# US 11,113,689 B2 · Transaction policy audit

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

- Number: US 11,113,689 B2
- Document ID: US11113689B2
- Status: Granted
- Kind: B2
- Application number: 16/577,997
- Filed: 2019-09-20
- Published: 2021-01-07
- Granted: 2021-09-07
- Earliest claimed: 2019-07-03
- Assignees: Sap SE
- Applicants: Sap SE
- Inventors: Michael Stark; Evan Adkins; Adithya Kumar; Suchitra Sundararaman; Jesper Lind
- Web page: https://patentgrep.com/patent/US11113689B2

## CPC

- G06Q 40/12: Accounting
- G06F 18/24: Classification techniques
- G06F 40/284: Lexical analysis, e.g. tokenisation or collocates
- G06F 40/30: Semantic analysis
- G06N 3/02: Neural networks
- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- G06N 3/08: Learning methods
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 3/09: Supervised learning
- G06N 20/00: Machine learning
- G06Q 20/045: {using payment protocols involving tickets}
- G06Q 20/389: {Keeping log of transactions for guaranteeing non-repudiation of a transaction}
- G06Q 20/40: Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
- G06Q 20/4016: {involving fraud or risk level assessment in transaction processing}
- G06T 7/0002: {Inspection of images, e.g. flaw detection}
- G06T 7/74: {involving reference images or patches}
- G06V 20/95: {Pattern authentication; Markers therefor; Forgery detection}
- G06V 30/224: of printed characters having additional code marks or containing code marks
- G06V 30/40: Document-oriented image-based pattern recognition
- G06V 30/413: Classification of content, e.g. text, photographs or tables
- G06V 30/414: Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text
- G06V 30/418: Document matching, e.g. of document images
- G06F 16/24564: {Applying rules; Deductive queries}
- G06T 2207/20061: Hough transform
- G06T 2207/30176: Document
- 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
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

The present disclosure involves systems, software, and computer implemented methods for transaction auditing. One example method includes receiving receipt data associated with an entity. Policy questions associated with the entity are associated with at least one policy question answer that corresponds to a conformance or a violation of a policy selected by the entity. For each policy question, a machine learning policy model is identified for the policy question that includes, for each policy question answer, receipt data features that correspond to the policy question answer. The machine learning policy model is used to automatically determine a selected policy question answer to the policy question by comparing features of extracted tokens to respective receipt data features of the policy question answers that are included in the machine learning policy model. In response to determining that the selected policy question answer corresponds to a policy violation, an audit alert is generated.

## Claims

### Claim 1

A computer-implemented method comprising:
receiving receipt data associated with a request associated with a first policy-enforcer entity, wherein the receipt data includes tokens extracted from at least one receipt;
identifying policy questions associated with the first policy-enforcer entity, wherein each policy question is associated with at least one policy question answer, and wherein each policy question answer corresponds to a conformance or a violation of a policy selected by the first policy-enforcer entity, wherein the identified policy questions associated with the first policy-enforcer entity include a first set of policy questions specific to the first policy-enforcer entity and a second set of policy questions common to multiple policy-enforcer entities, wherein the multiple policy-enforcer entities include the first policy-enforcer entity and at least a second policy enforcer entity that is a different entity than the first-policy enforcer entity; and
for each respective policy question in the identified policy questions:
identifying a machine learning policy model for the respective policy question based on a mapping associated with the first entity that maps policy questions to machine learning policy models, wherein the machine learning policy model is trained based on historical determinations of policy question answers for the respective policy question for historical receipt data, wherein the historical determinations of policy question answers includes different historical determinations for different policy-enforcer entities of the multiple policy-enforcer entities, and wherein the machine learning policy model includes, for each policy question answer, receipt data features that correspond to the policy question answer;
using the machine learning policy model to automatically determine a selected policy question answer to the respective policy question by comparing features of the extracted tokens to respective receipt data features of the policy question answers that are included in the machine learning policy model; and
in response to determining that the selected policy question answer corresponds to a policy violation, generating an audit alert.

### Claim 2 (depends on 1)

The method of claim 1, wherein the historical determination of policy question answers to the respective policy question are determined, based on the historical receipt data, by previous executions of the machine learning policy model.

### Claim 3 (depends on 2)

The method of claim 2, wherein the historical receipt data includes information from multiple policy-enforcer entities, wherein the multiple policy-enforcer entities include the second policy-enforcer entity and the first policy-enforcer entity.

### Claim 4 (depends on 1)

The method of claim 1, further comprising:
failing to identify, for a first policy question, a machine learning policy model for the first policy question; and
forwarding the receipt data to a secondary review process in response to failing to identifying a machine learning policy model for the first policy question.

### Claim 5 (depends on 1)

The method of claim 1, wherein different policy questions that have a same semantic meaning are mapped to a same machine learning policy model.

### Claim 6 (depends on 1)

The method of claim 1, wherein identifying the machine learning policy model comprises identifying a keyword-based model.

### Claim 7 (depends on 6)

The method of claim 6, wherein the keyword-based model is configured to identify one or more keywords in the extracted tokens.

### Claim 8 (depends on 1)

The method of claim 1, wherein identifying the machine learning policy model comprises identifying a neural network model.

### Claim 9 (depends on 8)

The method of claim 8, wherein the neural network model is a recurrent neural network model.

### Claim 10 (depends on 8)

The method of claim 8, wherein the neural network model is configured to perform character analysis of the receipt data to identify features that indicate a policy violation or a policy conformance.

### Claim 11 (depends on 10)

The method of claim 10, wherein the features include keyword patterns, receipt text format, and receipt layout.

### Claim 12

A system comprising:
one or more computers; and
a computer-readable medium coupled to the one or more computers having instructions stored thereon which, when executed by the one or more computers, cause the one or more computers to perform operations comprising:
receiving receipt data associated with a request associated with a first policy-enforcer entity, wherein the receipt data includes tokens extracted from at least one receipt;
identifying policy questions associated with the first policy-enforcer entity, wherein each policy question is associated with at least one policy question answer, and wherein each policy question answer corresponds to a conformance or a violation of a policy selected by the first policy-enforcer entity, wherein the identified policy questions associated with the first policy-enforcer entity include a first set of policy questions specific to the first policy-enforcer entity and a second set of policy questions common to multiple policy-enforcer entities, wherein the multiple policy-enforcer entities include the first policy-enforcer entity and at least a second policy enforcer entity that is a different entity than the first-policy enforcer entity; and
for each respective policy question in the identified policy questions:
identifying a machine learning policy model for the respective policy question based on a mapping associated with the first entity that maps policy questions to machine learning policy models, wherein the machine learning policy model is trained based on historical determinations of policy question answers for the respective policy question for historical receipt data, wherein the historical determinations of policy question answers includes different historical determinations for different policy-enforcer entities of the multiple policy-enforcer entities, and wherein the machine learning policy model includes, for each policy question answer, receipt data features that correspond to the policy question answer;
using the machine learning policy model to automatically determine a selected policy question answer to the respective policy question by comparing features of the extracted tokens to respective receipt data features of the policy question answers that are included in the machine learning policy model; and
in response to determining that the selected policy question answer corresponds to a policy violation, generating an audit alert.

### Claim 13 (depends on 12)

The system of claim 12, wherein the historical determination of policy question answers to the respective policy question are determined, based on the historical receipt data, by previous executions of the machine learning policy model.

### Claim 14 (depends on 13)

The system of claim 13, wherein the historical receipt data includes information from multiple policy-enforcer entities, wherein the multiple policy-enforcer entities include the second policy-enforcer entity and the first policy-enforcer entity.

### Claim 15

A computer program product encoded on a non-transitory storage medium, the product comprising non-transitory, computer readable instructions for causing one or more processors to perform operations comprising:
receiving receipt data associated with a request associated with a first policy-enforcer entity, wherein the receipt data includes tokens extracted from at least one receipt;
identifying policy questions associated with the first policy-enforcer entity, wherein each policy question is associated with at least one policy question answer, and wherein each policy question answer corresponds to a conformance or a violation of a policy selected by the first policy-enforcer entity, wherein the identified policy questions associated with the first policy-enforcer entity include a first set of policy questions specific to the first policy-enforcer entity and a second set of policy questions common to multiple policy-enforcer entities, wherein the multiple policy-enforcer entities include the first policy-enforcer entity and at least a second policy enforcer entity that is a different entity than the first-policy enforcer entity; and
for each respective policy question in the identified policy questions:
identifying a machine learning policy model for the respective policy question based on a mapping associated with the first entity that maps policy questions to machine learning policy models, wherein the machine learning policy model is trained based on historical determinations of policy question answers for the respective policy question for historical receipt data, wherein the historical determinations of policy question answers includes different historical determinations for different policy-enforcer entities of the multiple policy-enforcer entities, and wherein the machine learning policy model includes, for each policy question answer, receipt data features that correspond to the policy question answer;
using the machine learning policy model to automatically determine a selected policy question answer to the respective policy question by comparing features of the extracted tokens to respective receipt data features of the policy question answers that are included in the machine learning policy model; and
in response to determining that the selected policy question answer corresponds to a policy violation, generating an audit alert.

### Claim 16 (depends on 15)

The computer program product of claim 15, wherein the historical determination of policy question answers to the respective policy question are determined, based on the historical receipt data, by previous executions of the machine learning policy model.

### Claim 17 (depends on 16)

The computer program product of claim 16, wherein the historical receipt data includes information from multiple policy-enforcer entities, wherein the multiple policy-enforcer entities include the second policy-enforcer entity and the first policy-enforcer entity.