Granted Patent B2

US 12,524,817 · App. 17/842,523 · Granted Jan 13, 2026

Inventors: Delia Rusu (Wellington, NZ); Hayden Jeune (Wellington, NZ); Rebecca Dridan (Wellington, NZ); Soon-Ee Cheah (Wellington, NZ); Brett Calcott (Wellington, NZ); Zhimin Wang (Wellington, NZ); Quentin-Gabriel Thurier (Wellington, NZ); Fubiao Qin (Wellington, NZ); Niklas Patrick Pechan (Wellington, NZ)

Assignee: Xero Limited

G06Q40/12G06F40/279G06F40/284G06F40/30G06N20/00

Abstract

Claims (78)

1 . A computer-implemented method comprising:

determining, by an accounting system comprising memory, and one or more processors configured to execute instructions stored in the memory, a candidate financial record associated with a transaction between a first entity and a second entity;

determining, by the accounting system, one or more first substrings from a character string of the financial record;

training a numerical representation generation model using training data comprising a corpus generated from historical transaction records;

providing, by the accounting system, the one or more first substrings to the numerical representation generation model;

generating, by the numerical representation generation model, a numerical representation of the candidate financial record based on the one or more first substrings;

training a first entity prediction model using training data comprising example financial records annotated with a respective identifier to generate a first trained entity prediction model:

providing, by the accounting system, the numerical representation of the candidate financial record as an input to the first trained entity prediction model;

determining, by the first trained entity prediction model, a first predicted entity identifier;

training a second entity prediction model using training data comprising example financial records annotated with a respective identifier to generate a second trained entity prediction model, wherein the second trained entity prediction model is a multi-class classifier and wherein the multi-class classifier is configured to classify a candidate financial record as being associated with one of a first set of entity identifiers;

providing, by the accounting system, the numerical representation of the candidate financial record as an input to the second trained entity prediction model;

determining, by the second trained entity prediction model, a second predicted entity identifier;

combining, by the accounting system, the first predicted entity identifier and the second predicted entity identifier to generate a combined predicted entity identifier by determining a longest common sequence between the first predicted identifier and the second predicted identifier;

determining, by the accounting system, a first set of suggested entity identifiers for the candidate financial record based on the combined predicted entity identifier; and

using, by the accounting system, a suggested entity identifier from the first set of suggested entity identifiers to: (i) reconcile the candidate financial record with a respective accounting record of the accounting system; or (ii) create a new accounting record in the accounting system;

wherein generating the numerical representation of the candidate financial record comprises:

generating, by the accounting system, a first set of tokens by tokenising each of the one or more first substrings;

generating, by the numerical representation generation model, a numerical representation of each token of the first set of tokens; and

determining the numerical representation of the candidate financial record as a function of the numerical representations of each token of the first set of tokens.

2 . The method of claim 1 , wherein the first set of suggested entity identifiers for the candidate financial record is or comprises the combined predicted entity identifier.

3 . The method of claim 1 , wherein the first set of entity identifiers is derived from: (i) a global contact list of the accounting system, (ii) a local contact list specific to a user of the accounting system; or (iii) the global contact list of the accounting system and the local contact list specific to a user of the accounting system.

4 . The method of claim 1 , wherein determining by the accounting system, a first set of suggested entity identifiers for the candidate financial record based on the combined predicted entity identifier comprises comparing the combined predicted entity identifier with a second set of entity identifiers.

5 . The method of claim 4 , wherein the second set of entity identifiers is derived from: (i) a global contact list of the accounting system, (ii) a local contact list specific to a user of the accounting system; or (iii) the global contact list of the accounting system and the local contact list specific to a user of the accounting system.

6 . The method of claim 4 , wherein comparing the combined predicted entity identifier with the second set of entity identifiers comprises using fuzzy matching techniques.

7 . The method of claim 1 , further comprising determining the suggested entity identifier from the first set of suggested entity identifiers, wherein determining the suggested entity identifier comprise providing to a user interface of the accounting system the first set of suggested entity identifiers and receiving via the user interface a user selected entity identifier of the first set of suggested entity identifiers, wherein the user selected entity identifier is the suggested entity identifier.

8 . The method of claim 1 , further comprising:

determining one or more second substrings from the character string of the financial record, wherein the one or more second substrings are different from the one or more first substrings; and

providing the one or more second substrings with the one or more first substrings to the numerical representation generation model to generate the numerical representation of the candidate financial record;

wherein generating the numerical representation of the candidate financial record further comprises:

generating, by the accounting system, a second set of tokens by tokenising each of the one or more second substrings;

generating, by the numerical representation generation model, a numerical representation of each token of the second set of tokens; and

determining the numerical representation of the candidate financial record as a function of the numerical representations of each token of the first and second sets of tokens.

9 . An accounting system comprising:

one or more processors; and

memory comprising computer executable instructions, which when executed by the one or more processors, causes the accounting system to:

determine a candidate financial record associated with a transaction between a first entity and a second entity;

determine one or more first substrings from a character string of the financial record;

train a numerical representation generation model using training data comprising a corpus generated from historical transaction records:

provide the one or more first substrings to the numerical representation generation model;

generate, by the numerical representation generation model, a numerical representation of the candidate financial record based on the one or more first substrings;

train a first entity prediction model using training data comprising example financial records annotated with a respective identifier to generate a first trained entity prediction model;

provide the numerical representation of the candidate financial record as an input to the first trained entity prediction model;

determine, by the first trained entity prediction model, a first predicted entity identifier;

train a second entity prediction model using training data comprising example financial records annotated with a respective identifier to generate a second trained entity prediction model, wherein the second entity prediction model is a multi-class classifier, and wherein the multi-class classifier is configured to classify a candidate financial record as being associated with one of a first set of entity identifiers;

provide the numerical representation of the candidate financial record as an input to the second trained entity prediction model;

determine, by the second trained entity prediction model, a second predicted entity identifier;

combine the first predicted entity identifier and the second predicted entity identifier to generate a combined predicted entity identifier by determining a longest common sequence between the first predicted identifier and the second predicted identifier;

determine a first set of suggested entity identifiers for the candidate financial record based on the combined predicted entity identifier; and

use a suggested entity identifier from the first set of suggested entity identifiers to: (i) reconcile the candidate financial record with a respective accounting record of an accounting system; or (ii) create a new accounting record in the accounting system;

wherein generating the numerical representation of the candidate financial record comprises:

generating a first set of tokens by tokenising each of the one or more first substrings;

generating, by the numerical representation generation model, a numerical representation of each token of the first set of tokens; and

determining the numerical representation of the candidate financial record as a function of the numerical representations of each token of the first set of tokens.

10 . The system of claim 9 , wherein the first set of entity identifiers is derived from: (i) a global contact list of the accounting system, (ii) a local contact list specific to a user of the accounting system; or (iii) the global contact list of the accounting system and the local contact list specific to a user of the accounting system.

11 . The system of claim 9 , wherein determining by the accounting system, a first set of suggested entity identifiers for the candidate financial record based on the combined predicted entity identifier comprises comparing the combined predicted entity identifier with a second set of entity identifiers.

12 . The system of claim 9 , wherein the second set of entity identifiers is derived from: (i) a global contact list of the accounting system, (ii) a local contact list specific to a user of the accounting system; or (iii) the global contact list of the accounting system and the local contact list specific to a user of the accounting system.

13 . A computer-readable storage medium storing instructions that, when executed by a computer, cause the computer to perform operations including:

determining a candidate financial record associated with a transaction between a first entity and a second entity;

determining one or more first substrings from a character string of the financial record;

training a numerical representation generation model using training data comprising a corpus generated from historical transaction records;

providing the one or more first substrings to a numerical representation generation model;

generating, by the numerical representation generation model, a numerical representation of the candidate financial record based on the one or more first substrings;

training a first entity prediction model using training data comprising example financial records annotated with a respective identifier to generate a first trained entity prediction model;

providing, the numerical representation of the candidate financial record as an input to the first trained entity prediction model;

determining, by the first trained entity prediction model, a first predicted entity identifier;

training a second entity prediction model using training data comprising example financial records annotated with a respective identifier to generate a second trained entity prediction model, wherein the second trained entity prediction model is a multi-class classifier and wherein the multi-class classifier is configured to classify a candidate financial record as being associated with one of a first set of entity identifiers;

providing the numerical representation of the candidate financial record as an input to the second trained entity prediction model;

determining, by the second trained entity prediction model, a second predicted entity identifier;

combining the first predicted entity identifier and the second predicted entity identifier to generate a combined predicted entity identifier by determining a longest common sequence between the first predicted identifier and the second predicted identifier;

determining a first set of suggested entity identifiers for the candidate financial record based on the combined predicted entity identifier; and

using a suggested entity identifier from the first set of suggested entity identifiers to: (i) reconcile the candidate financial record with a respective accounting record of an accounting system; or (ii) create a new accounting record in the accounting system;

wherein generating the numerical representation of the candidate financial record comprises:

generating a first set of tokens by tokenising each of the one or more first substrings;

generating, by the numerical representation generation model, a numerical representation of each token of the first set of tokens; and

determining the numerical representation of the candidate financial record as a function of the numerical representations of each token of the first set of tokens.

14 . The computer-readable storage medium of claim 13 , wherein the first set of entity identifiers is derived from: (i) a global contact list of the accounting system, (ii) a local contact list specific to a user of the accounting system; or (iii) the global contact list of the accounting system and the local contact list specific to a user of the accounting system.

15 . The computer-readable storage medium of claim 13 , wherein determining by the accounting system, a first set of suggested entity identifiers for the candidate financial record based on the combined predicted entity identifier comprises comparing the combined predicted entity identifier with a second set of entity identifiers.

16 . The computer-readable storage medium of claim 13 , wherein the second set of entity identifiers is derived from: (i) a global contact list of the accounting system, (ii) a local contact list specific to a user of the accounting system; or (iii) the global contact list of the accounting system and the local contact list specific to a user of the accounting system.

Assignments (1)

Priority Claims (1)

Continuity (5)

References Cited (85)