Watson Puts On A Show At COMMON
May 8, 2017 Dan Burger
Watson, the anthropomorphic face of IBM cognitive computing, is used to being on the big stage. Fame came quickly and easily with the TV quiz show Jeopardy! Then there was recognition for work in cancer research and fighting cybercrime. IBM has invited Watson to all its major conferences and even changed the name, and some of the focus, of its Power Systems group into the Cognitive Systems business unit.
Yesterday, Watson showed up at the COMMON Annual Meeting and Exposition, the world’s largest gathering of IBM i professionals. The first impression was a good one. It emphasized easy integration and straight-forward objectives, and that Watson analytics and IBM i transactional capabilities are just getting acquainted.
At the forefront of the Watson and IBM i comingling is an increased focus on in-depth analytics and providing existing customers with strategies to extend the business value of their applications using new and important technologies.
IBM executives Stephanie Chiras and Steve Sibley handled the Watson introduction to the audience of IBM i advocates at the opening session. Chiras noted the strategic changes at IBM relative to the evolution to the cognitive era–moving from a mastery of data, reporting and context to include deeper insights. Sibley went in the direction of system design and high performance interconnects, while mentioning the introduction of Power9 servers would come late in 2017, with IBM i versions unavailable until early 2018.
Sibley also mentioned working with several large IBM i shops that are leveraging Web services and running large-scale deep learning projects involving ETL, data lakes, and running with the data closer to the transactions. Client skills are evolving, the whole industry is evolving, he says.
Chiras described Watson’s role like a page out of a Harry Potter book. Readers familiar with the Potter fantasy novels know about the sorting hat that magically assigns new students to one of four school houses through a process of mind reading, which groups them with students of similar likes and dislikes.
Watson demonstrated similar sorting hat capabilities using a combination of speech to text and social media data to sort COMMON attendees who stopped by the IBM booth in the expo area at the conference. (There’s a rumor that Watson sorted out COMMON attendees who still refer to the IBM i as the AS/400 and those individuals were marched into a room and sternly lectured by Trevor Perry.)
Demonstrations and educational sessions include interactions with RPG applications, using DB2 for i and Bluemix connectors, Watson’s capabilities for building probability tables based on searches of non-traditional data such as blogs, Facebook and other places where people make comments revealing what they like and dislike.
One of the Watson/IBM i business cases that was noted involved a connection with DB2 Web Query using DataMigrator to pull data from multiple sources, pass the data to Watson Analytics and then into Web Query.
The additional of Watson analytics, IBM says, brought predictive analysis capabilities that Web Query alone could not provide. Web Query was built to analyze current transactional data and compare it with historical transactional data and identify trends, but it lacks a predictive component. Watson delivers data from sources that go beyond the transactions. That makes Watson, according to IBM, more than an analytics engine and capable of learning from the world around it.
“The message behind what we are doing with Watson at COMMON is that things can be easily and quickly built that connect with IBM i. The Harry Potter sorting hat idea was a fun way to show this,” says Alison Butterill, IBM i offering manager. Butterill led a Sunday morning session on Watson and IBM i collaborative projects with IBM i technical wizards Scott Forstie and Jesse Gorzinski providing insights into the integration efforts already in operation.
As much as possible, the demos and the sessions are designed to show what can be accomplished without a lot of effort. The emphasis is on gradual progress and how it is easier and more fun to learn something new when you keep it simple at the beginning rather than trying to accomplish complex tasks too quickly.
Knowledge of the latest technologies used to interact with programs or descriptions of data will be a factor in how easy it will be for IBM i developers to connect their systems to Watson.
“Many of the technologies used to connect to Watson are outgrowths of things we have been doing in non-cognitive programming,” IBM i chief architect Steve Will noted when he spoke with IT Jungle last week. “You need to know how to formulate a request to a remote server–things like XML and JSON services and knowledge of SQL – will be used to tie Watson and IBM i. People have been learning these technologies for their normal job anyway.
There is a segment of IBM i customers who don’t know SQL and Web services. Those people are behind. Those who know SQL won’t find this to be all that difficult. If you’ve figured out how to use XML or JSON Web services to make data requests, that’s what it takes to talk with Watson.
There’s a little bit more to learn about what you’ll get back from Watson and how you might process it. It’s not going to be a normal transaction processing of a ledger entry. It’s going to be an analysis of information and how much you want to expose to the user or use in your query.”
APIs that allow retrieval and converging of data are available. Using facial recognition technology as an example, the camera knows how to take the picture. It creates a jpg digital file, for instance. That’s a format the Watson API is expecting so it can do its facial recognition. Making an API call to a camera is no different than making an API call to a program. The output from the camera is then input into another program, which is a service running in Watson. That service retrieves answers. Programming determines what answers you get from Watson. It could be a definitive answer or multiple answers that are weighted.
One thing to keep in mind is that the integration of Watson and business computing is in the early stages. The Watson team is building requirements for specific industries based on requests for customers in those industries. As more APIs become available, the priority is not driven by platform or language, the priorities are based on industry requirements.
The requirements for the IBM i development team, according to the chief architect is to integrate new languages and libraries into the programming environment to make it easier for traditional applications to make use of the cognitive capabilities.
“Today we are at the infancy. We know we have to be able to do Web services calls and SQL calls. There are languages developing around talking to Watson and we might need to do some of those languages. There are packages being developed that make it easier to do cognitive thinking. We may have to port some of them. These are enablers that allow us to connect our traditional business applications to what they can do in the future,” Will says.
The intimate details of building cognitive applications with visual recognition are covered in volume three of a seven-piece collection of IBM Redbooks titled Building Cognitive Applications with IBM Watson Services. The series includes an overview of specific Watson services with their associated architectures and simple code examples. Each volume describes how to use and implement these services. The series includes Redbooks that specialize in getting started, conversation services, visual recognition, natural languages classifier; language translators; speech to text and text to speech; and natural language understanding.
The Focused Education Roadmap for Watson includes the following sessions: Watson and IBM i, Overview of Open Source on IBM i, What’s New in DB2 for i, Snake Handling – Building Python Web Apps, REST Web Services for the RPG World, Constellations on i – Orion, Git and RPG, Untangling Your Data for Analytics and Reporting Systems, Watson and PHP Inventory Integration, and What’s New With DB2 Web Query.