Transitioning into 'Machine Learnable' Data Management

"I'll be back" - says Schwarzenegger in one of my all-time favorite series, the Terminator. The story and the fiction, at that point in time of release, had an 'awe' element to everything that we saw on the screen. Cut to the recent past of connected devices, being able to see robots perform complex tasks, learn cooking, surgeries and what not, reality is closer to fiction. So, if a bot can author a book, can it help manage my enterprise data better? If it does can we see some real tangible benefits? In all of this, data is the underlying fabric which helps the bots. Be it a cooking recipe, the ingredients, the precision in length, depth and breadth of a cut to perform a surgery and so on. There is still a long way to go.

For enterprise data management and our kind, the 'cogs' behind that fiction will help make our customer businesses better, via efficient procurement, better financial insights and more so free up business time to manage and maintain accurate asset, customer, product, chart of accounts and facilities data. While most 'learning' algorithms use mathematical models and the ability to improve efficiency based on looking at historical data, these are going to help ease the master data management (MDM) related aspects. When we apply machine learning (ML) to MDM and look at a MLDM, the possibilities are quite interesting. If a business steward and a stewardship service can spend more time improving process efficiency with minimal or no intervention to mundane workflow activities in MDM, maybe that Christmas sale or thanksgiving sale turnover might have been better?

The MDM to MLDM transition can have various business benefits. Some of them are: quicker procurement times due to closer & quicker material matches; increased market (regional/zonal/geo) penetration due to the much closer & faster accuracy of customer profiles, demographics and segmentation; automation of mundane stewardship activities which will lead to better business resource utilizations. The faster shelf replenishment of the higher revenue products for retailers, improving the e-commerce efficiency for e-tailers and maybe handling & procuring the right parts of a prototype from a vast set of suppliers to beat the competition on a launch. The benefits can be endless.

All of this is made possible, as the system can now look at reducing the number of iterations that it takes to arrive at the right threshold based on business priorities and inputs. Over a period of time, the data stewardship activity can be automated to enable workflows to 'learn to adjust' and route the information for approval OR take a business call based on historical preferences. Of course, there has to and will be and has to be a 'place of intervention' for business to rectify if needed. But the idea is to make time for process and not workflow tasks going forward. The match rules and weights assignment, should be determined by the systems from where the data comes from and the definition of the entity itself. This should drive the process and then the algorithm would take a mathematical view of the next best choice in some time. This would be evident in match accuracy and data elements selection to define an entity becomes more robust and makes a difference to the business. The difference is in the time to serve customers and time to procure material from suppliers, for example.

MLDM exists today in some MDM players with deeper and richer relationship views with a more analytical driven MDM apps available. While this provides a foundation, it has become essential for MLDM to be the driver for better business outcomes where MDM is involved. In the era of open source, IoT, AI and big data MDM becoming MLDM is the way forward and will still act as the bearer of quality, governed and more business relevant application in the enterprise. This will still be the right partner to the other players of the ecosystem to get to that elusive 'Golden' record.

What has been your experiences with your customers? Do you see various industries at various stages of adoption? Your valuable comments are welcome to further this discussion.

About the Author

Venkataraman Ramanathan has more than 15 years of experience in the areas of MDM, DQ, Data Governance, CRM, BI and Java. He has been involved in various consulting and engagements across geographies for a variety of Fortune 500 companies. He is involved in the areas of customer re-engineering, machine learning, data quality, data governance, sales and service efficiency, business strategy and business analytics/intelligence.

Service Offerings

Client Themes

Service Offerings

Client Themes

Our Global Presence Global Site

Our Global Presence Global Site

Our Showcase Pages Locations

Transitioning into 'Machine

Learnable'Data Management

About the Author

Service Offerings

Industries

Digital

HOLMES

Innovation

Contact Us

About Wipro

Location

Leadership

Investors

News and Events

Blogs

Analyst Speak

Partner Ecosystem

Careers

Sustainability

Contact Wipro