Cognitive Computing: Real-world applications
for an emerging technology
Emulating human reasoning
Part of the promise of cognitive computing is that it can address issues that span the globe or focus on a single individual. The Enterprise Cognitive System (ECS) from Enterra Solutions can manage risk in global supply chains, determine how to optimize trade promotions or identify a flavor that matches a consumer’s personal tastes, among other applications. “A tremendous amount of data is being collected today by organizations, but the challenge is that there are not enough data scientists to analyze it quickly,” says Stephen DeAngelis, CEO of Enterra. “We saw a requirement for systems that could not only ingest a lot of data, but also reason, calculate and generate actionable insights at machine speed in order to take actions within decision cycles.”
“Cognitive computing is the solution to big data,” says Venkat Rajan, principal analyst at Frost & Sullivan. “Companies are collecting a lot of information and are having trouble processing it, but if there is a cognitive layer you can more accurately detect an issue.” For example, a medical problem might need to be brought to a doctor’s attention, or data from a wearable device could be compared to data obtained in a hospital to detect trends in a medical condition.
Enterra ECS is the cognitive computing engine in an application called FlavorPrint developed for McCormick & Company. FlavorPrint gathers information about a consumer’s preferences through an initial survey and subsequent user interactions with the site, and then compares that data to its body of information about food—the taste and feel that is determined from the ingredients and the preparation. Now spun off as Vivanda, the application is part of Vivanda’s initiative to provide positive food experiences for both consumers and the food industry.
Other applications of Enterra’s ECS include Conair’s Retailer Compliance and Pre-Inventory Prediction systems, which gather information from internal and external sources on inbound shipments, distribution center operations and customer orders to predict and prioritize potential problem areas and assist operators in determining resolutions early in the process.
“We developed a system to solve real-world problems much like a human would,” continues DeAngelis, “by generating hypotheses and reasoning plans, executing calculations and then interpreting the results in context.” The components of Enterra ECS include a commonsense ontology that represents general knowledge and relationships, narrower and deeper specialized ontologies that relate to a particular domain, a rules-based inference system, a hypothesis engine, natural language processing and several advanced calculation engines.
Flexible and resilient
The big advance that cognitive computing has over early artificial intelligence (AI) systems, aside from vastly greater computing power, is resiliency. “Expert systems of the past lacked flexibility and could not take into account the exceptions to the rule—they were brittle,” says DeAngelis. “However, there will always be exceptions, and since cognitive computing functions on a probabilistic basis rather than an absolute basis, it can reason without having to break the rule.”
DeAngelis cites drug discovery as one of the most promising areas for cognitive computing. “Drug discovery time could be greatly reduced because the design of the experiment process is now able to be disrupted by the cognitive computing and advanced mathematical techniques. Today we can mathematically model the high dimensional interactions of molecules within a compound and find 4th or 5th order combinations of molecules that explain the behaviors of the data. We then can dramatically speed up the development cycle,” DeAngelis explains. “Thus the trial-and-error part of the drug discovery process could be eliminated in favor of a more data-driven process.”
DeAngelis also points to applications in consumer products and retail, value chain optimization and trade promotion management as strong areas where cognitive computing can have a game-changing impact. “We believe this industry will create billions of dollars of value in the next five years,”
Cognitive Computing Consortium
As is the case for many emerging technologies, a consortium has been founded to help develop the field of cognitive computing. The Cognitive Computing Consortium ?(cognitivecomputingconsortium.com) was formed to establish a community in which to share ideas and information and to ?support research through partnerships. It will also sponsor research and events such as workshops and conferences.
Members have expertise in a wide variety of related disciplines, including machine learning, big data and search. Among the group’s first achievements was the publication of a definition of cognitive computing. Current research initiatives include an assessment of the impact of cognitive computing on organizations and individuals.