With the data deluge data managers face today, smarter options for managing, enhancing, and connecting customer data points need to find their way into the data toolkit. AI capabilities are playing a new role in this effort. Initially developed to support advanced scientific research, tools such as machine reasoning are powering up key operations that include identity verification, customer onboarding, and product-focused data quality. For the range of stakeholders in the enterprise data chain—compliance officers, database managers, marketers, and more—this supports a complex set of operations aimed at stemming fraud and meeting compliance regulations while also identifying customer opportunities and improving the customer experience. Data quality is at the core of these capabilities; new value comes from a greater focus on uncovering data relationships and their impact on the business.
Smart data discovery
Customer data illustrates the challenge. Ideally, big picture data is aggregated into a single customer view (SCV), or a single source of truth regarding its customer based on accurate internal and external data, unified across channels, locations, and business silos. Data channels alone represent a challenge, considering the twelve distinct channels identified by HubSpot: website, videos, blogs, gated content, Messenger, Twitter, live chat, phone, a ‘contact us’ form, self-service, email and Slack. When all this data is integrated, businesses can analyze past behavior to better understand customer needs at each touch point and personalize future interactions.
Going beyond simple verification to provide a 360-degree SCV is useful not only for operations such as identity verification and fraud detection, but also for sales/marketing or customization that meets specific needs. These are business risks as well as opportunities, calling for higher quality data and smarter software. AI-enabled platforms and applications can automate data quality and improve interoperability, providing a shorter path to a 360-degree SCV.
Understanding semtech and where AI adds value
Rather than risk an overly ambitious AI effort based on deep learning and massive datasets, enterprise applications must be more goal oriented and well defined. Best practices in enterprise data management include more cost-effective, practical solutions. It is unnecessary to apply supercomputers and deep learning tools across huge amounts of data in order to realize value from AI—narrow initiatives are more concrete and therefore achievable.
Semantic technology, or semtech, is an extension of the current web that has been defined over the past decade by the World Wide Web Consortium (W3C) in collaboration with Stanford University, MIT and others. Semantic technology stands out in its design which facilitates universal data interoperability. In comparison, HTML (whose specifications are also managed by W3C) is purely designed to allow universal access to documents on the web.
Along with global data interoperability, semantic technology delivers a form of AI that associates words with meanings and recognizes relationships between them. This allows the business to tap into greater understanding of its customers by making powerful, real-time connections among the data in their records. Essentially, as business professionals digitize, they should also transform customer data—making it findable, accessible, interoperable and reusable (FAIR). This can be achieved by using AI-enabled semantics and machine reasoning to normalize, harmonize, and richly connect information.
The role of machine reasoning
Machine reasoning applies knowledge about a subject of interest captured in the form of ontologies—these are formal descriptions of classes, entities, and relationships relevant to computing in a specific area of interest. This promotes context and reasoning, or the ability to make inferences, assumptions based on logic applied to data. For example, this ensures properly validated identities as well as broader data quality and integrity. AI-enabled machine reasoning can also quickly open up opportunities for rich pattern recognition and responses. By taking all available data into account, the process reduces false positives—problematic situations that slow down business and have the potential to annoy customers.
Semantically-enabled software platforms also complement machine learning, a more commonly known form of AI. Machine learning applies supervised and unsupervised algorithms to analyze training datasets; its goal is to identify features that are, or seem to be, related to outcomes. This powers the generation of inductive hypotheses and facilitates pattern identification used for decision support.