As artificial intelligence (AI) and machine learning (ML) begin to move out of academia into the business world, there’s been a lot of focus on how they can help business intelligence (BI). There are a lot of potential in systems that use natural language search to help management more quickly investigate corporate information, perform analysis, and define business plans. A previous column discussing “self-service” business intelligence (BI) briefly mentioned two areas of focus where ML can help BI. While the user interface, the user experience (UX), matters, it’s visibility is only the tip of the iceberg. The data being supplied to the UX is even more important.
While that is important, being able to trust the information being displayed is even more critical. AI and machine learning can help address that challenge.
It Really Does Start With Data
While mainframes still exist, the day of the mainframe controlling all data and information is long gone. While the 1990s saw attempts at data warehouses, information is a fluid commodity that exists in too many places to ever make the warehouse the “single version of truth” that some hoped. Today’s data lake is just the operational data store on steroids. It will help but it will no more be a single repository than have the previous attempts at the same thing.
Data exists in so many systems and the growth of IoT and cloud computing means data keeps extending further away from the core of on-premises computing. Working to track all the data and decide what is information is an increasingly complex problem.
Therefore, the enterprise has three key issues with the modern explosion in data:
- Where is the data?
- Which data is important enough to be tracked as information?
- Which people should have what access to all those pieces of information?
Without addressing those issues, the business is at risk through poor decision making based on inaccurate data and from increasingly strong data compliance regulations.
Don’t Re-invent The Wheel
Given the challenge, a solution is needed. Fortunately, there is no need to start from scratch. Rather, there are techniques in other areas of software that can be leveraged and adapted to the problem. ML concepts and other tools can be borrowed from other areas of IT to help both compliance and business decision making.
Machine learning is making inroads in network and application security. Trained deep learning systems are investigating transactions to look for anomalies and identify attacks and other security risks. At same time, asset management systems are being pushed by both the explosion of mobile devices and the growth of SaaS applications to better understand what physical and intellectual property assets are connected to the corporate networks and infrastructure.
Those techniques can be used to query network nodes looking for data sources in order to help build an improved corporate metadata model. Transactions on the network can be interrogated for new information and for appropriate usage.
Helping Self-Service Through Data Management
Of critical importance, the ML system can help improve access to data alongside managing compliance. It’s not enough in BI to find exceptions and identify risk. If analytics are truly to become self-service, faster access to information is necessary.
In today’s model, compliance rules and analyst decisions set an employee’s access to databases and specific fields. That significantly restricts self-service through the simple fact that we can’t imagine all needs ahead of time.
As NLP provides an easy way for personnel to query business information, to understand business processes, and to discover new relations between business data, there will regularly come ideas based in intuition and insight. A manager will ask a question about data or relationships she hasn’t previously considered, request data not yet accessible, or otherwise attempt to extend past the hard-set information boundaries.
In the traditional process, that means the investigation comes to a sudden stop, emails must be sent to IT, discussions must happen and then systems must be adjusted to allow new access rules.
An ML system can significantly speed that process, using rules and experience to quickly find new data, see if existing data fits within compliance rules and grant immediate access, or flag the request for immediate review by a compliance officer.
This challenge is more complex than what is happening now with changes in the UX, but the challenge is just as important. It doesn’t matter how easily a manager can ask a question if there isn’t a quick way to understand where the information to answer the question resides and to decide if the questioner has the authority to know the answer.
Machine learning provides a potential to far better manage enterprise information in today’s distributed world. While the industry looks at ways to ask better questions, it needs to be looking at how to locate and manage the information that provides answers.
You Might Also Like: Artificial Intelligence And Prosthetics Join Forces To Create New Generation Bionic Hand