# US 2021/0342313 A1 · Autobuild log anomaly detection methods and systems

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0342313 A1
- Document ID: US20210342313A1
- Status: Published application
- Kind: A1
- Application number: 15/929,476
- Filed: 2020-05-04
- Published: 2021-11-04
- Earliest claimed: 2020-05-04
- Assignees: salesforce.com, inc.
- Applicants: salesforce.com, Inc.
- Inventors: Kevin Riddell
- Web page: https://patentgrep.com/patent/US20210342313A1

## CPC

- G06F 16/1873: {Versioning file systems, temporal file systems, e.g. file system supporting different historic versions of files}
- G06F 7/5443: {Sum of products (for applications thereof, see the relevant places, e.g. G06F17/10, H03H17/00)}
- G06F 11/368: {for test version control, e.g. updating test cases to a new software version}
- G06F 11/3692: {for test results analysis}
- G06F 16/1734: {Details of monitoring file system events, e.g. by the use of hooks, filter drivers, logs}
- G06F 16/1824: {implemented using Network-attached Storage [NAS] architecture (distributed or networked storage systems G06F3/067; protocols for distributed storage of data in a network H04L67/1097)}
- G06F 8/65: Updates (security arrangements therefor G06F21/57)
- G06F 11/3476: {Data logging (G06F11/14, G06F11/2205 take precedence)}
- G06F 2201/81: Threshold
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Computing systems, database systems, and related methods are provided for detecting anomalies within a log file. One method involves obtaining log data for test runs executed with respect to a compiled version of executable code for an application platform, filtering the log data based on one or more performance metrics to obtain reference log data, converting the reference log data to a corresponding numerical representation and generating a matrix of the numerical representation. For each line of test log data associated with an update to the executable code, the method converts the line into a numerical representation, determines a difference between the numerical representation and the matrix, and provides an indication of an anomaly when the difference is greater than a detection threshold.

## Claims

### Claim 1

A method of detecting anomalies with respect to an update to executable code for an application platform, the method comprising:
obtaining log data associated with a plurality of tests executed with respect to a compiled version of the executable code for the application platform;
filtering the log data based on one or more performance metrics associated with the plurality of tests to obtain reference log data;
converting the reference log data to a corresponding numerical representation;
generating a matrix of the numerical representation of the reference log data;
obtaining test log data associated with one or more tests executed against an updated compilation of the executable code including the update; and
for each line of the test log data:
converting the respective line of the test log data into a respective numerical representation;
determining a difference between the respective numerical representation and the matrix; and
providing an indication of an anomaly associated with the respective line of the test log data when the difference is greater than a detection threshold.

### Claim 2 (depends on 1)

The method of claim 1, wherein obtaining the test log data comprises obtaining a test results log file associated with the updated compilation of the executable code from a testing system.

### Claim 3 (depends on 2)

The method of claim 2, wherein:
converting the respective line of the test log data into the respective numerical representation comprises converting a log line of the test results log file into a numerical vector representation; and
determining the difference comprises calculating a value of a dot product between the numerical vector representation and the matrix.

### Claim 4 (depends on 3)

The method of claim 3, wherein providing the indication comprises providing a graphical representation of the log line and the value of the dot product when the value of the dot product is greater than the detection threshold.

### Claim 5 (depends on 4)

The method of claim 4, wherein:
obtaining the log data comprises obtaining a plurality of test result log files associated with the plurality of tests previously executed with respect to compiled versions of the executable code for the application platform prior to the update;
filtering the log data comprises identifying a reference subset of the plurality of test result log files having respective values for the one or more performance metrics satisfying one or more selection thresholds; and
the reference log data comprises the reference subset of test result log files.

### Claim 6 (depends on 5)

The method of claim 5, further comprising filtering nondescript information from the reference subset of test result log files to obtain filtered reference test result log files, wherein:
converting the reference log data to the corresponding numerical representation comprises converting remaining content of each respective log line of each respective one of the filtered reference test result log files to a corresponding sequence of numerical values associated with the respective log line of the respective filtered reference test result log file.

### Claim 7 (depends on 6)

The method of claim 6, wherein:
generating the matrix comprises constructing a multidimensional matrix configured to maintain the corresponding sequence of numerical values in association with the respective log line of the respective filtered reference test result log file; and
calculating the value comprises calculating the value of the dot product between the numerical vector representation and the multidimensional matrix.

### Claim 8 (depends on 1)

The method of claim 1, wherein:
obtaining the log data comprises obtaining a plurality of test result log files associated with the plurality of tests previously executed with respect to compiled versions of the executable code for the application platform prior to the update;
filtering the log data comprises identifying a reference subset of the plurality of test result log files having respective values for the one or more performance metrics satisfying one or more selection thresholds; and
the reference log data comprises the reference subset of test result log files.

### Claim 9 (depends on 8)

The method of claim 8, further comprising filtering nondescript information from the reference subset of test result log files to obtain filtered reference test result log files, wherein:
converting the reference log data to the corresponding numerical representation comprises converting remaining content of each respective log line of each respective one of the filtered reference test result log files to a corresponding sequence of numerical values associated with the respective log line of the respective filtered reference test result log file.

### Claim 10 (depends on 9)

The method of claim 9, wherein generating the matrix comprises constructing a multidimensional matrix configured to maintain the corresponding sequence of numerical values in association with the respective log line of the respective filtered reference test result log file.

### Claim 11 (depends on 1)

The method of claim 1, wherein:
determining the difference comprises determining a value of a dot product of the respective numerical representation and the matrix; and
providing the indication comprises providing the indication when the value of the dot product is greater than the detection threshold.

### Claim 12 (depends on 1)

The method of claim 1, wherein the application platform is capable of providing instances of a virtual application generated using data maintained in a database.

### Claim 13 (depends on 1)

The method of claim 1, further comprising generating a file comprising an anomalous subset of lines of the test log data, wherein the difference associated with each respective line of the anomalous subset is greater than the detection threshold.

### Claim 14

A non-transitory computer-readable medium having instructions stored thereon that, when executed by a processing system, cause the processing system to:
obtain log data associated with a plurality of tests executed with respect to a compiled version of executable code for an application platform;
filter the log data based on one or more performance metrics associated with the plurality of tests to obtain reference log data;
convert the reference log data to a corresponding numerical representation;
generate a matrix of the numerical representation of the reference log data;
obtain test log data associated with one or more tests executed against an updated compilation of the executable code including an update to the executable code; and
for each line of the test log data:
convert the respective line of the test log data into a respective numerical representation;
determine a difference between the respective numerical representation and the matrix; and
provide an indication of an anomaly associated with the respective line of the test log data when the difference is greater than a detection threshold.

### Claim at position 15

15-20. (canceled)

### Claim 21 (depends on 14)

The non-transitory computer-readable medium of claim 14, wherein the instructions cause the processing system to obtain the test log data by obtaining a test results log file associated with the updated compilation of the executable code from a testing system.

### Claim 22 (depends on 21)

The non-transitory computer-readable medium of claim 21, wherein the instructions cause the processing system to:
convert the respective line of the test log data into the respective numerical representation by converting a log line of the test results log file into a numerical vector representation; and
determine the difference by calculating a value of a dot product between the numerical vector representation and the matrix.

### Claim 23 (depends on 14)

The non-transitory computer-readable medium of claim 14, wherein:
the log data comprises a plurality of test result log files associated with the plurality of tests previously executed with respect to compiled versions of the executable code for the application platform prior to the update; and
the reference log data comprises a reference subset of the plurality of test result log files having respective values for the one or more performance metrics satisfying one or more selection thresholds.

### Claim 24 (depends on 14)

The non-transitory computer-readable medium of claim 14, wherein the instructions cause the processing system to:
determine the difference by determining a value of a dot product of the respective numerical representation and the matrix; and
provide the indication by providing the indication when the value of the dot product is greater than the detection threshold.

### Claim 25 (depends on 14)

The non-transitory computer-readable medium of claim 14, wherein the instructions cause the processing system to generate a file comprising an anomalous subset of lines of the test log data, wherein the difference associated with each respective line of the anomalous subset is greater than the detection threshold.

### Claim 26

A system comprising:
a data storage element; and
one or more computing devices coupled to the data storage element and a network to:
obtain log data associated with a plurality of tests executed with respect to a compiled version of executable code for an application platform via the network;
filter the log data based on one or more performance metrics associated with the plurality of tests to obtain reference log data;
convert the reference log data to a corresponding numerical representation;
generate a matrix of the numerical representation of the reference log data;
obtain test log data associated with one or more tests executed against an updated compilation of the executable code including an update to the executable code; and
for each line of the test log data:
convert the respective line of the test log data into a respective numerical representation;
determine a difference between the respective numerical representation and the matrix; and
provide an indication of an anomaly associated with the respective line of the test log data when the difference is greater than a detection threshold.
