QUERY TREES INCLUDING OR NODES FOR EVENT FILTERING

Patent №

US 6,918,124

Granted

2005-07-12

Filed 2000

Owner

MICROSOFT CORPORATION

AI components

3

ml · kr · hardware

Assignment

Recorded

Dataset

AIPD

2023_r1 edition

Application

09517895

Filtering trees for selectively notifying subscribers of events are provided, and are constructed with OR nodes to substantially reduce their size. The filtering trees have nodes representing event variables that ultimately branch to leaf nodes thereunder, and the leaf nodes identify which of a set of queries are satisfied by an actual event. A mechanism recursively merges nodes of trees into a single tree, and uses OR nodes when nodes cannot be combined, to essentially add a parallel path in the resulting tree to traverse. Nodes that can be combined are those that represent the same event variable, and may have data points that are merged into a combined node. Threshold gains in efficiency may be evaluated to determine whether the original trees should be kept instead of the resulting tree.

Machine learningKnowledge representationAI hardwareH04L 41/0604G06F 11/0709G06F 11/0781G06F 16/24568H04L 41/0213

AI classification

AI hardware1.00
Machine learning0.89
Knowledge representation0.68
Planning0.09
Vision0.01
Natural language0.01
Speech0.00
Evolutionary computation0.00

Ownership

MICROSOFT CORPORATION

assignment · 106610358

Assignors

NOVIK, LEV, HUDIS, IRENA, MCCOLLUM, RAYMOND W.

On an employer assignment, the assignors are typically the inventors.

© 2026 NYSGPT2525 LLC