WO2021119763A1

WO2021119763A1 - Schedule generation system and method

Info

Publication number: WO2021119763A1
Application number: PCT/AU2020/051405
Authority: WO
Inventors: Gabi Youssef
Original assignee: Your 1Hub Pty Ltd
Priority date: 2019-12-20
Filing date: 2020-12-18
Publication date: 2021-06-24
Also published as: AU2020406041A1

Abstract

The present invention relates generally to a system and method for generating a financial schedule for a user. The invention involves accessing data records relating to income and expenditure incurred by a user, using one or more character recognition techniques to identify a plurality of values embedded in the data records and indexing and storing same in a data structure to facilitate extraction of information for the purpose of generating a financial schedule include the application of deductions formed as a result of analysing the extracted information.

Description

SCHEDULE GENERATION SYSTEM AND METHOD

FIELD OF THE INVENTION

[0001] The present invention relates to a system and method for the generation of a financial schedule for a user. The invention is particularly useful for users who are confronted with monitoring and managing many documents pertaining to their finances and facilitates extraction of relevant information to generate income, expenditure and cash flow schedules according to deductions based upon extracted information.

BACKGROUND OF THE INVENTION

[0002] Maintaining a record of income and expenditure is a task that confronts most people who incur expenses and liabilities, and must satisfy those expenses and liabilities with the income they receive. There have been various book keeping solutions proposed that seek to enable users to keep track of their expenses with a view to ensuring that users do not incur expenses and liabilities that they cannot timely satisfy.

[0003] The introduction of unsecured credit facilities for average consumers has greatly complicated this task. In this regard, it is possible to obtain easy credit from banking and other financial institutions to enable users to afford purchases without available funds. However, it is often the case that users incur an ongoing debt arising from availing themselves of credit facilities, requiring the user to service interest payments in respect of the debt until such time that their debt is fully satisfied.

[0004] To assist users to manage their income and expenses, many software programs have been developed that require a user to manually enter details regarding their income and expenses in a format recognised by the software program. Subsequent to entering all of the relevant details regarding income and expenditure, known software programs provide detailed reports to enable the user to monitor their income and expenditure habits and manage their expenditure to ensure that they are able to timely settle invoices and/or liabilities as and when they are due.

[0005] However, previous software programs that assist users to manage their income and expenditure predominantly require invoices and income documents in a hard copy or printed format. Previously, users would generally receive an invoice/receipt or a document providing details regarding any income they received in the form of a printed document which the user could collect and, at a convenient time, enter the details of the income or expenditure detailed in the documents into the software program. The software program would then analyse the income and expenditure, and provide the user with detailed reports according to the facilities provided by the particular software program. With the advent of many organisations issuing invoices in electronic form (e.g. a document in the Portable Document Format [PDF] or a proprietary format used by the organisation) users are now receiving many of their documents in an electronic form and generally either as an attachment to an electronic mail message or a document that the user is invited to download (e.g. upon accessing a web-site to obtain an electronic copy of the relevant document). Flowever, many transactions continue to involve the generation of paper documents to evidence the transaction by providing the customer with a printed receipt.

[0006] Accordingly, users now commonly find the task of collecting various documents regarding income and expenditure to be difficult and tedious since the documents are now provided to the user in a range of formats including printed receipts and various electronic formats through various channels.

[0007] Of course, many organisations now generate documents including invoices and receipts in an electronic form in an attempt to reduce the usage of natural resources and provide those documents in a convenient form (such as attachments to electronic mail messages) so that users may receive those documents in a timely and secure manner and are not restricted to accessing those documents from a post box or letter box. Flowever, such documents are emailed to a user, the email can easily be deleted or become difficult to locate amongst hundreds or thousands of other emails.

[0008] As a result of recent developments, users are now confronted with a more difficult task regarding the monitoring of their income and expenditure and accordingly, there is a need for a system that enables the gathering of electronic documents relating to income and expenditure and the storage and extraction of same in a manner that facilitates the analysis and provision of results of such analysis. Of course, as the task of monitoring income and expenditure becomes increasingly difficult, users are increasingly unaware of the potential to have insufficient funds to satisfy a debt when it falls due.

SUMMARY OF THE INVENTION

[0009] In one aspect, the present invention provides a computer-implemented method of generating a schedule, the method including: accessing, by one or more processors, one or more data records pertaining to a particular user; processing, by the one or more processors, the data records to generate data from the data records including identifying, by the one or more processors, using one or more character recognition techniques, a plurality of values embedded in the data records; utilizing, by the one or more processors, a first machine learning model to allocate each of the plurality of values to one of a plurality of categories, the first machine learning model receiving, as input, the data and the identified plurality of values, and the first machine learning model outputting information that identifies the category or categories to which each value has been allocated, automatically generating, by the one or more processors, a data structure associated with the particular user; storing, by the one or more processors, the data in the data structure thereby providing an index of the data according to the category, or categories, to which each value has been allocated; and extracting, by the one or more processors, based upon the categorisation of the values, information from the data structure to generate a schedule for the particular user.

[0010] In an embodiment, the one or more data records includes information pertaining to income and expenditure of the particular user.

[0011 ] In an embodiment, the plurality of values embedded in the data identified using the one or more character recognition techniques include words, numbers and/or symbols relevant to income or expenditure of the particular user. Examples of values that are likely to be identified as relevant to income include words such as “credit” and “earnings” and symbols such as “+”, and values that are likely to be identified as relevant to expenditure include words such as “debit” and “receipt” and symbols such as the character. [0012] In an embodiment, the plurality of categories includes one category identifying values that relate to an income of the particular user, and another category identifying values that relate to an expenditure of the particular user.

[0013] In an embodiment, the plurality of categories further include a category identifying values that are likely to be recurring values. In this way, recurring values that represent income and expenditure likely to be repeated in the future, e.g. wages and household bills, may be identified and distinguished from income and expenditure that is unlikely to be repeated in the future, e.g. a wage bonus, or an irregular shopping expense (e.g. clothing, gifts, etc.).

[0014] In an embodiment, the plurality of categories further include categories relating to the frequency of recurrence in respect of recurring values. For example, values that are likely to recur on a weekly basis may be categorised separately from values that are likely to recur monthly.

[0015] It will be appreciated that values can thereby have multiple categories, e.g. a value may be categorised as income, as recurring, and as having a recurrence frequency of 1 month.

[0016] In an embodiment, the method further includes: receiving, by the one or more processors, input from the particular user correcting the allocation of any one value to an incorrect category. For example, if the schedule shows a value that is income but has been incorrectly classified as expenditure, the user is provided an opportunity to correct the incorrect classification.

[0017] In an embodiment, the method further includes: utilizing the data input into the first machine learning model, the information output from the first machine learning model, and the user input correcting the allocation of any one value to an incorrect category, to train the first machine learning model.

[0018] In an embodiment, generating the schedule for the particular user includes: utilizing a second machine learning model, the second model receiving, as input, the data extracted from the data structure based upon the categorization of the values, and the second model outputting predictions relating to income and expenditure of the particular user in a future time period for inclusion in the schedule.

[0019] In an embodiment, the predictions relating to income and expenditure included in the schedule are based solely upon values that are categorized as recurring.

[0020] In an embodiment, the predictions relating to income and expenditure included in the schedule are also based upon values that are categorized as non-recurring, i.e. unlikely to be repeated.

[0021] In an embodiment, when predicting income and expenditure in the future time period based upon both recurring and non-recurring values, more weight is given to recurring values than non-recurring values. In this way, the schedule is generated based upon values reflecting regular income and expenditure, and to a lesser extent, values identified as irregular income and expenditure.

[0022] In an embodiment, the method further includes, during or subsequent to said future time period: receiving, by the one or more processors, input from the particular user correcting a prediction regarding income or expenditure in the schedule based upon an actual income generated or expenditure incurred. For example, if the schedule shows an income or expenditure that is inaccurate as compared with the actual amount of income generated or expenditure incurred, the user is provided an opportunity to correct the income or expenditure.

[0023] In an embodiment, the method further includes: utilizing the data input into the second machine learning model, the predictions output from the first machine learning model, and the user input correcting an inaccurate income or expenditure, to train the second machine learning model.

[0024] In an embodiment, the schedule further indicates a cash flow position of the particular user in the future time period based upon the predicted income and expenditure, the cash flow position calculated based upon the total income and total expenditure for the future time period. For example, the schedule may indicate a negative cash flow position if the total expenditure exceeds the total income, or may indicate a positive cash flow position if the total income exceeds the total expenditure.

[0025] In an embodiment, the method further includes: generating, by the one or more processors, one or more recommendations for the user to improve their cash flow position in the future time period.

[0026] In an embodiment, the recommendation includes, in the case of a negative cash flow position in the future time period, recommending that the particular user save a dollar amount that is based upon (e.g. equivalent to) the negative cash flow position.

[0027] In an embodiment, users may provide permission to entities generating invoices and/or receipts for users to direct those documents directly to a storage and processing system that receives and analyses income and expenditure documents pertaining to the user. By providing permission to direct documents to a central processing facility, users may avoid significant difficulties associated with personally receiving a large number of documents in a range of formats thereby reducing the likelihood that a user misplaces one or more documents thereby requiring those documents to be re-transmitted to the user. Of course, this benefit significantly reduces frustration for the user and also reduces the inefficient consumption of computer and network resources associated with repeated requests for copies of misplaced or mistakenly deleted documents. Further, permission may be provided to the central processing facility to release data and information to an online Accounting software solution to assist users to populate the necessary fields of an Accounting database for Accounting purposes such as the preparation of taxation returns.

[0028] In an embodiment, the plurality of categories further include a category identifying values that characterise a type of income or expenditure. For example, the word “electricity” may automatically be categorised as “expenditure” and further categorised as a value that characterises a type of expenditure.

[0029] In an embodiment, two or more values may be linked according to whether they are related, e.g. related to income or related to expenditure. [0030] In an embodiment, two or more values are linked if they are identified within the same document, or within a particular vicinity or region of the same document. For example, the value “electricity” identified in an electricity bill may be linked to a dollar amount (number) present in the same document, or where a document includes a summary of multiple bills, the word “electricity” identified in the summary may be linked to an adjacent dollar amount (number), or a dollar amount (number) in the same row, which is likely to identify the amount relating to electricity usage.

[0031] In an embodiment, the method further includes: generating, by the one or more processors, a survey for completion by the particular user, the survey including queries based upon the values that characterise a type of income or expenditure; transmitting, by the one or more processors, the survey to a device associated with the particular user; and receiving, by the one or more processors, the user’s response to the survey.

[0032] In an embodiment, the second model is further trained based upon the survey results. For example, a survey question may ask a user what type of television they are using, or to specify how many hours they operate a television for on average, on the basis that the word “electricity” was identified as a value that characterises a type of expenditure incurred by the particular user. If, based upon multiple survey results, a particular type of television or a particular amount of viewing is identified as consuming electricity beyond a particular threshold, then any new predictions generated by the second model will take into account this additional information. For example, if a new query is executed by the second model in respect of a new user who uses the same television or operates their television for a similar amount of time on average, the predictions generated by the second model relating to expenditure may be adjusted accordingly.

[0033] More importantly, according to embodiments of the invention, users are provided with a schedule that enables users to understand the patterns of their income and expenditure and more particularly, are able to determine those forthcoming periods during which they may suffer a negative cash flow position. During these periods, users will need to either draw upon cash reserves (such as a bank or savings account) or alternatively, rely upon credit. [0034] In an embodiment, the method further includes: generating, by the one or more processors, one or more recommendations based upon the survey results. For example, a recommendation may be for a user to reduce the amount of time they watch television or to consider purchasing another television that consumes less electricity.

[0035] In an embodiment, the schedule provides analytical information and includes one or more of: a spreadsheet, a calendar, and a graph.

[0036] As will be appreciated by skilled readers, the generation of a data structure that provides an indexation of the data results in a particularly efficient storage of the data such that analysis and retrieval of data stored in the structure significantly reduces the computing resource that would otherwise be required to extract and sort / analyse data subsequent to storage.

[0037] In a second aspect, the present invention provides a system for generating a schedule, the system including: one or more memories, and one or more processors operable to: access one or more data records pertaining to a particular user; process the data records to generate data from the data records including identifying, using one or more character recognition techniques, a plurality of values embedded in the data records; utilize a first machine learning model to allocate each of the plurality of values to one of a plurality of categories, the first machine learning model receiving, as input, the data and the identified plurality of values, and the first machine learning model outputting information that identifies the category or categories to which each value has been allocated; automatically generate a data structure associated with the particular user; store the data in the data structure thereby providing an index of the data according to the category, or categories, to which each value has been allocated; and extract, based upon the categorisation of the values, information from the data structure to generate a schedule for the particular use.

[0038] In a third aspect, the present invention provides a computer-readable medium having a plurality of instructions executable by one or more processors to: access one or more data records pertaining to a particular user; process the data records to generate data from the data records including identifying, using one or more character recognition techniques, a plurality of values embedded in the data records; utilize a first machine learning model to allocate each of the plurality of values to one of a plurality of categories, the first machine learning model receiving, as input, the data and the identified plurality of values, and the first machine learning model outputting information that identifies the category or categories to which each value has been allocated; automatically generate a data structure associated with the particular user; store the data in the data structure thereby providing an index of the data according to the category, or categories, to which each value has been allocated; and extract, based upon the categorisation of the values, information from the data structure to generate a schedule for the particular use.

BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 illustrates a schematic overview of an embodiment of the invention that provides a system for generating a schedule for a particular user.

[0040] Figure 2 illustrates a central server of the system of Figure 1 .

[0041] Figure 3 illustrates an exemplary flow diagram of a process that enables a user downloading and installing a software application that enables the user to subsequently log in, or register, to use the software application for interacting with the system of Figure 1.

[0042] Figure 4 illustrates an exemplary flow diagram of a process that enables bill handling, allowing a user to enter bills, schedule payment of bills and receive notifications regarding new bills.

[0043] Figure 5 illustrates an exemplary flow diagram of a process that enables transfer of funds for payment of bills from a bank or credit facility.

[0044] Figure 6 illustrates an exemplary flow diagram of a process that enables redirection of bills from billing entities direct to the system, allowing automation of bill processing.

[0045] Figure 7 illustrates an exemplary flow diagram of a process that enables analytical information such as a schedule to be provided to the user regarding their bills and financial position, as well as access to archived information.

DETAILED DESCRIPTION OF EMBODIMENT(S) OF THE INVENTION

[0046] For simplicity and illustrative purposes, the present disclosure is described by referring to an embodiment thereof. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present disclosure. It will be readily apparent however, that the present disclosure may be practiced without limitation to these specific details. In other instances, some methods and structures have not been described in detail to avoid unnecessarily obscuring the present disclosure.

[0047] As used herein, the terms “a” and “an” are intended to denote at least one of a particular element, the term “includes” means includes but not limited to, the term “including” means including but not limited to, and the term “based on” means based at least in part on.

[0048] The present invention provides a computer-implemented system and method for generating a financial schedule for a user (100), utilising a central server (20) in communication with one or more user devices (110). The central server (20) maintains one or more processors and/or databases for performing functions including accessing one or more data records (150, 230) pertaining to the particular user (100), processing the data records to generate data from the data records, identifying (using one or more character recognition techniques) a plurality of values embedded in the data, and allocating each of the plurality of values to one of a plurality of categories. The allocation of values to categories is achieved using an analytic engine (80) utilizing a first machine learning model that receives, as input, the data and information relating to the identified plurality of values, and outputting information that identifies the category or categories to which each value has been allocated. The server (20) then automatically generates a data structure (40) associated with the particular user (100) and stores the data in the data structure to provide an index of the data according to the category or categories to which each value has been allocated. In order to generate a schedule (180) that provides predictions with respect to a future time period, the server (20) extracts data from the data structure based upon the particular categorization of values by the first machine learning model.

[0049] In a particular embodiment, the schedule that is generated is a schedule (180) relating to income and expenditure of a user (100), wherein the plurality of values embedded in the data include words, numbers and/or symbols that are likely to be relevant to income or expenditure of the particular user, and where for example the plurality of categories includes one category identifying values that relate to an income of the particular user, and another category identifying values that relate to an expenditure of the particular user. Whilst the generation of a schedule that assists user (100) in managing income, expenditure and cashflow, this represents one example.

[0050] Figure 1 provides a schematic overview of a particular embodiment of the invention that is divided into segments 200 to 700. The core of the system is segment 200 which provides a central server (20) supporting functions such as user accounts, bill storage and payment scheduling, managing funds and bill payments, notifications to users and analytics. The server (20) may interact with a user (100) via a web or smart device software application (120), as well as financial institutions (52, 54) and billing entities (240). Segment 300 allows a user to download and install the user application (120), create an account with the system and login to access the functionality of the system. Segment 400 provides bill handling functionality, allowing a user to enter bills, scheduled payments of bills and receive notifications of new bills. Segment 500 handles transferring funds for payment of bills from a bank or credit facility. Segment 600 provides for redirection of bills from suppliers of goods or services direct to the system allowing automation of bill processing. Segment 700 provides a schedule (180) and analytical information to the user (100) regarding their bills and financial position, as well as access to archived bills. Segments 200 to 700 are described in greater detail below with reference to Figures 2 to 7.

[0051] Figure 2 shows in greater detail the components of segment 200 which includes the server component (20) of the system and its interfaces with the user (100), financial institutions (52), (54), billing entities (240) and cloud storage providers (90). The server (20) is implemented on infrastructure (10) and may take a variety of forms. The infrastructure may be a local or cloud-based processor running one or more computer applications on one or more platforms. The server (20) provides multiple functions to support the system including user account function (30), data structure (40) for indexing income and expenditure according to the allocated categories, bill scheduling function (45), funds management function (50), bill payment function (60), bill notification function (70) and analysis module (80). These functions will be described below, and then in greater detail further below in the context of Figures 3 to 6. [0052] The user account function (30) stores user account information and thereby allows users to register with the system and to subsequently login to use the system.

[0053] The data structure (40) indexes details of a users’ income and expenditure, including for example payslips, bills and invoices, to facilitate access to such information for the purpose of extracting and analysing same using the analytic engine (80) which operates a second machine learning model to make predictions as described in greater detail below.

[0054] The bill scheduling function (45) maintains a schedule of when a user’s bills are to be paid, including as a result of manual scheduling by the user (100) and by automatic scheduling wherein predictions regarding income and expenditure in a future time period as determined by the analytic engine (80) are incorporated into the schedule (180).

[0055] The funds management function (50) manages funds to pay bills when due, obtaining the funds required from the user’s bank account (52), a credit facility (54), or a combination thereof. The credit facility may be used when the user (100) does not have sufficient funds available to pay their bills or to smooth bill payments over a time period such as a year. Users (100) will typically be charged interest on any credit used to pay bills.

[0056] The bill payment function (60) handles the payment of bills to a supplier of goods or services, either manually or in accordance with a schedule (180) managed by the bill scheduling function (45), using funds from the funds management function (50).

[0057] The bill notification function (70) issues notifications to users. Typically this will advise the user (100) of a new bill requiring payment but may also be a summary or an alert that the user needs more funds to pay their bills.

[0058] An analytic engine (80) for creating an income, expenditure and cashflow schedule (180) for the user (100) and provide any relevant analytical information including a cashflow position of the user (100). This functionality includes utilizing a first machine learning model to categorise income and expenditure, and utilizing a second machine learning model to predict future income and expenses. This analysis assists users by allowing them to smooth bill payments and understand future payment requirements. [0059] Figure 2 also shows that the server (20) is configured to enable communication (140) with user devices (110) and in particular with software application (120) operating on the user devices (110).

[0060] Figure 3 shows in greater detail segment 300 of Figure 1 which supports a user (100) downloading and installing the user application (130), creating an account with the system and logging in.

[0061] A user (100) may access the system via a user application (120) on a variety of devices (110). For access using a smart phone or tablet a dedicated application (120) may be used which a user first downloads and installs (130) on their device (110). A web based user application (120) may be accessed via a standard web browser for laptop or PC use.

[0062] Before using the user application (120), a user will be requested to register (140) their details with the user account function (30). Users may need to provide details such as name, address, date of birth, email address, phone number and mobile number. A one off, or recurring fee, may be charged for the user (100) to access the system. Once registered the user (100) can login to access the various functions of the system.

[0063] Figure 4 shows segment 400 of Figure 1 in which a user (100) enters a bill or billers into the system, schedules the payment of bills, and is notified of new bills.

[0064] A user (100) can enter the details of a bill (150) that they have received from a supplier of goods or services with the aid of a document submission module (160) of the user application (120). This may be achieved by capturing an image (165) of the bill, scanning bill codes (155) (e.g. barcode or QR codes) or manually entering details. The document submission module allows the user to enter further information regarding the bill, for example the user can specify whether this a ‘one-off bill for payment, or if they would like the system to register with the biller so that future bills are sent directly to the system for automated processing. Details of bills handled directly by the system are discussed further below with reference to Figure 6.

[0065] The user (100) may direct all of their documents pertaining to income and expenditure to the document submission module (160) which is operable to receive and recognise the formats of all of the documents the user directs to the document submission module (160). Alternatively, relevant documents relating to income or expenditure may be stored externally and therefore need to be retrieved by the software application (120). It is preferable that a user directs all of their documents to the document submission module (160), and that all relevant documents stored externally be retrieved, to improve the analysis conducted by the system and present to the user the schedule (180) and most accurate representation of the user’s cash flow position with respect to income and expenditure including likely income and expenditure into the future.

[0066] The analytic engine (80) utilises a first machine learning model to allocate values recognised in each document (utilising a character recognition technique) to one of a plurality of categories. The values that are detected by the character recognition technique may include words, numbers and/or symbols that are likely to be relevant to income or expenditure of the particular user. Examples of values that are likely to be identified in a document as relevant to income include words such as “credit” and “earnings” and symbols such as “+”, and values that are likely to be identified as relevant to expenditure include words such as “debit” and “receipt” and symbols such as

[0067] The first machine learning model receives, as input, all data and information relating to the identified plurality of values, and outputs information that identifies the category or categories to which each value has been allocated. For example, the character recognition technique (e.g. an optical character recognition (OCR) facility) extracts sufficient detail from the data records to identify differences in values, and thereby differences between documents (or parts thereof) pertaining to income as compared with documents pertaining to expenditure. In this way, the character recognition facility may extract sufficient information to enable the first machine learning model to categorise values embedded in the documents, and thereby the documents themselves, as income or expenditure.

[0068] The plurality of categories may further include a category identifying values that are likely to be recurring values. In this way, recurring values that represent income and expenditure likely to be repeated in the future, e.g. wages and household bills, can be identified and distinguished from income and expenditure that is unlikely to be repeated in the future, e.g. a wage bonus, or an irregular shopping expense. The plurality of categories may further include categories relating to the frequency of recurrence in respect of recurring values. For example, values that are likely to recur on a weekly basis may be categorised separately from values that are likely to recur monthly. It will be appreciated that values can thereby have multiple categories, e.g. a value may be categorised as income, as recurring, and as having a recurrence frequency of 1 month.

[0069] The plurality of categories may further include a category identifying values that characterise a type of income or expenditure. For example, the word “electricity” may automatically be categorised as “expenditure” and further categorised as a value that characterises a type of expenditure. Further, two or more values may be linked according to whether they are related, e.g. related to income or related to expenditure. Two or more values may be linked if they are identified within the same document, or within a particular vicinity or area of the same document. For example, the value “electricity” identified in an electricity bill may be linked to a dollar amount (number) present in the same document, or where a document includes a summary of multiple bills, the word “electricity” identified in the summary may be linked to an adjacent dollar amount (number), or a dollar amount (number) in the same row, which is likely to identify the amount relating to electricity usage.

[0070] It will also be appreciated that the first machine learning model can be trained, and the accuracy of output will increase over time. The user (100) may be prompted to correct any instances of values being incorrectly allocated. For example, if the schedule (180) that is eventually generated shows a value that is income but has been incorrectly classified as expenditure, the user (100) may be provided an opportunity to correct the incorrect classification, and such corrections may be used to train the first machine learning model.

[0071] The categorised values are stored in the data structure (40) which indexes the values accordingly, and thereby facilitates subsequent extraction and analysis thereof.

[0072] If a bill is received directly by the system the user, depending on preferences set, may be notified (170) of the bill by the bill notification function (70) via a push notification, text message, or an email summary of new bills and expenditure.

[0073] After values have been extracted from relevant income and expenditure records, categorized using the analytic engine (80) operating the first machine learning model, and indexed in the data structure (40), the analytic engine (80) may generate a schedule (180) for the user. The schedule (180) includes predictions relating to future income and expenditure, the predictions generated by the analytic engine (80) using a second machine learning model that receives, as input, the data extracted from the data structure (40) based upon the categorization of the values, and outputs predictions relating to income and expenditure of the particular user in a future time period for inclusion in the schedule (180). By providing such predictions, the schedule (180) can automatically schedule the payment of bills on the user’s behalf by the bill payment function (60). The user can also manually schedule payments . Details of the bills may be stored by the bill storage function (40), and the income and expenditure schedule (180) may be stored by the bill scheduling function (45).

[0074] The predictions relating to income and expenditure included in the schedule (180) may be based solely upon values that are categorized as recurring, or may be based upon both recurring and non-recurring values. In respect of the latter, more weight may be given to recurring values than non-recurring values. In this way, the schedule (180) may be generated based upon values reflecting regular income and expenditure, and to a lesser extent, values identified as reflecting irregular income and expenditure. The user (100) may input a correction to an inaccurate prediction regarding income or expenditure in the schedule based upon an actual income generated or expenditure incurred. For example, if the schedule shows an income or expenditure that is inaccurate compared with the actual amount of income generated or expenditure incurred, the user may be prompted to correct the income or expenditure, and such inputs may be used to train the second machine learning model.

[0075] The application (120) may be configured to generate a survey for presentation to the user (100), the survey including queries based upon the values that characterise a type of income or expenditure. Once the survey is completed, the user’s response to the survey may be used to further train the second machine learning model. For example, a survey question may ask a user what type of television they are using, or to specify how many hours they operate a television for on average on the basis that the word “electricity” was identified as a value that characterises a type of expenditure incurred by the particular user. If, based upon multiple survey results, a particular type of television or a particular amount of viewing is identified as consuming electricity beyond a particular threshold, then any new predictions generated by the second model may factor in same. For example, if a new query is run by the second model in respect of a new user who uses the same television or operates their television for a similar amount of time on average, the predictions generated by the second model relating to expenditure may be generated or adjusted accordingly. [0076] The schedule (180) may further indicate a cash flow position of the particular user (100) in the future time period based upon the predicted income and expenditure, the cash flow position calculated based upon the total income and total expenditure for the future time period. For example, the schedule may indicate a negative cash flow position if the total expenditure exceeds the total income, or may indicate a positive cash flow position if the total income exceeds the total expenditure.

[0077] Recommendations may also be generated by the system and presented to the user (100), and may include recommendations to improve the user’s cash flow position in the future time period. For example, the recommendation may include, in the case of a negative cash flow position in the future time period, a recommendation that the particular user (100) save a dollar amount that is based upon (e.g. equivalent to) the negative cash flow position. The recommendations may also be based upon the survey results. For example, a recommendation may be for a user to reduce the amount of time they watch television or to consider purchasing another television that consumes less electricity. Similarly, irrespective of a cash flow position (i.e. positive or negative), the system may recognise opportunities to improve cash flow by reducing expenditure without sacrificing any goods or services regularly purchased. For example, a user may be responsible for electricity charges at more than a single property and the system may recommend using the same service provider for electricity supply to reduce overall charges . For instances where a group of users (e.g. two or more) have a common interest in respect of income and/or expenditure, the system may group the users according to the common interest and provide schedules to the grouped users. In this regard, the grouped users may each receive notifications and schedules (including income, expenditure and cash flow schedules) for consideration.

[0078] The schedule (180) and any associated analytical information may be presented to the user via a graphical user interface and in any suitable format that assists the user to interpret the information presented, e.g. in a spreadsheet, calendar and/or graph. In the case of cash flow information, the graphical user interface may present information to enable the user to understand the effect on their cash flow when scheduling bills for payment in accordance with the schedule. Since changes and/or adjustments to the user’s income or expenditure are reflected in the data structure (40) and hence the schedule (180), the schedule (180) provides the user (100) with a real-time reflection of their financial position including updates to the cash flow position such that the user can appreciate and understand the impact of the schedule upon their cash flow.

[0079] In instances where an unexpected or irregular invoice is received by the system which does not fall within the scope of the schedule (180), the user (100) may be invited to confirm that the system should attend to timely settlement of the invoice and in addition to inviting the user (100) to authorise the system to attend to settlement of the invoice, the system may present to the user (100) the cash flow effect of attending to payment of the irregular invoice on or before the due date for settlement of same.

[0080] In the event the user (100) avails themselves of the credit facility of the system, the system may automatically update the schedule to settle the interest and/or other charges associated with the user availing themselves of short term credit from the credit facility operated by the on-line system. Once again, the on-line system may establish or update the schedule (180) for the credit facility and display to the user the effect of attending to payment for the credit charges with a representation, e.g. graphically, representing the effect upon the user’s cash flow.

[0081] Figure 5 details the components of segment 500 of Figure 1 which handles transferring funds by the funds management function (50) for payment of bills from a user bank account (190) or credit facility (54).

[0082] Users (100) may either allow the system to make direct debits (210) from their nominated bank account or accounts (190), either when a bill is due or as regular payments to the system to help to even out cash flow for the user. This ensures that the system has the funds available to make bill payments when scheduled on behalf of the user. A user’s wages (230) is typically deposited into their bank account (52) to supply funds for bill payments according to the schedule (180).

[0083] Alternatively, if approved users may access a credit facility (54) to draw credit (220) for the payment of bills when they don’t have available funds. This credit facility will allow users to smooth their bill payments and avoid overdue bills. The credit facility may be provided by the system or a financial institution. Users may be charged interest on credit which would be billed through the system.

[0084] The user may link one or more bank accounts to the on-line system and automate payment of bills according to a pre-defined and stored bill payment schedule in the on- line system. In this regard, once a user has established a schedule (180) with which they are satisfied does not negatively impact their cash flow and represents an affordable bill payment schedule, the user (100) may confirm payments from one or more bank accounts to accommodate the schedule (180) that has been entered and stored in the system.

[0085] In addition to a user (100) directing their invoices to the document submission module of the on-line system, they may also cause the system to effect payments to the supplier of the invoice at a time on or before the scheduled due date for settlement of the invoice. In this regard, the user (100) may direct their income to an account controlled by the system (or at least a bank account in respect of which the on-line system has authorised access) such that the system may make use of available funds from the user in the form of income and use those funds to attend to addressing expenditure in the schedule.

[0086] In addition to regularly scheduled and predicted income and expenditure items, the system may also allow users to provide “one off” cash deposits such as those received in the form of a gift, bonus payment and/or other irregular receipts such as a taxation return and also to allocate the application to received “one off cash receipts to outstanding bills including outstanding bills in relation to the fees and/or charges associated with the user availing themselves of the credit facility operated by the system.

[0087] In the event a user (100) experiences an un-scheduled expenditure or an unusual event that reduces their income, the system may provide details to the user confirming the extent to which they will suffer a negative cash flow position and the period of time for which their expenditure and ability to settle debt exceeds their available funds. In this instance, the system may use the credit facility such that funds may be drawn to accommodate the schedule (180), i.e. accommodating any shortfall in funds to satisfy the expenditure in the schedule with funds drawn from the external credit facility. The funds may be pre-activated by the user (100) and the system to be available for those periods during which the user may suffer a negative cash flow and be in a position during which they temporarily cannot maintain their schedule (100).

[0088] Figure 6 shows the components of segment 600 of Figure 1 detailing redirection of bills from suppliers of goods or services direct to the system allowing automation of bill processing and payment of bills on behalf of a user. Bills are typically for goods and services supplied (248) to the user by a supplier (240), for example utilities (242), rent (244), and car and retail expenditure (246).

[0089] Once configured by the user (100), the system may initiate communication (260) with a biller (240) on behalf of the user requesting that the supplier’s billing system (250) issue bills (270) for the user directly to the system. Billers can then electronically submit bills to the system, saving on bill delivery costs. Users (100) may also be eligible for discounts due to the ‘buying power’ and payment guarantees that are likely to be available as a result of implementation of the present invention.

[0090] Once a bill is received by the system, details of the bill may be stored by the bill storage function (40) with the payment due date stored by the bill scheduling function (45), and the user may be notified of the bill by the bill notification function (70).

[0091] The bill payment function (60) may make bill payments (280) directly on behalf of the user when the bill is scheduled to be paid, with funds sourced by the funds management function from either the user’s bank account, from a credit facility or a combination of both. The biller’s financial institution (290) may receive the payment and notify (295) the biller that payment has been received.

[0092] Figure 7 provides greater detail regarding the components of segment 700 of Figure 1 which provides analytical information (320) to the user (100) regarding their bills and financial position, as well as access to archived bills.

[0093] In this regard, in addition to providing a schedule (180), the analysis module (80) may provide analytical expenditure information. Such information may include details relating to value categories, the predicted future income and expenses, budgeting information and recommendations. This analysis may assist users by allowing them to smooth bill payments and understand future payments.

[0094] Bills entered into the system by the user (100) or received directly by the system may optionally be copied to a cloud storage solution (90) such as OneDrive, Google Drive or Dropbox. User (100) may subsequently retrieve the bills for review.

[0095] The analysis module (80) may conduct an analysis of the user’s expenditure and compare same with the user’s income. In addition to predicting future income and expenses, the analysis module (80) may generate a representation for display to a user detailing the income and current and future expenses to depict the user’s ability to afford the future likely expenses as and when they occur. In this regard, the prediction of user expenses may include predicting both the timing and amount of those future expenses.

[0096] An example of predicting a future expense both in respect of timing and amount includes analysing utility bills and comparing same with previous utility bills over an extended historical period. The analysis module (80) and in particular the second machine learning model may calculate the likely future expense according to the time of the year and may apply additional parameters when conducting such an analysis to take into account seasonal factors and changed circumstances (such as more or less people occupying the user’s dwelling). The analysis module (80) may also predicts the user’s future income both in relation to timing and amount which is particularly useful for users whose income varies from month to month. Again, the predicted capability of the analysis module may predict the user’s income taking into account a range of factors including past income for similar historical periods of time.

[0097] The analysis module (80) may also analyse the cost associated with a user availing themselves of credit as compared with the penalty payments that apply as a result of failure to settle debts as and when they are due and may determine the best allocation of funds to provide the user with the least cost option with respect to settling debts.

[0098] In the instance of discounts applied to invoices in the event of early settlement, the analysis module (80) may also consider early payment of invoices and the cost impact of attending to early payment as compared with the user’s other outstanding commitments and determine the best use and deployment of funds to provide the best outcome for the user. In an embodiment, the user (100) may elect to enable the system to automatically attend to payments with available funds and availing itself of credit from an external credit facility, as and when required, to place the user in a position wherein the best use is made of their financial resources for the purposes of settling invoices.

[0099] As will be appreciated, attending to timely settlement of invoices as and when settlement is required ensures that users are not deprived of services, including essential services, such as electricity, gas, local government services and telephone / data services. Similarly, other expenses that do not necessarily fall into the category of “essential services” may also be adequately serviced by the system according to the present invention such as school fees and/or club membership fees such that the user is not subject to cancellation of services provided by schools, universities and/or clubs of which they are a member which can significantly disrupt a user’s lifestyle and cause significant embarrassment.

[0100] As described above, the invention provides a system enabling a user to submit documents relating to income and expenditure and view the results of a first machine learning model and in the same embodiment, the result of a second machine learning model in an interface in a format such that the user can compare available funds, future income and expenses (including likely future expenses) to determine whether sufficient funds are available at the time required for the user to timely settle debts by the due date.

[0101] It will be appreciated by persons skilled in the relevant field of technology that numerous variations and/or modifications may be made to the invention as detailed in the embodiments without departing from the spirit or scope of the invention as broadly described. The present embodiments are, therefore, to be considered in all aspects as illustrative and not restrictive.

[0102] Throughout this specification and claims which follow, unless the context requires otherwise, the word “comprise”, and variations such as “comprises” and “comprising”, will be understood to imply the inclusion of a stated feature or step, or group of features or steps, but not the exclusion of any other feature or step or group of features or steps.

Claims

The claims defining the invention are as follows:

1. A computer-implemented method for generating a schedule, the method including: accessing, by one or more processors, one or more data records pertaining to a particular user; processing, by the one or more processors, the data records to generate data from the data records including identifying, by the one or more processors, using one or more character recognition techniques, a plurality of values embedded in the data records; utilizing, by the one or more processors, a first machine learning model to allocate each of the plurality of values to one of a plurality of categories, the first machine learning model receiving, as input, the data and the identified plurality of values, and the first machine learning model outputting information that identifies the category or categories to which each value has been allocated, automatically generating, by the one or more processors, a data structure associated with the particular user; storing, by the one or more processors, the data in the data structure thereby providing an index of the data according to the category, or categories, to which each value has been allocated; and extracting, by the one or more processors, based upon the categorisation of the values, information from the data structure to generate a schedule for the particular user.

2. A method according to claim 1 , wherein the one or more data records includes information pertaining to income and expenditure of the particular user.

3. A method according to claim 2, wherein the plurality of values embedded in the data that are identified using the one or more character recognition techniques include words, numbers and/or symbols that are likely to be relevant to income or expenditure of the particular user.

4. A method according to either claim 2 or claim 3, wherein the plurality of categories includes one category identifying values that relate to an income of the particular user, and another category identifying values that relate to an expenditure of the particular user.

5. A method according to any one of claims 2 to 4, wherein the plurality of categories include a category identifying values that are likely to be recurring values.

6. A method according to any one of claims 2 to 5, wherein the plurality of categories include categories relating to a frequency of recurrence in respect of recurring values.

7. A method according to any one of the preceding claims, the method further including: receiving, by the one or more processors, input from the particular user correcting the allocation of any one value to an incorrect category.

8. A method according to claim 7, the method further including: utilizing the data input into the first machine learning model, the information output from the first machine learning model, and the user input correcting the allocation of any one value to an incorrect category, to train the first machine learning model.

9. A method according to any one of the preceding claims, wherein generating the schedule for the particular user includes: utilizing a second machine learning model, the second model receiving, as input, the data extracted from the data structure based upon the categorization of the values, and the second model outputting predictions relating to income and expenditure of the particular user in a future time period for inclusion in the schedule.

10. A method according to claim 9, wherein the predictions relating to income and expenditure included in the schedule are based solely upon values that are categorized as recurring.

11. A method according to claim 10, wherein the predictions relating to income and expenditure included in the schedule are also based upon values that are categorized as non-recurring.

12. A method according to claim 11 , wherein when predicting income and expenditure in the future time period based upon both recurring and non-recurring values, more weight is given to recurring values than non-recurring values.

13. A method according to any one of claims 9 to 11 , the method further including, during or subsequent to said future time period: receiving, by the one or more processors, input from the particular user correcting a prediction regarding income or expenditure in the schedule based upon an actual income generated or expenditure incurred.

14. A method according to claim 14, the method further including: utilizing the data input into the second machine learning model, the predictions output from the first machine learning model, and the user input correcting an inaccurate income or expenditure, to train the second machine learning model.

15. A method according to any one of claims 9 to 14, wherein the schedule further indicates a cash flow position of the particular user in the future time period based upon the predicted income and expenditure, the cash flow position calculated based upon the total income and total expenditure for the future time period.

16. A method according to claim 15, the method further including: generating, by the one or more processors, one or more recommendations for the user to improve their cash flow position in the future time period.

17. A method according to claim 16, the method further including: generating, by the one or more processors, a survey for completion by the particular user, the survey including queries based upon the values that characterise a type of income or expenditure; transmitting, by the one or more processors, the survey to a device associated with the particular user; receiving, by the one or more processors, the user’s response to the survey; and training, by the one or more processors, the second machine learning model based upon the survey results, or generating, by the one or more processors, one or more recommendations based upon the survey results.

18. A method according to any one of the preceding claims, wherein the schedule includes one or more of: a spreadsheet, a calendar, and a graph.

19. A system for generating a schedule, the system including: one or more memories, and one or more processors operable to: access one or more data records pertaining to a particular user; process the data records to generate data from the data records including identifying, using one or more character recognition techniques, a plurality of values embedded in the data records; utilize a first machine learning model to allocate each of the plurality of values to one of a plurality of categories, the first machine learning model receiving, as input, the data and the identified plurality of values, and the first machine learning model outputting information that identifies the category or categories to which each value has been allocated; automatically generate a data structure associated with the particular user; store the data in the data structure thereby providing an index of the data according to the category, or categories, to which each value has been allocated; and extract, based upon the categorisation of the values, information from the data structure to generate a schedule for the particular use.

20. A computer-readable medium having a plurality of instructions executable by one or more processors to: access one or more data records pertaining to a particular user; process the data records to generate data from the data records including identifying, using one or more character recognition techniques, a plurality of values embedded in the data records; utilize a first machine learning model to allocate each of the plurality of values to one of a plurality of categories, the first machine learning model receiving, as input, the data and the identified plurality of values, and the first machine learning model outputting information that identifies the category or categories to which each value has been allocated; automatically generate a data structure associated with the particular user; store the data in the data structure thereby providing an index of the data according to the category, or categories, to which each value has been allocated; and extract, based upon the categorisation of the values, information from the data structure to generate a schedule for the particular use.