Is there a case for federated data models?

By Gus Verzosa

With so much useful data available to organisations today that doesn't reside in-house combined with the reality that true data sharing is still a long way off, is there a need to move to federated data models?

We believe there is a strong case for federated data models, particularly within the public sector and we have been working with some Australian federal agencies on refining what this might look like and how it would work.  As a result Altis have developed a Federated Unification Method (FUM) - a method that shares and integrates data from multiple organisations using limited, secured data from each organisation on a “need-to-access” basis for a specific set of objectives.  This is a paradigm shift from the traditional centralised approach and has a profound impact on business intelligence and analytics. This approach uses the following steps:

  • Data sets (both structured and unstructured) from various source systems are processed using rules and algorithms to generate a subset of key alert or critical indicator records which are relevant to the objective of the whole BI/analytics program. This results in a significantly-reduced data set per source system and data is de-identified at source to preserve privacy.
  • The reduced data sets are then consumed, consolidated, cleansed, and integrated for reporting and interrogation by the business intelligence platform.
  • The integrated data set is then fed into the analytics platform for mining, prediction, and prescription. A key component in this step is the use of semantic computing via an integration ontology (a computer-based representation of the entities and the relationships among entities for a given business or knowledge domain).
  • The ontology then is able to visually show and report on points of convergence in the alert/critical indicators (red flags/green flags). These points of convergence represent actionable intelligence that narrows down the effort and focus of exploration and investigation into high-probability scenarios or patterns.

The FUM has the following advantages:

  • Requires data on “need” basis – only what is needed and when it is needed
  • Handles privacy and security through indexing at source but anonymous/masked/de-identified/aggregated data from each organisation
  • Data requirement from each participating organisation is driven by objectives and use cases - which is far more efficient than the  inefficient “give me all you have” approach
  • Integrates using user-friendly ontology

The FUM has been applied to 2 high-profile use cases presented in recent public sector conferences: Countering terrorism using information analytics; and Boosting economic productivity through analytics & data sharing in the public sector. Feedback from conference participants as well as further research has revealed its potential application in pre-emptive intervention for both domestic violence and child abuse.

Altis has developed a FUM approach and would be more than happy to discuss these with you and they could improve your current business intelligence and analytics applications, contact us to find out more.

We welcome your comments and thoughts on whether you believe there is a need for a paradigm switch from a traditional centralised data analysis approach to a federated approach to enable better decision making in organisations.

Join the conversation

Captcha Image

Comments


Post has no comments.

Would you like to be kept in the loop on courses, events and other related topics?

Simply complete your details and we’ll add you to our list.