Rete algorithm

The Rete algorithm (Latin rete, net ',' Network ' ) is an algorithm and expert system for pattern recognition and mapping of system processes via rules. It was developed by U.S. computer scientist Charles Forgy as part of his doctoral thesis at Carnegie Mellon University, which was to lead him to the title in 1979. The algorithm is free, as the U.S. Defense Department was sufficiently at its creation involved with.

Objectives of the algorithm

The Rete algorithm, has been developed from the viewpoint of ensuring a highly efficient processing rule. In addition, large rule sets can be treated performant. When developed, it was superior to the existing systems by a factor of 3000.

Use and variations

RETE today forms the basis of many control systems, such as various Prolog interpreter or so-called business rules engines, and has established itself for this as the de facto standard.

So far, there are two direct descendants Rete II and Rete III. Both are about 50 times faster than the original approach. Rete III includes a few extensions that can easily increase its efficiency further. The development was closely modeled on the platform DEC XCON and found its implementation first in OPS systems, especially in OPS2 or later in OPS5. Save millions of dollars have been designated. The implemented rules had about 10,000 members in its final form.

Drools is an example of a business rule engine, which was based on the Rete algorithm. Meanwhile (version 6.0 ) is in Drools a successor algorithm ( PHREAK ) active.

Disadvantages

The high speed of the algorithm is at the expense of the shared memory.