Back to results
US 12,664,582 B2Granted
Computing apparatus, method and program for data reconciliation
OwnerRodger McNab, Juan Carlos Rada-Vilela +1inventors
InventorsRodger McNab, Juan Carlos Rada-Vilela, Stanley Roache
Life of this patent
Estimating the term…
Abstract
A mechanism for reconciling unreconciled bank statement lines with corresponding accounting book entries in an online bookkeeping service. A machine-configurable similarity measurement model is applied to quantify similarity of an unreconciled bank statement line to each of a set of historical bank statement line reconciliations for a specific user, to identify a most similar line reconciliation, and upon satisfaction of a threshold, to automatically reconcile the unreconciled bank statement line with extracted accounting book entry details from the identified reconciliation.
Claims15
A computing apparatus comprising processor hardware, memory hardware, and network interface hardware, the memory hardware storing processing instructions which, when executed by the processor hardware, cause the computing apparatus to provide a reconciliation engine for reconciling unreconciled bank statement lines with corresponding accounting book entries in an online bookkeeping service of an accounting system having one or more users, the unreconciled bank statement lines being imported into the accounting system and representing a payment in which the user is a participant;
- the reconciliation engine comprising an historical reconciliation tool, being operable, for each of one or more of the unreconciled bank statement lines, to use a similarity measurement model to quantify similarity of the unreconciled bank statement line to each of a set of historical bank statement line reconciliations for a same user, to identify a most similar historical bank statement line reconciliation to the unreconciled bank statement line, to extract one or more accounting book entry details from the said most similar historical bank statement line reconciliation, and upon satisfaction of a threshold minimum similarity, to automatically or semi-automatically reconcile the unreconciled bank statement line with the extracted one or more accounting book entry details;
- the computing apparatus being operable to execute a machine learning algorithm to configure the similarity measurement model, the similarity measurement model comprising one or more input vector components, with respective configurable weights, and a configurable bias;
- wherein the machine learning algorithm is a logistic regression machine learning algorithm operable to configure the weights of each of the one or more input vector components;
- and
- wherein the one or more input vector components comprises two input vector components:
- a first input vector component being normalized Levenshtein distance between a first unreconciled text string, representing the unreconciled bank statement line, and a first reconciled text string representing the historical bank statement line reconciliation, the first unreconciled text string and the first reconciled text string each being composed of a concatenation of text strings being values of same bank statement line elements in a same order and from which words containing numeric characters are excluded,
- a second input vector component being normalized Levenshtein distance between a second unreconciled text string, representing the unreconciled bank statement line, and a second reconciled text string representing the historical bank statement line reconciliation, the second unreconciled text string and the second reconciled text string each being composed of a concatenation of text strings being values of the same bank statement line elements in the same order and from which words not containing numeric characters are excluded.
4 more elements
A computer-implemented method comprising providing a reconciliation engine for reconciling unreconciled bank statement lines with corresponding accounting book entries in an online bookkeeping service of an accounting system having one or more users, the unreconciled bank statement lines being imported into the accounting system and representing a payment in which the user is a participant;
- the method including an historical reconciliation process, for each of one or more of the unreconciled bank statement lines, comprising:
- using a similarity measurement model to quantify similarity of the unreconciled bank statement line to each of a set of historical bank statement line reconciliations for a same user, to identify a most similar historical bank statement line reconciliation to the unreconciled bank statement line, to extract one or more accounting book entry details from the said most similar historical bank statement line reconciliation, and upon satisfaction of a threshold minimum similarity, to automatically or semi-automatically reconcile the unreconciled bank statement line with the extracted one or more accounting book entry details;
- the method further comprising executing a machine learning algorithm to configure the similarity measurement model, the similarity measurement model comprising one or more input vector components, with respective configurable weights, and a configurable bias;
- wherein the machine learning algorithm is a logistic regression machine learning algorithm operable to configure the weights of each of the one or more input vector components;
- and
- wherein the one or more input vector components comprises two input vector components:
- a first input vector component being normalized Levenshtein distance between a first unreconciled text string, representing the unreconciled bank statement line, and a first reconciled text string representing the historical bank statement line reconciliation, the first unreconciled text string and the first reconciled text string each being composed of a concatenation of text strings being values of same bank statement line elements in a same order and from which words containing numeric characters are excluded,
- a second input vector component being normalized Levenshtein distance between a second unreconciled text string, representing the unreconciled bank statement line, and a second reconciled text string representing the historical bank statement line reconciliation, the second unreconciled text string and the second reconciled text string each being composed of a concatenation of text strings being values of the same bank statement line elements in the same order and from which words not containing numeric characters are excluded.
5 more elements
A computer program which, when executed by a computing apparatus having processor hardware and memory hardware, causes the computing apparatus to perform a method comprising providing a reconciliation engine for reconciling unreconciled bank statement lines with corresponding accounting book entries in an online bookkeeping service of an accounting system having one or more users, the unreconciled bank statement lines being imported into the accounting system and representing a payment in which the user is a participant;
- the method including an historical reconciliation process, for each of one or more of the unreconciled bank statement lines, comprising:
- using a similarity measurement model to quantify similarity of the unreconciled bank statement line to each of a set of historical bank statement line reconciliations for a same user, to identify a most similar historical bank statement line reconciliation to the unreconciled bank statement line, to extract one or more accounting book entry details from the said most similar historical bank statement line reconciliation, and upon satisfaction of a threshold minimum similarity, to automatically or semi-automatically reconcile the unreconciled bank statement line with the extracted one or more accounting book entry details;
- the method further comprising executing a machine learning algorithm to configure the similarity measurement model, the similarity measurement model comprising one or more input vector components, with respective configurable weights, and a configurable bias;
- wherein the machine learning algorithm is a logistic regression machine learning algorithm operable to configure the weights of each of the one or more input vector components;
- and
- wherein the one or more input vector components comprises two input vector components:
- a first input vector component being normalized Levenshtein distance between a first unreconciled text string, representing the unreconciled bank statement line, and a first reconciled text string representing the historical bank statement line reconciliation, the first unreconciled text string and the first reconciled text string each being composed of a concatenation of text strings being values of same bank statement line elements in a same order and from which words containing numeric characters are excluded,
- a second input vector component being normalized Levenshtein distance between a second unreconciled text string, representing the unreconciled bank statement line, and a second reconciled text string representing the historical bank statement line reconciliation, the second unreconciled text string and the second reconciled text string each being composed of a concatenation of text strings being values of the same bank statement line elements in the same order and from which words not containing numeric characters are excluded.
5 more elements
A computer program which, when executed by a computing apparatus comprising processor hardware and memory hardware, causes the computing apparatus to;
- provide a reconciliation engine for reconciling unreconciled bank statement lines with corresponding accounting book entries in an online bookkeeping service of an accounting system having one or more users, the unreconciled bank statement lines being imported into the accounting system and representing a payment in which the user is a participant;
- and
- perform an historical reconciliation process, for each of one or more of the unreconciled bank statement lines, comprising:
- using a similarity measurement model to quantify similarity of the unreconciled bank statement line to each of a set of historical bank statement line reconciliations for a same user, to identify a most similar historical bank statement line reconciliation to the unreconciled bank statement line, to extract one or more accounting book entry details from the said most similar historical bank statement line reconciliation, and upon satisfaction of a threshold minimum similarity, to automatically or semi-automatically reconcile the unreconciled bank statement line with the extracted one or more accounting book entry details;
- and
- executing a machine learning algorithm to configure the similarity measurement model, the similarity measurement model comprising one or more input vector components, with respective configurable weights, and a configurable bias;
- wherein the machine learning algorithm is a logistic regression machine learning algorithm operable to configure the weights of each of the one or more input vector components;
- and
- wherein the one or more input vector components comprises two input vector components:
- a first input vector component being normalized Levenshtein distance between a first unreconciled text string, representing the unreconciled bank statement line, and a first reconciled text string representing the historical bank statement line reconciliation, the first unreconciled text string and the first reconciled text string each being composed of a concatenation of text strings being values of same bank statement line elements in a same order and from which words containing numeric characters are excluded,
- a second input vector component being normalized Levenshtein distance between a second unreconciled text string, representing the unreconciled bank statement line, and a second reconciled text string representing the historical bank statement line reconciliation, the second unreconciled text string and the second reconciled text string each being composed of a concatenation of text strings being values of the same bank statement line elements in the same order and from which words not containing numeric characters are excluded.
8 more elements
Dependencies are inferred from the claim text.