existing systems so that cognitive processes can be embedded into existing solutions. The intent is not to rip and replace existing systems as this will be expensive and cumbersome, but to embed cognitive computing into existing processes.
The key components of the platform include:
The following figure illustrates the use of the cognitive platform in the context of a telecommunications network operations environment.
Having such a platform allows the implementation of multiple cognitive services using existing systems and data. Future tutorials in this series will build on this basic platform so that you will understand how to incorporate cognitive computing into existing and emerging enterprise solutions.
We have provided a brief overview of cognitive computing to set the overall context of why it is an important emerging field that will soon impact all aspects of solution development. The key characteristic of a cognitive system is the ability to understand, reason, and learn. When you are implementing such a system, it is important to consider the cognitive platform that will be used to ensure that the systems are built effectively and can be used to run multiple use cases. The next tutorial in the series will provide sample cognitive use cases that can be built on such a platform. Future tutorials will take a deeper dive into this platform using some of the defined examples so that you will be able to implement your own cognitive solution.
April 8, 2019
May 8, 2019
Get an overview of computer vision with deep learning and learn how it can help your applications recognize what an…
Artificial intelligenceDeep learning+
Back to top