Products

Mentys™ consists of six products that utilize Markov chains and Bayesian networks:

Learner

This product is designed to learn knowledge bases from data to explain or predict changes to data. It also incorporates prior knowledge from experts and application domain experts and runs only on the desktop. It currently does not have an API implementation.

Prospector

This product is designed to use knowledge bases generated by Learner to detect and track down errors according to their implications. These implications are "reversible" in the sense of time. That is, Prospector can predict forward implications as well as post-dict backward implications. Prospector has both desktop and API implementation.

Proactor

This product is a specific implementation of Prospector to work directly with very large tables, i.e., to scan an entire database. As this may involve many thousands, perhaps even millions of records, Proactor runs on the desktop non-interactively with checkpointing and restart capabilities. Proactor does not have an API implementation.

Refractor

This product is designed to learn knowledge bases from data. The primary difference between Refractor and Learner is that Refractor's knowledge base can be used to test, classify, correct, and fill data. Unlike Learner's knowledge base, it does not "explain" data. Therefore Refractor's knowledge base is very different from Learner considering its purpose is different. Refractor works in tandem with Proactor and Imputor. It has both desktop and API implementations.

Imputor

This product is designed to test, classify, correct, and fill data using knowledge bases generated by Refractor. Imputor has both desktop and API implementations.

Mentor

This product is designed to automate parameter selection and otherwise fine-tune the functions of the five other products. It is embedded within the desktop versions of those products and does not have an API implementation.

View the Mentys Installation Procedure.

Mentys applications can be developed in either C++ or Java, and will run on several platforms including Microsoft Windows®, Linux, and Sun Solaris®.

Mentys currently supports several third party data structures. The simplest are CSV (comma-separated value) files, which can be exported directly from most modern spreadsheets, including Microsoft Excel®. Tables may also be imported from any database for which there is a JDBC (Java) interface, such as Oracle or MySQL.

White Papers

• Mentys for Letter of Credit AML
• Monitoring Trading Activity
• Market Data Correction
• Credit Risk Application
• Data Recovery Application
• U.S. Treasuries Trading