Does Business Intelligence (BI) have a future outside of traditional transactional reporting? And just how much is artificial intelligence (AI) going to change the way BI is typically deployed? To learn more about the future of BI, IDM spoke with Glen Rabie, CEO and founder of Yellowfin, an Australian Business Intelligence (BI) vendor with a global footprint, currently boasting more than 2 million users in more than 70 countries. Founded in 2003 in response to the complexity and costs associated with implementing and using traditional BI tools, Yellowfin is a 100 percent Web-based reporting and analytics solution. 

IDM: The term ‘predictive analytics’ is one that’s causing a lot of discussion and change in the legal fraternity particularly with e-discovery. You have introduced predictive analytics with the Yellowfin 7.2 release, how do you understand that term?

GR: Predictive analytics forms part of this next wave of BI. Today, if you’re a business user and you want an answer, either the answer is pre-baked for you, so someone else has built it, or you have to use an analytic tool to try and get that answer yourself.  Now natural language processing, which could be voice or text based, would simply be me asking that question in a way that makes sense for me to ask it, for instance how many contracts did we close last year?  The analytic engines need to be smart enough and to have that semantic layer that sits behind it to understand, firstly, what you’re asking and secondly, what data is out there that supports that question. It also must be able to marry the two together so that you as a user don’t have to use a GUI that seems complicated to a business user. Instead of being exposed to lots of buttons and lots of fields, you can just simply ask the question. The fundamental challenge is how do you automate some of the tasks that data analysis does today.  So a business user can get to their insights faster without the need for an intermediary to build content and to do that discovery for them.  I think that’s it’s something you’ll see more of; I don’t know if it’ll happen this year but I think that’s certainly the trend you’re starting to see in industry.

IDM: Yellowfin offers Virtualised Integrated Data Preparation Model, which promises to provide structure to your data; is this where you’re utilising a lot of those tools and how is this providing structure?

GR: At the core of enabling people to ask questions is an analytics engine that understands what people are saying which gets down to the semantic layer. When you’re building these models, you need to be very clear about what things mean. For instance if I ask about the profitability of my business, there has to be a defined calculation somewhere and a rule set for what profitability is. That’s what that data preparation does, it basically enables those rule sets to be defined for end users to not need to think about them. One of the big questions in banking was, well what defines a customer?  Is a customer someone who’s got an open account with you?  Is it a customer that has transacted with you?  Is it a customer with a certain level of loans accounts?

If you think about BI and analytics as fundamentally as content engines, then you can see a whole lot of use cases where AI machine learning would be applied to that content and deliver a kind of content syndication engine which is able to deliver curated and personalised content to a particular user. I think that that makes a whole lot of sense in BI, if I am a salesperson I’d really want to see sales content delivered to me, I don’t want to see necessarily product content, and the same for my analytics: I’d want to see analytics that was really targeted to me, who my customers are, those we should be targeting, all that kind of stuff, and I think that really is the next evolution of analytics.

"There has been a broadening of what constitutes a data source that an organisation’s interested in." - YellowfinCEO and founder, Glen Rabie

IDM:  As the application of BI technology is broadening are you finding that you’re having conversations with different people, whereas in the past it was just finding people reporting on financial or transactional or operations, are there now people who are looking at governance and risk and other aspects?

GR:  Very much so. When I started in BI  20-odd years ago but it was a finance function, so if you sold product you sold it to finance.  These days, corporate risk is a huge part of the buy of analytic technologies. There is more focus on the unstructured data as well. Organisations need a fundamental understanding of all the documentation they have and must be able to categorise, functionalise and then report off it as well. For a form-driven business like banking, you need to extract as much value out of those documents as humanly possible.  You also want to identify risk from emails as possible. We’ve partnered with a number of people, specifically in the document space, around providing document analytics,

There has been a broadening of what constitutes a data source that an organisation’s interested in. In the past finance owned BI because they were the keepers of the financial metrics and that was the most obvious are to get reporting on.  That problem’s been largely solved, I don’t think anyone’s too excited by doing just analytics on financials.  Now, as an organisation matures it starts to ask where are these other pockets of data that could, if I understood it, could have a huge impact on my business?

IDM: BI is traditionally used to report on transactional data, the stuff that lives in as tables or columns and rows in a standard database, whereas 85% of corporate data that’s unstructured and lives in file shares, EDRMS, email or paper; how can Yellowfin help here and where does it play in that space?

GR: That’s a really good question. The answer is that all unstructured data can become structured data and there are now plenty of toolsets out there that can accomplish. At Yellowfin we partner with a number of the vendors of these tools to undertake tasks ranging from simple structuring of content up to full text mining. One of our customers is a large bank which employs these techniques to go through emails, go through contracts and to actually identify risk within those and to structure those risk metrics so that they can be reported on and monitored and taken action against.  These tools will allow organisations to deal with the propensity for employees to talk to competitors, for instance.

One of the partners we have been working with locally is solutions provider Birnam Wood using the  Worldwide Applications Smart Data Platform.

Sentiment analysis is an interesting challenge, as depending on the audience, sentiment and what drives sentiment can be really vastly different. If you’ve got quite a sarcastic audience then it’s very difficult to pick up whether the person’s statements are positive or negative based, unless you are able to contextualise. There is a lot of work and a lot of thought that has to go into doing it properly, but it can be done.