CN118312599A - Financial task execution method, apparatus, device, medium and program product - Google Patents

Abstract

The disclosure provides a financial task execution method, and relates to the field of artificial intelligence. The method is used for the financial intelligent agent, wherein the financial intelligent agent comprises a large language model, a planner and an executor, and specifically comprises the following steps: inputting a user instruction into the large language model, wherein the large language model is obtained in advance through training according to a user instruction sample corresponding to a historical financial task; invoking the large language model to predict user intent based on the user instructions; based on N interfaces and M groups of processes which are configured in advance, invoking the planner to determine an action sequence matching the user intention, wherein the action sequence is constrained by at least one group of processes, and the invoking object comprises at least one interface N, M which is an integer greater than or equal to 1 respectively; and calling the executor to execute the action sequence to complete the financial task indicated by the user instruction.

Description

Financial task execution method, apparatus, device, medium and program product

Technical Field

The present disclosure relates to the field of artificial intelligence, and more particularly, to a financial task execution method, apparatus, device, medium, and program product.

Background

In the financial arts, financial digital employees refer to software-based automated tools designed to mimic the performance of work tasks or business processes by human employees. Currently, the financial industry widely uses various automated tools to handle tasks such as transactions, data analysis, and customer service. These tools typically operate based on preset algorithms and rules. For example, in the customer service field, a chat robot or other digital assistant is used to automatically respond to customer consultation, thereby reducing the manpower requirements.

In implementing the disclosed inventive concepts, the inventors have found that existing automated tools often require a large amount of manual configuration and maintenance, and often lack sufficient flexibility to cope with rapid changes and complications in the financial field, such as limited effectiveness in processing complex natural language, inability to provide deep and accurate answers, and reduced user experience.

Disclosure of Invention

In view of the foregoing, the present disclosure provides financial task execution methods, apparatus, devices, media, and program products.

According to a first aspect of the present disclosure, there is provided a financial task execution method for a financial agent, the financial agent including a large language model, a planner and an executor, the method comprising: inputting a user instruction into the large language model, wherein the large language model is obtained in advance through training according to a user instruction sample corresponding to a historical financial task; invoking the large language model to predict user intent based on the user instructions; based on N interfaces and M groups of processes which are configured in advance, invoking the planner to determine an action sequence matching the user intention, wherein the action sequence is constrained by at least one group of processes, and the invoking object comprises at least one interface N, M which is an integer greater than or equal to 1 respectively; and calling the executor to execute the action sequence to complete the financial task indicated by the user instruction.

According to an embodiment of the present disclosure, pre-configuring the N interfaces and the M groups of flows includes: acquiring a configuration instruction described in natural language; analyzing the configuration instruction to obtain the N interfaces and M groups of processes; and configuring the N interfaces and the M groups of processes in a tool set of the financial intelligent agent.

According to an embodiment of the present disclosure, the acquiring the configuration instruction described in the natural language includes: providing a visual configuration interface for a user; and responding to the configuration operation of the user on the visual configuration interface, and obtaining the configuration instruction described in natural language.

According to an embodiment of the present disclosure, the visual configuration interface includes an input box, and the configuration operation includes an input operation at the input box; wherein, responding to the configuration operation of the user on the visual configuration interface, the configuration instruction described in natural language comprises the following steps: and acquiring natural language input by the user when the user executes the input operation so as to obtain the configuration instruction.

According to an embodiment of the present disclosure, the visual configuration interface includes a function list and a preview panel, and the configuration operation includes a drag operation on at least one function in the function list; wherein, responding to the configuration operation of the user on the visual configuration interface, the configuration instruction described in natural language comprises the following steps: acquiring at least one function dragged to the preview panel in the process of executing the dragging operation by the user; and generating natural language description content based on at least one function dragged to the preview panel to obtain the configuration instruction.

According to an embodiment of the present disclosure, parsing the configuration instruction to obtain the N interfaces and M sets of flows includes: matching S task templates based on an analysis result of analyzing the configuration instruction, wherein the S task templates are in one-to-one correspondence with S financial tasks, and S is an integer greater than or equal to 1; and extracting the N interfaces and M groups of processes from the S task templates.

According to an embodiment of the disclosure, invoking the actuator to perform the sequence of actions includes: after any one action in the action sequence is executed, a financial data processing result corresponding to the action is obtained; when the financial data processing result meets a preset condition, continuing to execute the next action in the action sequence, wherein the preset condition is determined according to the next action; and when the financial data processing result does not accord with the preset condition, calling the planner to update the action sequence based on the N interfaces and M groups of processes which are pre-configured.

Another aspect of an embodiment of the present disclosure provides a financial task execution apparatus for a financial agent, the financial agent including a large language model, a planner, and an executor, the apparatus comprising: the instruction input module is used for inputting user instructions into the large language model, wherein the large language model is obtained in advance through training according to user instruction samples corresponding to historical financial tasks; the model calling module is used for calling the large language model to predict the user intention based on the user instruction; the action planning module is used for calling the planner to determine an action sequence matching the user intention based on N interfaces and M groups of processes which are pre-configured, the action sequence is constrained by at least one group of processes, and a calling object comprises at least one interface N, M which is an integer greater than or equal to 1 respectively; and the task execution module is used for calling the executor to execute the action sequence so as to complete the financial task indicated by the user instruction.

Another aspect of an embodiment of the present disclosure provides an electronic device, including: one or more processors; and a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the method as described above.

Another aspect of the disclosed embodiments provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, causes the processor to perform the method as described above.

Another aspect of the disclosed embodiments provides a computer program product comprising a computer program which, when executed by a processor, implements a method as described above.

One or more of the above embodiments have the following advantages: a financial agent is provided to take charge of financial task execution, a large language model can be utilized to predict user intention based on user instructions, a planner is called to determine an action sequence, and an executor is called to execute the action sequence so as to complete financial tasks indicated by the user instructions. Based on N interfaces and M groups of processes which are configured in advance, the execution process of the financial intelligent agent can be restrained and guided, so that the execution accuracy of financial tasks is improved, the adaptability and the flexibility are higher, the capability of processing complex natural language queries and providing deep and accurate answers is improved, and the maintenance and configuration burden is reduced.

Drawings

The foregoing and other objects, features and advantages of the disclosure will be more apparent from the following description of embodiments of the disclosure with reference to the accompanying drawings, in which:

FIG. 1 schematically illustrates an application scenario diagram of a financial task execution method according to an embodiment of the present disclosure;

FIG. 2 schematically illustrates a flow chart of a financial task execution method according to an embodiment of the present disclosure;

FIG. 3 schematically illustrates a flow chart of a preconfigured interface and flow according to an embodiment of the present disclosure;

FIG. 4 schematically illustrates a flow diagram for retrieving configuration instructions described in natural language in accordance with an embodiment of the present disclosure;

FIG. 5 schematically illustrates a flow diagram for retrieving configuration instructions described in natural language in accordance with another embodiment of the present disclosure;

FIG. 6 schematically illustrates a flow chart of resolving the configuration instruction get interface and flow according to an embodiment of the disclosure;

FIG. 7 schematically illustrates a flow chart of a method of financial task execution according to another embodiment of the present disclosure;

FIG. 8 schematically illustrates a block diagram of a financial task execution device according to an embodiment of the present disclosure;

fig. 9 schematically illustrates a block diagram of an electronic device adapted to implement a financial task execution method according to an embodiment of the present disclosure.

Detailed Description

Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood that the description is only exemplary and is not intended to limit the scope of the present disclosure. In the following detailed description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the present disclosure. It may be evident, however, that one or more embodiments may be practiced without these specific details. In addition, in the following description, descriptions of well-known structures and techniques are omitted so as not to unnecessarily obscure the concepts of the present disclosure.

In the technical scheme of the invention, the related user information (including but not limited to user personal information, user image information, user equipment information, such as position information and the like) and data (including but not limited to data for analysis, stored data, displayed data and the like) are information and data authorized by a user or fully authorized by all parties, and the related data are collected, stored, used, processed, transmitted, provided, disclosed, applied and the like, all comply with related laws and regulations and standards, necessary security measures are adopted, no prejudice to the public order is provided, and corresponding operation entries are provided for the user to select authorization or rejection.

In the scenario of using personal information to make an automated decision, the method, the device and the system provided by the embodiment of the invention provide corresponding operation inlets for users, so that the users can choose to agree or reject the automated decision result. If the user selects refusal, the expert decision flow is entered. The expression "automated decision" here refers to an activity of automatically analyzing, assessing the behavioral habits, hobbies or economic, health, credit status of an individual, etc. by means of a computer program, and making a decision. The expression "expert decision" here refers to an activity of making a decision by a person who is specializing in a certain field of work, has specialized experience, knowledge and skills and reaches a certain level of expertise.

Fig. 1 schematically illustrates an application scenario diagram of a financial task execution method according to an embodiment of the present disclosure. It should be noted that fig. 1 is only an example in which embodiments of the present disclosure may be applied to help those skilled in the art understand the technical content of the present disclosure, but does not mean that embodiments of the present disclosure may not be used in other devices, systems, environments, or scenarios.

As shown in fig. 1, an application scenario 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used as a medium to provide communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.

The user may interact with the server 105 via the network 104 using the terminal devices 101, 102, 103 to receive or send messages or the like. Various communication client applications, such as shopping class applications, web browser applications, search class applications, instant messaging tools, mailbox clients, social platform software, etc. (by way of example only) may be installed on the terminal devices 101, 102, 103.

The terminal devices 101, 102, 103 may be a variety of electronic devices having a display screen and supporting web browsing, including but not limited to smartphones, tablets, laptop and desktop computers, and the like.

The server 105 may be a server providing various services, such as a background management server (by way of example only) providing support for websites browsed by users using the terminal devices 101, 102, 103. The background management server may analyze and process the received data such as the user request, and feed back the processing result (e.g., the web page, information, or data obtained or generated according to the user request) to the terminal device. For example, the server 105 may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing a basic cloud computing service such as a cloud service, a cloud computing service, a network service, or a middleware service.

It should be noted that the financial agent may be loaded in at least one of the terminal device or the server, for example, all of the terminal device or the server, or part of the server (e.g., a large language model), and part of the terminal device (e.g., the planner and the executor). Accordingly, the financial task execution method provided by the embodiments of the present disclosure may be generally executed by at least one of a terminal device or a server. Accordingly, the financial task execution device provided by the embodiments of the present disclosure may be generally provided in at least one of a terminal device or a server.

It should be understood that the number of terminal devices, networks and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.

The financial task execution method according to the embodiment of the present disclosure will be described in detail with reference to fig. 2 to 7 based on the scenario described in fig. 1.

Fig. 2 schematically illustrates a flowchart of a financial task execution method according to an embodiment of the present disclosure.

As shown in fig. 2, the financial task execution method of this embodiment is used for a financial agent including a large language model, a planner, and an executor.

An AI Agent (AI Agent) is illustratively a software entity implemented using artificial intelligence techniques that is capable of performing a particular task or service with a degree of autonomy, intelligence, and interactivity. The present disclosure relates to a financial agent that is an AI agent trained using financial data and is used for financial task execution.

LLM (large language model) is a deep learning based natural language processing model/generic language model that is capable of learning the grammar and semantics of natural language so that human readable text can be generated. LLM is typically based on neural network models, trained using large-scale financial corpora, such as using massive text data in banking systems. These models typically possess billions to trillions of parameters that are capable of handling various natural language processing tasks such as dialog systems, intelligent customer service, natural language generation, text classification, text summarization, machine translation, speech recognition, and the like. The core idea of LLM is to learn the pattern and language structure of natural language through extensive unsupervised training, which can learn the human language cognition and generation process to some extent. LLM is better able to understand and generate natural text than traditional NLP (Natural Language Processing ) models, while also exhibiting some logic thinking and reasoning capabilities.

The planner may include abstracted software components responsible for planning actions and deciding on the next actions of the agent. The planner may generate one or more possible sequences of actions based on the objectives and the current state of the agent. An executor may comprise an abstracted, software component that is responsible for performing the actual acts or operations, such as, for example, sending information, user interface operations, file operations, network communications, data processing, automation tasks, control hardware, trigger events, status updates, error handling, and error handling.

It is understood that a planner or an actuator may also refer to a physical tool that a computer device can control, such as an external device tool connected to the computer device, or a remote control tool that the computer device can remotely control. Or the planner or executor may be built based on another large language model.

The financial task execution method of this embodiment includes:

in operation S210, a user instruction is input into a large language model, wherein the large language model is trained in advance according to a user instruction sample corresponding to a historical financial task.

Illustratively, the user instruction is an instruction entered by a user, which may be at least one of text, audio, picture, video. The user instruction may be text information or voice information that the user describes a task desired to be performed using natural language. For example, the user instruction may be "please collect the last three years of financial reports of enterprise A, and analyze financial status" or the like.

Alternatively, this step may be performed by an electronic device, which may be a terminal device or a server. The electronic device can deploy pre-trained financial agents.

Illustratively, a large language model may employ models of ChatGPT, laMDA (Language Model for DialogueApplications, dialog application language model), techGPT (Technology-Oriented GENERATIVE PRE-trained Transformer, technology-Oriented generation pre-training transducer), chatGLM (Chat GenerativeLanguage Model, chat generation language model), and the like in the case of legal use. And can be obtained by fine tuning training by using user instruction samples corresponding to historical financial tasks. The fine-tuning will eventually produce a new model instance, although initialized with the pre-trained model, whose parameters will differ from those after pre-training. For example, the present disclosure relates to a large language model obtained by fine-tuning a loaded pre-trained model over a dataset comprising a plurality of batches of user instruction samples.

In operation S220, a large language model is invoked to predict user intent based on user instructions.

Illustratively, the user intent includes a financial task that the user wants to instruct the financial agent to perform. The large language model may generate predicted user intent through intent analysis.

In operation S230, based on the preconfigured N interfaces and M sets of flows, the call planner determines an action sequence matching the user intention, the action sequence being constrained by at least one set of flows, and wherein the call object includes at least one interface N, M being an integer greater than or equal to 1, respectively.

An interface is illustratively an API (Application Programming Interface ) that is a set of predefined functions, protocols, and tools for building software applications. APIs act as intermediaries between different software, enabling developers to write code using some off-the-shelf functionality rather than starting from scratch when designing a product. The main purpose of the API is to provide a way for applications to communicate with each other. Each set of processes includes one or more steps for providing references to the planner to constrain the precedence relationship between the actions in the sequence of actions to be consistent with the relationship between the steps in each set of processes, thereby improving the accuracy and usability of the planning result. For example, the planner may determine the action sequence by way of keyword matching, or may invoke a large language model to generate action parameters and determine the action sequence accordingly.

In operation S240, the executor is invoked to perform a sequence of actions to complete the financial task indicated by the user instruction.

For example, in a financial analysis scenario, a financial analysis instruction is issued to a financial agent. The financial agent automatically analyzes the specific requirements and complexity of the task according to semantic understanding, automatically disassembles the task into N actions, selects a proper interface, and automatically invokes and assembles the action sequence specific to the current task in series.

For example, in the financial analysis scene, skills such as registered annual report search, OCR (optical character recognition), mathematical calculation, external consultation search and the like are realized by a financial agent, and professional instructions in the business field of the digital staff are utilized, wherein the professional instructions comprise indexes which are required to be analyzed by the financial analysis key points, corresponding processing logic of the indexes (such as indexes of total assets, total asset growth rate, business income growth rate and the like), and operation instructions which are required to be subjected to standard flow on the professional scene (such as that annual report of enterprises is in pdf format and is required to be converted into characters through OCR recognition, and the pdf document is required to be analyzed by default, the annual report of enterprises of nearly 3 years is required to be analyzed, etc.), character targets are understood, skills can be called, a skill calling path is independently planned, and a return result is acquired by distributing and calling skills.

Specifically, in the financial analysis scenario, the following steps are performed by using a financial agent, the first step: and calling an enterprise annual report acquisition interface in the banking system to determine the company name and the year of the enterprise annual report. And a second step of: and calling OCR recognition skills to analyze the annual report information of the enterprise. And a third step of: and calling data processing skills according to the key analysis indexes, and calculating the growth rate by combining the recognition results. Fourth step: and analyzing the index data and the calculation result by combining the business requirement, outputting objective description of the data, and performing further business risk analysis on the enterprise to assist in actual business decision.

According to the embodiment of the disclosure, a financial agent is provided to be responsible for financial task execution, a large language model can be utilized to predict user intention based on user instructions, and a planner is called to determine an action sequence, and then an executor is called to execute the action sequence so as to complete financial tasks indicated by the user instructions. Based on N interfaces and M groups of processes which are configured in advance, the execution process of the financial intelligent agent can be restrained and guided, so that the execution accuracy of financial tasks is improved, the adaptability and the flexibility are higher, the capability of processing complex natural language queries and providing deep and accurate answers is improved, and the maintenance and configuration burden is reduced.

Fig. 3 schematically illustrates a flow chart of a preconfigured interface and flow according to an embodiment of the disclosure.

As shown in fig. 3, this embodiment includes:

in operation S310, a configuration instruction described in a natural language is acquired.

Illustratively, the configuration instructions provided by the user or administrator in natural language form are received, facilitating configuration by non-technical users as well.

In operation S320, the configuration instruction is parsed to obtain N interfaces and M sets of flows.

The configuration instructions described in natural language can be converted into configuration information which can be understood by a computer system, and interfaces and flows which need to be configured are defined.

In operation S330, the N interfaces and the M sets of processes are configured in the tool set of the financial agent.

For example, the tool set may store information of interfaces and flows, and the interface information may include at least one of the following information: interface name, interface type, interface description, interface version, interface parameters, URL (Uniform ResourceLocator ), runtime environment, installation style, and use case. The flow information may include at least one of the following: flow name, flow type, flow description, flow parameters, task samples, and use cases. The interface information and the flow information may be acquired in operation S310, for example, the interface information and the flow information are described in natural language to form a configuration instruction, thereby completing the automated configuration.

In addition to natural language description interface information and flow information, in other embodiments, task goals may be described in natural language, for example, in an online loan approval system of a bank, an administrator may wish a financial agent to enable loan risk screening to speed up approval of a small loan. The administrator inputs a configuration instruction 'risk examination during examination of each small loan' through natural language, the server automatically analyzes and configures a credit interface, a balance interface, a personal information interface, a flow interface and the like, and configures a risk examination flow, namely automatically configures interface information and flow information according to a task target described by the natural language.

In some embodiments, the user provides real-time feedback and verification portals during the configuration process to help optimize the configuration content.

According to the embodiment of the disclosure, through an intelligent configuration mode based on natural language, skill assembly and decision support conforming to task targets can be realized for the financial intelligent body, and the efficiency and accuracy of financial business processing are improved.

Fig. 4 schematically illustrates a flowchart of obtaining configuration instructions described in natural language, according to an embodiment of the disclosure.

As shown in fig. 4, this embodiment is one of the embodiments of operation S310, including:

in operation S410, a visual configuration interface is provided to a user.

In operation S420, a configuration instruction described in a natural language is obtained in response to a configuration operation of the user at the visual configuration interface.

Illustratively, the visual configuration interface includes an interface that a user can operate through a graphical interface, typically containing elements such as buttons, menus, dialog boxes, etc., so that the user can configure the system through intuitive operations such as text input, audio input, gesture input, etc. Configuration operations refer to a series of configuration actions performed by a user on a visual configuration interface, such as clicking, selecting, entering, etc. The user operation on the visual interface can be captured, the user configuration operation is converted into the instruction described by the natural language, and the configuration instruction is generated according to the configuration operation.

According to the embodiment of the disclosure, the threshold of configuration can be reduced, and the configuration interface is visualized so that a user with no technical background can easily realize configuration.

In some embodiments, the visual configuration interface includes an input box, and the configuration operation includes an input operation at the input box. Wherein, responding to the configuration operation of the user on the visual configuration interface, the configuration instruction described in the natural language comprises the following steps: and acquiring natural language input by a user when the user executes the input operation so as to obtain the configuration instruction.

For example, the input box is a component in a visual configuration interface that allows a user to input text information. Input operations refer to actions performed by a user in an input box, such as typing text.

For example, the user inputs "the requirement is to customize financial products for more than 100 thousands of users in the month stream through the input box of the visual configuration interface, please configure", analyze this natural language input, and automatically configure the associated interface and flow.

In some embodiments, a machine learning model may be trained to parse natural language and implement automatic configuration. In other embodiments, the configuration phase may be engaged using a financial agent, such as resolving configuration instructions using a large language model, and configuration planning by a planner, and finally executed by an executor.

According to the embodiment of the disclosure, a user can freely describe the configuration requirements according to the needs, so that the time of manual configuration is reduced, the degree of automation of configuration is improved, and the error rate is reduced.

Fig. 5 schematically illustrates a flowchart for retrieving configuration instructions described in natural language according to another embodiment of the present disclosure.

The visual configuration interface comprises a function list and a preview panel, and the configuration operation comprises a drag operation on at least one function in the function list.

The list of functions has one or more configurable functions listed therein from which the user may select. The preview pane is a display area to which a user can drag a function in the list of functions to preview or confirm the configuration effect.

As shown in fig. 5, in response to a configuration operation of a user in the visual configuration interface, the obtaining a configuration instruction described in a natural language in this embodiment includes:

At least one function dragged to the preview panel during the user' S execution of the drag operation is acquired in operation S510.

In operation S520, natural language description contents are generated based on at least one function dragged to the preview panel to obtain a configuration instruction.

Referring to fig. 5, a user may be allowed to select and configure a desired function through a direct interactive manner. After identifying which functions the user drags onto the preview pane, the user's operations are converted into easy-to-understand natural language descriptions to generate configuration instructions.

For example, the financial analysis corresponds to a annual report search function, a mathematical calculation function, an index extraction function, an external consultation search function, and the like. Customized configuration of the financial analysis task may be achieved by dragging one or more functions, and natural language descriptions may be generated to obtain configuration instructions. For example, according to the drag function, a natural language description is generated using a preset template. If the user drags the "annual report search" function, the use of the template may generate "add a function for searching for annual reports". Each function may be preconfigured with a corresponding template. In addition, the user may modify and confirm the generated natural language description.

In some embodiments, the configuration instructions may also be assembled in combination with text entered by the user in the input box, and one or more functions of the drag.

According to the embodiment of the disclosure, the operation intuitiveness of a user is improved, the understanding difficulty of a complex configuration flow is reduced through drag operation, intuitive selection can be provided for the user, the condition that the configuration is inaccurate only through inputting natural language is avoided, and the configuration process is more visualized and accurate.

Fig. 6 schematically illustrates a flow chart of a parse configuration instruction acquisition interface and flow in accordance with an embodiment of the disclosure.

As shown in fig. 6, this embodiment is one of the embodiments of operation S320, including:

in operation S610, S task templates are matched based on the analysis result of the analysis configuration instruction, where the S task templates correspond to the S financial tasks one by one, and S is an integer greater than or equal to 1.

Illustratively, the task template includes a predefined task framework for directing completion of a particular financial task. Financial tasks include specific tasks performed in the financial arts, such as transaction processing, risk assessment, and the like.

For example, a loan approval process template, a transaction monitoring process template, a property management process template, a customer service process template, a compliance check process template, a financial reporting process template, and the like are preconfigured.

The interface information in the loan approval process template comprises information of interfaces such as a client information interface, a credit scoring interface, a property evaluation interface and the like, and the process information comprises a client submitting a loan application. The system verifies the applicant identity through the customer information interface. The credit risk of the applicant is assessed using a credit scoring interface. The asset assessment interface evaluates the mortgage. And automatically or manually deciding whether to approve the loan according to the evaluation result.

The interface information in the transaction monitoring flow template comprises information of interfaces such as a transaction data interface, a risk assessment interface, an alarm system interface and the like, and the flow information comprises transaction records which are monitored in real time and flow in through the transaction data interface. The risk assessment interface analyzes the transaction for fraud or violations. If the fault exists, the alarm system interface triggers an alarm and notifies the compliance department.

The interface information in the asset management flow template comprises information of interfaces such as a market data interface, a portfolio management interface, a risk management interface and the like, and the flow information comprises the latest market information acquired through the market data interface. Asset allocation is adjusted according to market changes through the portfolio management interface. And evaluating risk exposure of the investment portfolio through a risk management interface and making risk control suggestions.

The interface information in the client service flow template comprises information of interfaces such as a client consultation interface, a service request interface, a feedback collection interface and the like, and the flow information comprises that a client submits a problem or consultation through the client consultation interface. The service demands of the clients are recorded through the service request interface and distributed to the corresponding service team. And after the service is finished, acquiring satisfaction feedback of the client through a feedback collection interface.

The interface information in the compliance checking flow template comprises information of interfaces such as a compliance rule interface, a transaction record interface, a report generation interface and the like, and the flow information comprises the latest compliance requirements and standards obtained through the compliance rule interface. Historical transaction data is provided through the transaction record interface for compliance analysis. Compliance reports are made periodically using the report generating interface for review by regulatory authorities.

The interface information in the financial report flow template comprises information of interfaces such as an accounting data interface, a report generation interface, an audit interface and the like, and the flow information comprises the step of collecting all financial transaction records through the accounting data interface. And automatically generating a financial statement according to the accounting criteria through the statement generation interface. And enabling an internal or external auditor to audit the financial statement through an audit interface.

In operation S620, N interfaces and M sets of flows are extracted from the S task templates.

For example, the configuration instruction may be "add VIP customer transaction monitoring flow". The system parses this instruction, understanding its meaning. And matching the transaction monitoring flow template from a database which is pre-stored with a plurality of task templates. The required interfaces (such as a transaction data interface, a risk assessment interface, an alarm system interface and the like) and flows (such as data acquisition, risk analysis and the like) are extracted from the transaction monitoring flow template.

According to the embodiment of the disclosure, the possibility of configuration errors in the configuration process is reduced by automatically analyzing the natural language and matching the task templates, so that the deployment speed and accuracy are improved. Can be quickly adjusted according to different configuration instructions so as to adapt to different tasks.

Fig. 7 schematically illustrates a flowchart of a financial task execution method according to another embodiment of the present disclosure.

As shown in FIG. 7, the embodiment includes operations S210-S240, wherein one embodiment of operation S240 includes operations S710-S740

In operation S710, after any one of the action sequences is performed, a financial data processing result corresponding to the action is obtained.

In operation S720, when the financial data processing result meets the preset condition, the next action in the action sequence is continuously executed, wherein the preset condition is determined according to the next action.

In operation S730, when the financial data processing result does not meet the preset condition, the planner is invoked to update the action sequence based on the N interfaces and the M sets of processes configured in advance.

The assembled workflow (i.e. action sequence) is executed by the financial agent, and the assembled workflow can be optimized and adjusted according to the task execution result and the user feedback, including disassembling a new task flow and selecting a new interface or flow. And meanwhile, the task progress and result clear report is provided by effectively interacting with the user.

For example, in a financial analysis scenario, if a call fails and the call result does not meet expectations, the financial agent may re-plan the path based on the actual return. Such as: when the annual report information skill is called, the interface report error indicates that the preset condition is not met, the parameters for executing the next action are difficult to obtain, and the planner can re-determine to call the annual report acquisition interface according to the actual situation. During management risk analysis, searching is found by combining with industry information, real-time information is insensitive after a large language model is returned, the management analysis result is evaluated to be improved, and a planner can plan and call an interface of an external information result by self according to actual conditions, so that a material basis is enhanced, and analysis quality is improved.

According to the embodiment of the disclosure, the automatic updating is performed according to the execution result, so that the efficiency and accuracy of financial business processing can be improved.

Based on the financial task execution method, the present disclosure also provides a financial task execution device. The device will be described in detail below in connection with fig. 8.

Fig. 8 schematically illustrates a block diagram of a financial task execution device according to an embodiment of the present disclosure.

As shown in fig. 8, the financial task execution device 800 of this embodiment includes an instruction input module 810, a model invoking module 820, an action planning module 830, and a task execution module 840.

The instruction input module 810 may perform operation S210 for inputting the user instruction into a large language model, wherein the large language model is trained in advance according to the user instruction samples corresponding to the historical financial tasks.

The model invoking module 820 may perform operation S220 for invoking the large language model to predict the user intention based on the user instruction.

The action planning module 830 may perform operation S230, where the action planning module is configured to invoke the planner to determine an action sequence matching the user intention based on the preconfigured N interfaces and M sets of flows, the action sequence is constrained by at least one set of flows, and the invocation target includes at least one interface N, M that is an integer greater than or equal to 1, respectively.

The task execution module 840 may perform operation S240 for invoking an executor to perform a sequence of actions to complete the financial task indicated by the user instruction.

In some embodiments, the task execution module 840 may execute the operations S710 to S730, which are not described herein.

In some embodiments, the financial task execution device 800 may further include a configuration module, which may execute operations S310-S330, operations S410-S420, operations S510-S520, and operations S610-S620, which are not described herein.

For parts of the device not mentioned, it can be understood with reference to the various embodiments of the method described above. That is, the apparatus portion comprises means for performing the steps of any of the method embodiments described above, respectively. In addition, the implementation manner, the solved technical problems, the realized functions and the realized technical effects of each module/unit/subunit and the like in the apparatus part embodiment are the same as or similar to the implementation manner, the solved technical problems, the realized functions and the realized technical effects of each corresponding step in the method part embodiment, and are not described herein again.

Any of the instruction input module 810, the model invocation module 820, the action planning module 830, and the task execution module 840 may be combined in one module to be implemented, or any of them may be split into multiple modules, according to embodiments of the present disclosure. Or at least some of the functionality of one or more of the modules may be combined with, and implemented in, at least some of the functionality of other modules.

At least one of the instruction input module 810, the model invocation module 820, the action planning module 830, and the task execution module 840 may be implemented, at least in part, as hardware circuitry, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a system on a chip, a system on a substrate, a system on a package, an Application Specific Integrated Circuit (ASIC), or by hardware or firmware, such as any other reasonable manner of integrating or packaging the circuitry, or in any one of or a suitable combination of three of software, hardware, and firmware, in accordance with embodiments of the present disclosure. Or at least one of the instruction input module 810, the model invocation module 820, the action planning module 830, and the task execution module 840 may be at least partially implemented as computer program modules that, when executed, perform the corresponding functions.

Fig. 9 schematically illustrates a block diagram of an electronic device adapted to implement a financial task execution method according to an embodiment of the present disclosure.

As shown in fig. 9, an electronic device 900 according to an embodiment of the present disclosure includes a processor 901 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 902 or a program loaded from a storage portion 908 into a Random Access Memory (RAM) 903. The processor 901 may include, for example, a general purpose microprocessor (e.g., a CPU), an instruction set processor and/or an associated chipset and/or a special purpose microprocessor (e.g., an Application Specific Integrated Circuit (ASIC)), or the like. Processor 901 may also include on-board memory for caching purposes. Processor 901 may include a single processing unit or multiple processing units for performing the different actions of the method flows according to embodiments of the present disclosure.

In the RAM 903, various programs and data necessary for the operation of the electronic device 900 are stored. The processor 901, the ROM 902, and the RAM 903 are connected to each other by a bus 904. The processor 901 performs various operations of the method flow according to the embodiments of the present disclosure by executing programs in the ROM 902 and/or the RAM 903. Note that the program may be stored in one or more memories other than the ROM 902 and the RAM 903. The processor 901 may also perform various operations of the method flow according to embodiments of the present disclosure by executing programs stored in one or more memories.

According to an embodiment of the disclosure, the electronic device 900 may also include an input/output (I/O) interface 905, the input/output (I/O) interface 905 also being connected to the bus 904. The electronic device 900 may also include one or more of the following components connected to the I/O interface 905: including an input portion 906 of a keyboard, mouse, etc. Including an output portion 907 such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), etc., and a speaker, etc. Including a storage portion 908 of a hard disk or the like. And a communication section 909 including a network interface card such as a LAN card, a modem, or the like. The communication section 909 performs communication processing via a network such as the internet. The drive 910 is also connected to the I/O interface 905 as needed. A removable medium 911 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is installed as needed on the drive 910 so that a computer program read out therefrom is installed into the storage section 908 as needed.

The present disclosure also provides a computer-readable storage medium that may be embodied in the apparatus/device/system described in the above embodiments. Or may exist alone without being assembled into the apparatus/device/system. The computer-readable storage medium carries one or more programs which, when executed, implement methods in accordance with embodiments of the present disclosure.

According to embodiments of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example, but is not limited to: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this disclosure, a computer-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. For example, according to embodiments of the present disclosure, the computer-readable storage medium may include ROM 902 and/or RAM 903 and/or one or more memories other than ROM 902 and RAM 903 described above.

Embodiments of the present disclosure also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowcharts. The program code, when executed in a computer system, causes the computer system to perform the methods provided by embodiments of the present disclosure.

The above-described functions defined in the system/apparatus of the embodiments of the present disclosure are performed when the computer program is executed by the processor 901. The systems, apparatus, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the disclosure.

In one embodiment, the computer program may be based on a tangible storage medium such as an optical storage device, a magnetic storage device, or the like. In another embodiment, the computer program may also be transmitted, distributed, and downloaded and installed in the form of a signal on a network medium, via communication portion 909, and/or installed from removable medium 911. The computer program may include program code that may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the foregoing.

In such an embodiment, the computer program may be downloaded and installed from the network via the communication portion 909 and/or installed from the removable medium 911. The above-described functions defined in the system of the embodiments of the present disclosure are performed when the computer program is executed by the processor 901. The systems, devices, apparatus, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the disclosure.

According to embodiments of the present disclosure, program code for performing computer programs provided by embodiments of the present disclosure may be written in any combination of one or more programming languages, and in particular, such computer programs may be implemented in high-level procedural and/or object-oriented programming languages, and/or assembly/machine languages. Programming languages include, but are not limited to, such as Java, c++, python, "C" or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, partly on a remote computing device, or entirely on the remote computing device or server. In the case of remote computing devices, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., connected via the Internet using an Internet service provider).

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

Those skilled in the art will appreciate that the features recited in the various embodiments of the disclosure and/or in the claims may be provided in a variety of combinations and/or combinations, even if such combinations or combinations are not explicitly recited in the disclosure. In particular, the features recited in the various embodiments of the present disclosure and/or the claims may be variously combined and/or combined without departing from the spirit and teachings of the present disclosure. All such combinations and/or combinations fall within the scope of the present disclosure.

The embodiments of the present disclosure are described above. These examples are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the embodiments cannot be used advantageously in combination. The scope of the disclosure is defined by the appended claims and equivalents thereof. Various alternatives and modifications can be made by those skilled in the art without departing from the scope of the disclosure, and such alternatives and modifications are intended to fall within the scope of the disclosure.

Claims (11)

1. A method of performing a financial task for a financial agent, the financial agent comprising a large language model, a planner, and an executor, the method comprising:

Inputting a user instruction into the large language model, wherein the large language model is obtained in advance through training according to a user instruction sample corresponding to a historical financial task;

invoking the large language model to predict user intent based on the user instructions;

Based on N interfaces and M groups of processes which are configured in advance, invoking the planner to determine an action sequence matching the user intention, wherein the action sequence is constrained by at least one group of processes, and the invoking object comprises at least one interface N, M which is an integer greater than or equal to 1 respectively;

and calling the executor to execute the action sequence to complete the financial task indicated by the user instruction.

2. The method of claim 1, wherein pre-configuring the N interfaces and M sets of flows comprises:

acquiring a configuration instruction described in natural language;

analyzing the configuration instruction to obtain the N interfaces and M groups of processes;

and configuring the N interfaces and the M groups of processes in a tool set of the financial intelligent agent.

3. The method of claim 2, wherein the obtaining configuration instructions described in natural language comprises:

Providing a visual configuration interface for a user;

And responding to the configuration operation of the user on the visual configuration interface, and obtaining the configuration instruction described in natural language.

4. The method of claim 3, wherein the step of,

The visual configuration interface comprises an input box, and the configuration operation comprises an input operation in the input box;

Wherein, responding to the configuration operation of the user on the visual configuration interface, the configuration instruction described in natural language comprises the following steps:

And acquiring natural language input by the user when the user executes the input operation so as to obtain the configuration instruction.

5. The method according to claim 3 or 4, wherein,

The visual configuration interface comprises a function list and a preview panel, and the configuration operation comprises a drag operation on at least one function in the function list;

Wherein, responding to the configuration operation of the user on the visual configuration interface, the configuration instruction described in natural language comprises the following steps:

Acquiring at least one function dragged to the preview panel in the process of executing the dragging operation by the user;

And generating natural language description content based on at least one function dragged to the preview panel to obtain the configuration instruction.

6. The method of claim 2, wherein parsing the configuration instruction to obtain the N interfaces and M sets of flows comprises:

Matching S task templates based on an analysis result of analyzing the configuration instruction, wherein the S task templates are in one-to-one correspondence with S financial tasks, and S is an integer greater than or equal to 1;

and extracting the N interfaces and M groups of processes from the S task templates.

7. The method of claim 1, wherein invoking the actuator to perform the sequence of actions comprises:

After any one action in the action sequence is executed, a financial data processing result corresponding to the action is obtained;

when the financial data processing result meets a preset condition, continuing to execute the next action in the action sequence, wherein the preset condition is determined according to the next action;

And when the financial data processing result does not accord with the preset condition, calling the planner to update the action sequence based on the N interfaces and M groups of processes which are pre-configured.

8. A financial task execution device for a financial agent, the financial agent comprising a large language model, a planner, and an executor, the device comprising:

the instruction input module is used for inputting user instructions into the large language model, wherein the large language model is obtained in advance through training according to user instruction samples corresponding to historical financial tasks;

The model calling module is used for calling the large language model to predict the user intention based on the user instruction;

The action planning module is used for calling the planner to determine an action sequence matching the user intention based on N interfaces and M groups of processes which are pre-configured, the action sequence is constrained by at least one group of processes, and a calling object comprises at least one interface N, M which is an integer greater than or equal to 1 respectively;

And the task execution module is used for calling the executor to execute the action sequence so as to complete the financial task indicated by the user instruction.

9. An electronic device, comprising:

One or more processors;

A memory for storing one or more computer programs,

Characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1-7.

10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, realizes the steps of the method according to any one of claims 1-7.

11. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, realizes the steps of the method according to any one of claims 1-7.