More Nuts and Bolts for Entity Resolution Success
Posted by 
Share this Post
Make sure you have the nuts and bolts of entity resolution success before you start your initiative.
Last week I began discussing some of the most important factors to consider when evaluating entity resolution. Today, I’m adding on to that list:
Flexible and Multi-Domain Data Models
Standard data model templates that can be easily modified to suit specific data needs should be provided as a baseline for quick delivery. In particular, the ability to resolve any type of “noun” such as persons, objects, locations and events (POLE) should be supported.
Rapid Implementation
Most software implementations take too long and can disrupt an organization’s ongoing investigations. Entity resolution technologies must be able to be quickly configured and deployed, support the ability to rapidly implement relevant user feedback, and achieve full operational capability (FOC) within an aggressive time period.
Easy and Open Integration
Agencies today are implementing large, complex solutions, instead of the disparate applications and systems of the past. These new deployments have complex requirements that require the integration of numerous vendor technologies.
Technologies that are part of these solutions must support multiple production-ready integration options including services oriented architecture (SOA), web services, standard ETL, batch-based interfaces, and other points of integration, such as software development kits (SDKs).
Dynamic Implementation Styles
A range of hub implementation styles should be supported to meet the unique data control, authorship and persistence requirements of individual missions and organizations. Implementation flexibility also enables the ability to extract incremental value by allowing the data to reside where it should, and where it makes the most sense.
Data security
Roles based access controls (RBAC) that protects data in accordance with established policy, guidance and system controls should be supported.
Data Privacy
The system need to match data across secure domains, including encrypted, anonymized and obfuscated data, and it also needs to be able to encrypt, anonymize or obfuscate data before sharing.
Reporting and Analytics
The system must support efficient and seamless integration with an organization’s analytics/reporting application of choice whether through an adaptor or coarse-grained services such as SOA and/or APIs.
There’s still one big factor to consider: total cost of ownership. I’ll discuss TCO in detail next week.
Other than that, have I missed anything? Add your additions or comments below.
3 Responses »
Leave a Response
Entries(RSS)
Great list, Jonathan.
Got a message when looking for the earlier post, but another key factor I'd add beyond basic entity resolution is finding hidden relationships that are several degrees distant (ala Six Degrees of Kevin Bacon). It's valuable to constrain searching to one degree, two degrees, or whatever level is appropriate to a particular problem in a specific domain.
Bob Barker, IdentityResolutionDaily.com
Bob, I've fixed the link to the previous post - apologies. http://blog.initiate.com/wordpress/index.php/2009/11/30/what-needed-for-entity-resolution-success/
Bob - thanks for the response. I will be addressing non-obvious relationships and degrees of separation in subsequent blogs. As I'm sure you're aware, accurate entity resolution is only the first step towards correlating entities (the discovery and linking of obvious and non-obvious relationships across data elements with varying degrees of separation). This particular blog was meant to be focused on entity resolution necessities which, in turn, enable accurate and flexible entity correlation capabilities.