Cognitive search, widely accepted as the next evolution of enterprise search, offers the potential for dramatic improvements in the accuracy, relevance, and efficiency of insight discovery. Although some see cognitive search as simply traditional search enhanced by machine learning and artificial intelligence, there is actually a sophisticated combination of capabilities that make it distinct from — and superior to — traditional enterprise search. Cognitive search goes well beyond search engines to bring together myriad data sources, along with sophisticated tagging automation and personalization, vastly improving how an organization’s employees find, discover and access the information they need to do their jobs.
This article provides an overview of the key elements of cognitive search, including data connectivity, meta tagging, user experience and contextual recommendations. It also explains how each element is augmented, rather than replaced by, machine learning technology.
Cognitive search provides a way for users to find integrated information from disparate sources both internal and external to the organization, whether on premise or in the cloud. Without that connectivity, users would have to determine where specific types of information reside and jump from one data platform to another to find it. Experience shows that under those circumstances they often just give up. Some of the on-premise and cloud-based enterprise systems that contain relevant information include Salesforce.com, Box, Microsoft Dynamics, SharePoint, Documentum, FileNet, Confluence, homegrown business applications and others.
The advantage of pulling together far-flung information sources is that it provides a foundational universe of digital assets that allows machine learning and AI to come into play. The value of separate data sources grows exponentially when they can be combined, standardized and auto-classified at the data object level in a way that facilitates complex collective analysis.
Given the rate at which unstructured content is growing, there is no way that manual tagging can keep up. Gartner estimates that unstructured data now represents as much as 80 percent of a typical organization’s knowledge repository. What makes this type of data truly useful is auto-tagging and-classification, which can increasingly be done on a contextual level at scale. Auto-classification enriches results using tags and metadata, and it has advanced capabilities like text analytics and named-entity recognition. Vast amounts of data can be classified and structured automatically, rendering it far more useful to enterprise users.
Microsoft’s AI platform and Google Cloud’s machine learning provide services for natural language processing, image and video recognition. The rapid speed at which new capabilities are being introduced in this area increasingly requires any cognitive search platform to ensure ease of integration, which in turn is stimulating wider adoption.
Using cognitive search, users can develop custom profiles and set criteria that shape the ongoing flow of information and recommended content. Cognitive search platforms also help identify and recommend experts based on users’ digital footprints — which may include such things as projects worked on, email contributions, and PowerPoint or Word documents authored by them, among other things — to foster collaboration and accelerate decision-making. This data can then be used automatically to provide more personalized and relevant results.
Recommendations can be provided directly to the user through an auto-suggest or type-ahead function, similar to what users see in Google search, or they can be offered transparently through personalized results. The more data fed back to the AI engine, the greater the relevance of future search results. Relevance can also be increased when system administrators use analytics about users and their search requirements to improve the system over time.
In addition to making contextual recommendations, cognitive search can proactively deliver “personalized information” that is particularly relevant based on factors like location, job title, subject, project, department or other attributes. Such a system becomes not only reactive, based on a user’s inquiries, but also proactive in that it attempts to anticipate a user’s needs based on a variety of characteristics related to past search efforts. This is where cognitive search begins to fulfill some of the promise of machine learning and artificial intelligence.
The Benefit of Cognitive Search For Your Most Important Customers: Employees
It’s important to remember that employees are valuable customers, too. All businesses try to engage customers and make it possible for them to find information about products and services. Many companies have specialized customer portals to make it easier to provide relevant information. But what about employee engagement? How much thought do we give to providing them with the information they need to do their jobs?
I recently spoke to an individual from a world-renowned company who told me, “I can find information needed to do my job faster on the Internet than I can on our own intranet.” Is that how we want our team members to see our organizations?
However those in the industry refer to cognitive search — Gartner calls it “insight engines,” and Forrester refers to the category as “cognitive search and knowledge discovery” — we need to heed the call to harness its power for employees. Unless we embrace and implement this new powerful form of enterprise search, its potential to turbo-charge our companies will come up short the same way that traditional search has — whether powered by AI or not.
If cognitive search incorporates the capabilities outlined in this article, it can provide a comprehensive solution that engages employees, makes them more productive, and allows them to focus their time on generating revenue rather than searching fruitlessly for information that should be readily available.