# US 2018/0322414 A1 · Machine learning auto completion of fields

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

- Number: US 2018/0322414 A1
- Document ID: US20180322414A1
- Status: Published application
- Kind: A1
- Application number: 15/674,353
- Filed: 2017-08-10
- Published: 2018-11-08
- Earliest claimed: 2017-05-04
- Assignees: ServiceNow, Inc.
- Applicants: ServiceNow, Inc.
- Inventors: Baskar Jayaraman
- Web page: https://patentgrep.com/patent/US20180322414A1

## CPC

- G06N 20/00: Machine learning
- G06F 40/174: Form filling; Merging
- G06F 40/30: Semantic analysis
- G06F 40/40: Processing or translation of natural language (natural language analysis G06F40/20; semantic analysis G06F40/30)
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06Q 10/00: Administration; Management
- G06N 5/04: Inference or reasoning models
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

Systems and methods for using a mathematical model based on historical natural language inputs to automatically complete form fields are disclosed. An incident report may be defined with a set of required parameter fields such as category, priority, assignment, and classification. Incident report submission forms may also have other free text input fields providing information about a problem in the natural vocabulary of the person reporting the problem. Automatic completion of these so-called parameter fields may be based on analysis of the natural language inputs and use of machine learning techniques to determine appropriate values for the parameter fields. The machine learning techniques may include parsing the natural language input to determine a mathematical representation and application of the mathematical representation to “match” historically similar input. Once matched the parameter values from the historically similar input may be used instead of generic default values.

## Claims

### Claim 1

A cloud-based computer system, comprising:
a memory partition;
a network interface communicatively coupled to one or more processing units and the memory partition, wherein the memory partition comprises computer instructions that when executed by the one or more processing units cause the cloud-based computer system to:
provide a hosted client instance communicatively coupled, via the network interface, with a remote client device;
receive a request from the remote client device for an incident reporting form;
obtain information in natural language text regarding an incident associated with the incident reporting form;
parse the natural language text to identify one or more N_Grams;
process the identified one or more N_Grams against a model to determine a predicted value and confidence level of the predicted value for one or more input fields of the incident reporting form;
determine that the confidence level of the predicted value is above a confidence level threshold associated with the one or more input fields; and
send the predicted value to the remote client device with an indication to use the predicted value as at least one field value in the incident reporting form rather than a generic default value.

### Claim 2 (depends on 1)

The cloud-based computer system of claim 1, wherein the computer instructions further cause the cloud-based computer system to receive an indication of user-override for the predicted value and use the indication of user-override to retrain the model.

### Claim 3 (depends on 1)

The cloud-based computer system of claim 1, wherein the computer instructions further cause the cloud-based computer system to periodically retrain the model using information obtained after initial use of the model.

### Claim 4 (depends on 1)

The cloud-based computer system of claim 1, wherein the information in natural language text regarding an incident comprises a short description field generically describing an incident.

### Claim 5 (depends on 1)

The cloud-based computer system of claim 1, wherein the at least one field value in the incident reporting form is a priority field.

### Claim 6 (depends on 1)

The cloud-based computer system of claim 1, wherein the at least one field value in the incident reporting form is a classification field.

### Claim 7 (depends on 1)

The cloud-based computer system of claim 1, wherein the at least one field value in the incident reporting form is an assignment field.

### Claim 8

A cloud-based computer system, comprising:
a memory partition;
a network interface communicatively coupled to one or more processing units and the memory partition, wherein the memory partition comprises computer instructions that when executed by the one or more processing units cause the cloud-based computer system to:
provide a hosted client instance communicatively coupled, via the network interface, with a remote client device;
receive a request to create a model configured to assist in machine learning prediction of input values;
extract historical data associated with the hosted client instance and representing completed incident reports;
process the extracted historical data to create pre-processed model input data;
transform the pre-processed model input data using transformation techniques to create transformed model data;
partition the transformed model data into training, tuning, and testing data sets; and
create a target matrix from the training data set, the target matrix comprising N_gram values associated with a target identifier as indicated by the extracted historical data.

### Claim 9 (depends on 8)

The cloud-based computer system of claim 8, wherein the request to create a model comprises a time-frame for which to extract.

### Claim 10 (depends on 8)

The cloud-based computer system of claim 8, wherein the instructions to process the extracted historical data comprise instructions to remove junk characters.

### Claim 11 (depends on 8)

The cloud-based computer system of claim 8, wherein the instructions to process the historical data comprise instructions to dedupe redundant information.

### Claim 12 (depends on 8)

The cloud-based computer system of claim 8, wherein the instructions to process the extracted historical data comprise instructions to adjust different references to a common item to a consistent reference to the common item.

### Claim 13 (depends on 8)

The cloud-based computer system of claim 8, wherein the transformation techniques include keyword extraction.

### Claim 14 (depends on 13)

The cloud-based computer system of claim 13, wherein the transformation techniques include preparing a mathematical model of English sentences in the pre-processed model input data.

### Claim 15 (depends on 14)

The cloud-based computer system of claim 14, wherein the mathematical model of English sentences is based at least in part on N_gram generation.

### Claim 16 (depends on 15)

The cloud-based computer system of claim 15, wherein both 1_gram and 2_gram values are used in the mathematical model.

### Claim 17 (depends on 8)

The cloud-based computer system of claim 8, wherein the training data set substantially comprises 80 percent of the transformed model and each of the tuning and testing data sets comprise substantially 10 percent.

### Claim 18 (depends on 8)

The cloud-based computer system of claim 8, further comprising computer instructions to cause the cloud-based computer system to:
create a confusion matrix reflecting predicted values against actual values using the tuning dataset, the confusion matrix indicating an accuracy of the model.

### Claim 19 (depends on 18)

The cloud-based computer system of claim 18, further comprising computer instructions to cause the cloud-based computer system to:
create a cost matrix associated with an error rate correlated with the accuracy of the model.

### Claim 20 (depends on 19)

The cloud-based computer system of claim 19, further comprising computer instructions to cause the cloud-based computer system to:
calculate a total cost based on the cost matrix and cost of misclassification; and
iteratively adjust confidence thresholds within the model and recalculate the total cost to identify confidence thresholds associated with a minimized cost.
