AnyConnection© is a powerful, unique approach to uncovering hidden relationships in large data bases. It is very useful for fraud detection and entity resolution because it allows seemingly disparate records to be joined and related, even if they are related only via another record. For example, suppose you have the following 4 records in your data base.
With the human eye, you may be able to tell that all four of the above records are related:
– Records A and B are related via similar name and same phone number.
– Records B and C are related via same bank account number.
– Records C and D are related via same phone number.
This is a simplistic example, but displays how link analysis works.
With AnyConnection©, all of the above four records would be systematically and pro-actively joined together in one cluster for further analysis and investigation.
This allows record A to be related to record D even though they do not relate to each other directly on any piece of data. This kind of record linkage is incredibly useful for identifying collusion, fraud, fictitious entities, duplicate records, and related entities.
Powerful Linkage Mechanisms
With AnyConnection©, you can literally find any connection between two records, on any field. If two records share the same or similar name, address, phone number, bank account number, SSN, Tax ID, email address, or any other piece of identifying information, our software can find the linkage using fuzzy matching logic. But AnyConnection does not stop with giving you pairs of matches, we follow each connection out to N number of degrees of separation, until there are no more connections to make. AnyConnection is link analysis without the distracting graphics which still leave you finding a needle in a haystack. With most off-the-shelf link analysis packages, you get a mess that looks like this:
With visual link analysis such as the above graph, attention is usually only drawn to the largest points of connections and you may miss out on other important connections. Plus, when your data base is very large, a graphical representation just loses its effectiveness. But with AnyConnection©, data is connected in data base format and linked with dollar amounts so that the most suspicious entities with the most dollars are detected first. AnyConnection is link analysis but we control the fuzzy matching underlying each match technique and customize the matching for each client. Each address, name, phone number, SSN, email, is matched using proprietary fuzzy matching techniques. Then the individual matches are funneled into the Entity Resolution piece of the software and grouped into families.
AnyConnection© also implements a master ranking schema, so that entities that have several connections are floated to the top of a master list. For example, when we conduct an Accounts Payable audit, we run several A/P fraud algorithms, then join them with the vendor/employee crosswalk, the vendor/mail drop matches, and other lists and rank each Vendor by the number of hits on each algorithm, and their total spend in dollars. Ranking saves time. Each potential fraud lead takes time to investigate, and we understand the time and budget constraints you may face.
Our fuzzy matching techniques include the Jaro-Winkler distance function, the Levenshtein edit distance function (Proc COMPLEV in SAS), and several user-defined functions that are proprietary to Automated Auditors, LLC. Our edit distance functions are order independent, meaning they can identify a match between “Christine Warner” and “Warner, Christy” – – the order of the words in the phrase do not matter. This is particularly useful for international addresses which often do not follow the same standardization as U.S. addresses.
Our software has been employed:
- to find terrorists on the Terrorist Watch List
- to identify excluded Medicare and Medicaid providers
- to match Medicaid and Medicare provider using fuzzy matching
- to cleanse the World Bank Vendor File
- to find fraud for the NFL, Dial, the World Bank, and other corporations
- to identify duplicate payments and duplicate health care claims
- to reconstruct an Accounts Receivable data base by linking data via student name
AnyConnection©: Linking entities A to Z even though they are only connected indirectly to entity M.