Knowledge Discovery and Dissemination (KDD) Program
The objective of KDD is to improve the production of actionable intelligence from multiple, disparate, large data sets, including new or unanticipated data sets. KDD is focused on two research tasks:
- Quickly align the terminology, organization and semantics of different data sets so that analysts can have a common model for queries and analysis.
- Develop advanced analytic algorithms that can effectively draw inferences across multiple, aligned data sets.
With these two capabilities, it may be possible for the IC to create “virtual” fusion centers where analysts can have many of the benefits of a fusion center without the costs of physically creating one.
Enabling Analysts to Produce Actionable Intelligence from Multiple Sources
The development approach is designed to ensure that the research developed under the KDD program is relevant to the IC and performs well on real data. Each year, KDD contractors are provided with several real data sets and sample problems to support their research. KDD contractors then deliver their research in the form of prototype systems that are tested by IC analysts using different sets of real data and different analytic problems.
The KDD program began in October 2010 with five research teams comprised of 17 academic and 12 commercial organizations. A Base Year evaluation in September 2011 was used to exercise the prototype systems, rehearse the evaluation process and give researchers insight into IC analysis. A formal evaluation was conducted in September 2012. Results are currently being assessed and should be completed in November 2012.