The Idea:

The same customer may appear as slightly different duplicates in the same dataset, or different records in different databases.

In these different instances of the same customer, certain fields may be missing; there may be changes in format, typographical errors, omissions, or alternative spellings, abbreviations.

Artificial Intelligence and Data Mining:

Chalcedon uses latest developments in artificial neural networks. It can be trained to learn the characteristics of data. Combining this with a powerful rule-based matching algorithm that uses fuzzy logic, it can detect similarities efficiently and with high precision.

Chalcedon calculates a Similarity Score between Records:

Chalcedon first calculates similarities between fields and then later combines them to calculate a total similarity score.

Record pairs with scores higher than a threshold are automatically matched, and those above a critical value may be listed for manual control.

Chalcedon is Robust:

Making use of artificial intelligence, matching is insensitive to changes in format, alternative spellings, or small typographical errors, as well as to missing data. Chalcedon adapts itself to such changes automatically.

Chalcedon can Learn from Examples:

Chalcedon is trained by using a small example set of manually matched record pairs. Rules learned from such pairs are then applied to the whole database.

Chalcedon is Flexible:

Efficient: Can run online or in batch mode.
Adaptable: New fields, formats, operators can be added.
Platform independent: Can run on any operating system/hardware and interface to any database/warehouse.
Scalable: On large databases, matching can be distributed on multiple processors/machines.

Click here to download the brochure