For mailRetrieval there are the following, typical use scenarios, which have in common the error-tolerant search for addresses in large databases.

In this use scenario, with just a few name and address fragments (example: last name/first name, only the name or only the city) you can make an error-tolerant search for an address stored in the address database and find it. Typically this is done when a customer contacts the company, but doesn't have his customer number at hand or conveys the wrong number.

mailRetrieval usually delivers the results in less than a second - even for databases containing several million addresses. And, mailRetrieval supplies optimal results even when there are errors in the information the user enters or in the information the database contains. The results are delivered back to the user according to the degree of agreement.

Automaticaly, without any user impetus, mailRetrieval checks just before address sets are entered or changed whether a person or a company has already been stored in the database. The system recognizes if this is the case even if there are some deviations between the entry and the stored data. The system also differentiates whether it is the question of a duplicate with very high probability or only with some probability. Depending on the application, the user can be informed about this issue and can thus clarify in the call center this suspicion of a duplicate, which only exists with a certain degree of probability, directly on the telephone with the caller.

For every new entry or change of addresses different "cluster views" of an address are automatically generated here and corresponding "cluster identifications" are placed into the address database. Thereby it is possible to generate different clusters both independently or in a hierarchy. Even with only "some" uncertainty it is possible to assign the address to a cluster and to store the "probability" with which an address belongs to a certain cluster along with it in the database.

Some practical examples for typical clusters

• for consumer addresses: the person, the household or all addresses in a building. 
• for business addresses: department/contact person in a company or all contacts within a company.
• creditworthiness and risk clusters.

With mailRetrieval several of the scenarios described above can be used combined in a single application.

All intelligent retrieval functions can be implemented in an error-tolerant way due to Uniserv's specially developed method not only for all name and address elements, but also for telephone numbers, e-mail addresses, dates of birth etc. according to your notions - and this in the shortest response time (related to that see also error-tolerant search methods).

Sitemap | Webmaster | Disclaimer | Privacy Policy | Imprint | © 2008 Uniserv GmbH