USING TRACE MATCHING TO IDENTIFY AND ANALYZE APPLICATION TRACES

Patent №

US 8,997,057

Granted

2015-03-31

Filed 2012

Owner

GOOGLE INC.

AI components

5

ml · nlp · kr · planning · hardware

Assignment

Recorded

Dataset

AIPD

2023_r1 edition

Application

13463584

Methods for identifying and analyzing performance traces are provided. Temporal logic formulas are patterns that can be compared with traces and individual events to identify the existence of certain behavior. Traces, sequences of time-stamped events in time order, are compared with one or more temporal logic formulas to identify the event sequences that match the formulas. The temporal logic formulas can be written in the simple temporal logic language that is presented. When a formula matches an event sequence, attributes from the event sequence are extracted and metric expressions are evaluated based on these attributes. The extracted attributes and the results of the metric expression are returned. This temporal logic pattern matching process can efficiently identify and analyze performance traces and significantly reduce manual effort for identifying performance problems.

AI classification

Machine learning1.00
AI hardware1.00
Planning1.00
Natural language0.98
Knowledge representation0.96
Vision0.00
Speech0.00
Evolutionary computation0.00

Ownership

GOOGLE INC.

assignment · 281660723

Assignors

DIWAN, AMER, RYCKBOSCH, FREDERICK, SITES, RICHARD LEE

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

© 2026 NYSGPT2525 LLC