Cognitive Computing: Film at Eleven

On a recent, very long plane trip, I watched a fellow passenger scroll through a number of movies on her mobile device. She would watch a few minutes of a film, then switch to another one. After she repeated this pattern for several films, she sighed and shook her head. It turns out her brother had recommended movies for her “watching pleasure” while flying over the Atlantic Ocean. Unfortunately, his take on good films to accompany her trip didn’t coincide with hers. “They’re all silly,” was her judgment.

Why did it immediately occur to me that what was needed in this situation was cognitive computing? I doubt my fellow passenger ever gave a thought to cognitive computing, and probably would not have known what I was talking about had I mentioned it to her, but I had just read Attivio CMO Lou Jordano’s white paper about Netflix as the future of cognitive search. His take is that movie recommendations from Netflix, which rely on machine learning, reflect user behavior as being predictive of preferences. Relying on people’s ratings of films is not nearly as accurate. This bears out what social scientists have known for years: What people say and what they do don’t always coincide.

Jordano also points out the importance of context. The time of day and day of the week, taken in the context of your work experience, can help a search engine predict the knowledge you need at that particular moment. He admits that, in the work world, it’s tricky to predict what people want to know when they search. Providing relevant results also has more impact, particularly on productivity, in a work environment rather with entertainment. He suggests that the recommendation model of Netflix could be applied to knowledge management. The technologies of machine learning and natural language processing (NLP), components of cognitive computing, can make major contributions to corporate strategy. Perhaps someone will make a movie about it.

Learning Aspects of Cognitive Computing

Learning permeates discussions of cognitive computing and the re-emergence of artificial intelligence (AI) in discussions about knowledge management. AI derives much of its power from deep learning, machine learning, and adaptive learning, with contributions from NLP. Simple queries don’t need much AI applied. “When are my taxes due?” “What hours are you open?” “Who’s my representative?” These questions are straightforward. It’s more complex queries that require some cognitive technology.

With eGain’s virtual assistants (VAs), says Anand Subramaniam, Senior Vice President, World Wide Marketing, eGain, more complicated questions can be handled. Not everything, of course, but enough to make the process more efficient—and customer pleasing—before escalating to a human. This is not your menu item hell where you “press 1 for this,” “press 2 for that,” and none of the “thises” or “thats” actually match up with what you want to know. Because the system learns as it goes along, understanding is enhanced and a more satisfactory answer is delivered.

Reasoning is the AI piece that guides answers to more complex questions. The capacity to learn from past experiences leads to better business decisions. Subramaniam warns against the use of rigid scripting and rule-based systems, believing they are too inflexible, particularly in today’s business world, which is full of ambiguous and incomplete information. Relying on last year’s solutions without any understanding of how things have changed is not at all intelligent.

Search Technologies

If the future of cognitive search is Netflix, Nika Mizerski, product manager or PoolParty Semantic Suite, thinks that the future of cognitive computing lies in a combined approach, since an all-encompassing AI technology doesn’t exist. The combination entails machine learning algorithms, data integration, NLP, and advanced data analytics.

We see all kinds of smart applications—and not just for our smart phones. Mizerski mentions chatbots, recommendation services, knowledge discovery tools, and analytics solutions. But smart isn’t synonymous with intelligent. Semantic technologies can use conceptual networks and knowledge graphs that mirror the human brain’s architecture. Subject matter experts contribute to the cognitive applications to create business value.

All three white papers on cognitive computing stress that this is not a static process. Companies do not simply install a system and leave it alone. Just as our brains continue to evolve and learn, so do business processes.

Film at Eleven

What about the woman on the plane with her choice of movies? My guess is that she doesn’t need an in-depth understanding of cognitive computing to know not to ask her brother for movie recommendations for her next flight. Would she prefer a comedy, a drama, or a film set in the country she intended to visit? Her brother didn’t know, but it’s possible that a cognitive computing application would.

Cognitive computing can be a powerful aid in delivering excellent customer service and solving business problems. Machine learning, AI, cognitive search, and semantic technologies can make the difference between happy, contented customers and those who walk away dissatisfied. Or, in the case of my seatmate, simply fall asleep.

Losing relevancy from search systems drives users away, since the knowledge provided doesn’t meet requirements. Thus, it’s imperative to do cognitive computing correctly so it will add value to organizations. The promise of cognitive computing is immense. Business cases spotlighting AI, NLP, machine learning, text analytics, and the other components of cognitive computing will continue to add to our knowledge about knowledge management and our ability to capitalize on these tantalizing technologies.