Program generation method and related device

Abstract

The invention discloses a proposal generation method and a related device, and relates to the field of data processing, session information of a user and an intention recognition cue word are input into a large language model, so that the large language model recognizes a user intention in the session information according to an intention recognition rule indicated by the intention recognition cue word. And inputting a parameter extraction cue word corresponding to the identified target intention identifier and the session information into a large language model, so that the large language model extracts a proposal parameter associated with the target intention identifier from the session information according to a parameter extraction rule indicated by the parameter extraction cue word. And then based on the target intention identifier and the proposal parameter, obtaining proposal associated content, finally inputting the proposal associated content, the proposal parameter and a proposal generation cue word corresponding to the target intention identifier into a large language model, and integrating the proposal associated content to obtain a proposal. The proposal which is highly accurate and meets the requirements of the user is efficiently generated.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a proposal generation method and related devices. Background Art

[0002] In the distribution of financial products like insurance, personalized recommendations need to be generated based on individual customer preferences, business scenarios, and other information to assist customers in purchasing financial products. However, traditional recommendation generation methods rely on fixed processes and steps, which are not only cumbersome but also require a high level of professional knowledge from the client. Consequently, the generated recommendations often fail to meet the user's personalized needs, reducing the user experience. Summary of the Invention

[0003] In view of the above problems, this application provides a proposal generation method and related devices to achieve the purpose of generating personalized proposals and improving user experience. The specific solution is as follows:

[0004] The first aspect of the present application provides a proposal generation method, comprising:

[0005] Get the user's session information;

[0006] Inputting the conversation information and the intent recognition prompt word into a large language model, causing the large language model to identify whether the conversation information contains the user intent according to the intent recognition rule indicated by the intent recognition prompt word, and obtaining intent recognition result data;

[0007] Inputting a parameter extraction prompt corresponding to a target intent identifier and the session information into the large language model, causing the large language model to extract a proposal parameter associated with the target intent identifier from the session information according to a parameter extraction rule indicated by the parameter extraction prompt, wherein the target intent identifier is an intent identifier corresponding to the user intent when the intent recognition result data indicates that the session information contains a user intent;

[0008] Acquiring proposal-related content based on the target intent identifier and the proposal parameters;

[0009] The proposal-related content, the proposal parameters, and the proposal generation prompt word corresponding to the target intent identifier are input into the large language model, so that the large language model integrates the proposal-related content according to the proposal generation rule indicated by the proposal generation prompt word to obtain the proposal.

[0010] In a possible implementation, before obtaining the proposal-related content based on the target intent identifier and the proposal parameters, the method further includes:

[0011] Determine whether the proposal parameters are complete based on the requirement parameters corresponding to the target intent identifier, and when it is determined that the proposal parameters are incomplete, initiate at least one round of parameter supplementation dialogue to the user, so that the user supplements the parameters according to the prompts of the parameter supplementation dialogue.

[0012] In one possible implementation, the intent recognition prompt includes: a first prompt content, a second prompt content, and a third prompt content, wherein the first prompt content is used to indicate the task description content, the second prompt content is used to indicate the schematic example, and the third prompt content is used to indicate the result output format. The large language model identifies whether the session information contains the user intent based on the intent recognition rule indicated by the intent recognition prompt, and obtains intent recognition result data including:

[0013] The large language model performs similarity recognition on the user's current intention in the session information based on the intention example, and outputs a similarity score and an identifier of the intention with the highest score based on the task description content;

[0014] The large language model outputs the similarity score and the highest-scoring intent identifier according to the result output format.

[0015] In one possible implementation, the intent recognition prompt further includes: fourth prompt content, the fourth prompt content is used to indicate intended restricted content, the large language model identifies whether the session information contains user intent based on the intent recognition rule indicated by the intent recognition prompt, and obtains intent recognition result data, further including:

[0016] When the large language model identifies at least two intent identifiers, one intent identifier is selected from all intent identifiers according to the intent restriction content.

[0017] In one possible implementation, the parameter extraction prompt includes: first parameter indication content, second parameter indication content, and third parameter indication content, where the first parameter indication content is used to indicate a parameter extraction range and a parameter output format, the second parameter indication content is used to indicate a parameter extraction type, and the third parameter indication content is used to indicate a parameter extraction location. Inputting the parameter extraction prompt corresponding to the target intent identifier and the session information into the large language model, causing the large language model to extract the proposal parameters associated with the target intent identifier from the session information according to the parameter extraction rule indicated by the parameter extraction prompt, includes:

[0018] The large language model searches for parameters in the session information according to the extraction range, and identifies target parameters according to the parameter extraction type. When the target parameters are repeated, the target parameters at the parameter extraction position are used as the proposal parameters and output according to the parameter output format.

[0019] In a possible implementation, obtaining proposal-related content based on the target intent identifier and the proposal parameters includes:

[0020] The target intention identifier and the proposal parameter are used as search keys, and a backend interface is called to filter out search values ​​matching the search keys from a database as the proposal-related content.

[0021] A second aspect of the present application provides a proposal generation device, comprising:

[0022] Session information acquisition module, used to obtain user session information;

[0023] an operation intention recognition module, configured to input the session information and the intention recognition prompt word into a large language model, causing the large language model to recognize whether the session information contains the user intention based on the intention recognition rules indicated by the intention recognition prompt word, and obtain intention recognition result data;

[0024] a suggestion parameter acquisition module, configured to input a parameter extraction prompt corresponding to a target intent identifier and the session information into the large language model, causing the large language model to extract suggestion parameters associated with the target intent identifier from the session information according to the parameter extraction rule indicated by the parameter extraction prompt, wherein the target intent identifier is an intent identifier corresponding to the user intent when the intent recognition result data indicates that the session information contains a user intent;

[0025] a suggestion content acquisition module, configured to acquire suggestion-related content based on the target intent identifier and the suggestion parameters; and

[0026] The suggestion content integration module is used to input the suggestion-related content, the suggestion parameters, and the suggestion generation prompt word corresponding to the target intent identifier into the large language model, so that the large language model integrates the suggestion-related content according to the suggestion generation rule indicated by the suggestion generation prompt word to obtain the suggestion.

[0027] A third aspect of the present application provides a computer program product, comprising computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements the proposal generation method according to the first aspect or any implementation of the first aspect.

[0028] A fourth aspect of the present application provides an electronic device, comprising at least one processor and a memory connected to the processor, wherein:

[0029] The memory is used to store computer programs;

[0030] The processor is configured to execute the computer program so that the electronic device can implement the proposal generation method of the first aspect or any implementation manner of the first aspect.

[0031] In a fifth aspect, the present application provides a computer storage medium carrying one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can generate a proposal according to the first aspect or any implementation of the first aspect.

[0032] By leveraging the above-mentioned technical solution, the proposal generation method provided in this application inputs the user's session information and intent recognition prompts into a large language model, which then identifies the user's intent in the session information according to the intent recognition rules indicated by the intent recognition prompts. The parameter extraction prompt corresponding to the identified target intent identifier and the session information are then input into the large language model, which then extracts the proposal parameters associated with the target intent identifier from the session information according to the parameter extraction rules indicated by the parameter extraction prompts. Based on the target intent identifier and the proposal parameters, the proposal-related content is then obtained. Finally, the proposal-related content, the proposal parameters, and the proposal generation prompt corresponding to the target intent identifier are input into the large language model, which then integrates the proposal-related content according to the proposal generation rules indicated by the proposal generation prompts to generate the proposal. Leveraging the deep understanding technology and generation capabilities of the large language model, user needs are analyzed and processed, enabling the efficient generation of highly accurate, professional, and user-specific proposals. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The above and other features, advantages, and aspects of the various embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that the originals and elements are not necessarily drawn to scale.

[0034] Figure 1 An architectural diagram of a proposal generation system provided for this application;

[0035] Figure 2 A flowchart of a proposal generation method provided for this application;

[0036] Figure 3Another flow chart of a proposal generation method provided for this application;

[0037] Figure 4 A structural diagram of a proposal generation device provided for this application;

[0038] Figure 5 This is a structural diagram of an electronic device provided in this application. DETAILED DESCRIPTION

[0039] The following describes the embodiments of the present application in conjunction with the accompanying drawings. The terms used in the implementation methods of the present application are only used to explain the specific embodiments of the present application and are not intended to limit the present application.

[0040] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0041] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, and this is merely a way of distinguishing the objects of the same attributes when describing them in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0042] See also Figure 1 , Figure 1 The schematic diagram of the architecture of a proposal generation system is shown. The system may include a terminal 100 and a server 200. The server 200 may provide the proposal generation method provided in the embodiment of the present application to one or more terminals.

[0043] Among them, the terminal 100 can be installed with an application for generating a proposal. The above application and web page can provide an interface. The terminal 100 can receive relevant parameters entered by the user on the proposal generation interface and send the above parameters to the server 200. The server 200 can obtain processing results based on the received parameters and return the processing results to the terminal 100.

[0044] It should be understood that in some optional implementations, the terminal 100 can also complete the action of obtaining the processing result based on the received parameters by itself without the need for the cooperation of the server, and the embodiments of the present application are not limited to this.

[0045] Next describe Figure 1 The product form of the mid-terminal 100;

[0046] The terminal 100 in the embodiment of the present application can be a mobile phone, a tablet computer, a wearable device, a vehicle-mounted device, an augmented reality (AR) / virtual reality (VR) device, a laptop computer, an ultra-mobile personal computer (UMPC), etc., and the embodiment of the present application does not impose any restrictions on this.

[0047] The terminal 100 may include components such as a radio frequency unit, a memory, an input unit, a display unit, a camera (optional), an audio circuit (optional), a speaker (optional), a microphone (optional), a headphone jack (optional), a processor, an external interface, and a power supply. Those skilled in the art will appreciate that the aforementioned components are merely examples and do not limit the terminal or multi-function device. The terminal or multi-function device may include more or fewer components, or may combine certain components or use different components.

[0048] The input unit can be used to receive input digital or character information and generate key signal input related to user settings and function control of the portable multi-function device. Specifically, the input unit may include a touch screen (optional) and / or other input devices. Specifically, other input devices may include, but are not limited to, one or more of a physical keyboard, function keys (such as volume control keys, power keys, etc.), a trackball, a mouse, a joystick, etc.

[0049] Among them, the input device can receive input data and so on.

[0050] The display unit can be used to display information input by the user or information provided to the user, various menus of the terminal, interactive interfaces, file display and / or playback of any multimedia file. In the embodiment of the present application, the display unit can be used to display the interface for generating the proposal, processing results, etc.

[0051] The memory may be used to store software codes related to the proposal generation method, the processor may execute the steps of the proposal generation method, and may also schedule other units (such as the above-mentioned input unit and display unit) to implement corresponding functions.

[0052] The radio frequency unit (optional) can be used to send and receive information or receive and send signals during a call.

[0053] In this embodiment of the present application, the radio frequency unit can send data to the server 200 and receive the processing results sent by the server 200.

[0054] It should be understood that the radio frequency unit is optional and can be replaced by other communication interfaces, such as a network port.

[0055] The terminal 100 further includes a power source (such as a battery) for supplying power to various components.

[0056] The terminal 100 also includes an external interface, which may be a standard Micro USB interface or a multi-pin connector, and may be used to connect the terminal 100 to other devices for communication, or to connect a charger to charge the terminal 100.

[0057] The server 200 includes a bus, a processor, a communication interface, and a memory. The processor, the memory, and the communication interface communicate with each other via the bus.

[0058] The memory may be used to store software codes related to the proposal generation method, the processor may execute the steps of the proposal generation method of the chip, and may also schedule other units to implement corresponding functions.

[0059] Current insurance proposal generation systems primarily rely on templates and simple rules. These systems rely on pre-set, fixed templates. Users fill in information through interactive interfaces by clicking, checking boxes, and filling in fields. The system then generates a proposal based on these user actions. This approach lacks the ability to understand complex user needs, resulting in fixed proposal formats that lack personalization and professionalism. Furthermore, these simple rules make it difficult to deeply analyze and present the complex terms and benefits of insurance products, resulting in a lack of professionalism and inability to effectively convince customers. Complicated processes and inaccurate proposals make it difficult for users to obtain valid information and reduce their satisfaction with insurance recommendations.

[0060] Furthermore, the commonly used proposal generation method, which combines rule engines with manual intervention, requires manual coding of numerous rules, making it difficult to cover complex and ever-changing sales scenarios. Furthermore, manual intervention increases costs and time, making it difficult to ensure accuracy and consistency.

[0061] In order to solve the above problems, the present invention provides a proposal generation method. The proposal generation method of the present invention is described in detail below with reference to the accompanying drawings.

[0062] Reference Figure 2 , Figure 2 A schematic diagram of a proposal generation method provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, a proposal generation method provided by an embodiment of the present application may include steps 201 to 205, and these steps are described in detail below.

[0063] 201. Obtain user session information.

[0064] Specifically, by monitoring the conversation information in the proposal generation interface on the terminal in real time, current and historical conversation information can be obtained. The conversation information can be in text or voice format. When the user is speaking in voice, a conversion program is required to convert the voice content into text.

[0065] It is understandable that when performing the above-mentioned speech-to-text conversion, a variety of models or methods can be used to convert the speech content, which will not be described in detail here.

[0066] 202. Input the conversation information and the intention recognition prompt word into the large language model, so that the large language model recognizes whether the conversation information contains the user intention according to the intention recognition rule indicated by the intention recognition prompt word, and obtains the intention recognition result data.

[0067] Specifically, real-time conversation information and intent recognition prompts are input into a large language model. The large language model is a large neural network model with powerful semantic understanding and generative reasoning capabilities, capable of processing and generating natural language text. Leveraging its semantic understanding capabilities, combined with intent recognition prompts, it analyzes whether the current conversation information contains the user's intent to generate a proposal. When designing the intent recognition prompts, the large language model can define the content and purpose of its processing in a task-oriented manner. This allows the large language model to identify the presence of the intent to generate a proposal in the conversation information based on the intent recognition prompts. For example, the large language model can be tasked with identifying the user's current intent based on the user's conversation. If the user's current intent is highly similar to a predefined system intent, a system intent identifier is generated and the similarity is scored. Example intent content corresponding to the corresponding intent identifier is also provided to assist the large language model in intent recognition.

[0068] 203. Input the parameter extraction prompt word corresponding to the target intent identifier and the session information into the large language model, so that the large language model extracts the proposal parameters associated with the target intent identifier from the session information according to the parameter extraction rule indicated by the parameter extraction prompt word. The target intent identifier is the intent identifier corresponding to the user intent when the intent recognition result data represents that the session information contains the user intent.

[0069] Specifically, after the large language model identifies the user's intent to generate a proposal, it needs to further obtain the proposal parameters required for proposal generation. Having already determined the corresponding target intent identifier, the large language model is fed the parameter extraction prompt corresponding to the target intent identifier and the session information. Guided by the parameter extraction prompt, the large language model identifies and extracts the required parameters in the session information, and then generates the proposal parameters corresponding to the target intent identifier. The proposal parameters are essential for proposal generation, such as age and insurance type.

[0070] 204. Based on the target intent identifier and the proposal parameters, obtain proposal-related content.

[0071] Specifically, based on the target intent identifier and corresponding proposal parameters, the corresponding proposal-related content is obtained from the background database according to the correspondence between the intent identifier and the proposal parameters and the proposal-related content. The proposal-related content here may include: customer personalized information (such as purchase history, risk preference, etc.), business scenario information (such as insurance product features, market trends, etc.) and dynamic knowledge (such as the latest insurance policies, industry dynamics, etc.).

[0072] 205. Input the proposal-related content, proposal parameters, and proposal generation prompt words corresponding to the target intent identifier into the large language model, so that the large language model integrates the proposal-related content according to the proposal generation rule indicated by the proposal generation prompt words to obtain the proposal.

[0073] Specifically, after obtaining the proposal-related content as described above, the language generation capability of the large language model is utilized. Under the guidance of the proposal generation prompt words corresponding to the target intent identifier, a proposal with richer content and more in line with the user's personalized needs is generated in accordance with the professional proposal format and logic.

[0074] This proposal generation method uses direct voice or text communication with the user, leveraging the deep understanding technology and text generation capabilities of a large language model to identify and process user conversational information to determine the user's proposal generation intention. Based on this, the proposed content is integrated based on the identified proposal parameters and related content, resulting in more precise proposals that meet the user's personalized needs. Furthermore, the voice or text interaction method effectively reduces user operation costs, speeds up proposal generation, and ultimately improves the circulation efficiency of financial products such as insurance.

[0075] In one embodiment, to further ensure the integrity of the proposal parameters and thereby improve the accuracy and reliability of subsequent proposal generation, before obtaining proposal-related content based on the target intent identifier and proposal parameters in step 204, the following steps are further included:

[0076] Whether the proposal parameters are complete is determined based on the requirement parameters corresponding to the target intent identifier, and when it is determined that the proposal parameters are incomplete, at least one round of parameter supplementation dialogue is initiated to the user, so that the user can supplement the parameters according to the prompts of the parameter supplementation dialogue.

[0077] Specifically, based on the identification of the target intent, the necessary proposal parameters for the generated proposal are also determined. Based on this, by comparing the proposal parameters identified in the session information with the necessary proposal parameters, it can be determined whether any parameters are missing. If the corresponding proposal parameters are missing, a session process for supplementing the proposal parameters is initiated with the user, prompting the user to supplement the corresponding proposal parameters. This achieves integrity verification of the proposal parameters, ensuring the completeness and accuracy of the subsequently generated proposal content.

[0078] As a specific implementation of step 202, the intent recognition prompt includes: a first prompt content, a second prompt content, and a third prompt content. The first prompt content is used to indicate the task description content, the second prompt content is used to indicate the schematic example, and the third prompt content is used to indicate the result output format. The large language model identifies whether the conversation information contains the user intent based on the intent recognition rules indicated by the intent recognition prompt, and obtains intent recognition result data, including:

[0079] Step 11: The large language model identifies the similarity of the user's current intent in the conversation information based on the intent examples, and outputs a similarity score and the highest-scoring intent identifier based on the task description content.

[0080] Step 12: The large language model outputs the similarity score and the highest-scoring intent identifier according to the result output format.

[0081] For example, the intent recognition prompt words may include the following:

[0082] #Task Description:

[0083] Your task is to identify the user's current intent based on their current question and past conversations. If the user's current intent is highly similar to a predefined intent, match it to the system intent identifier and score the similarity.

[0084] #Intent Example

[0085] •Intent ID: BUSINESS_009;

[0086] •Intent explanation: The user wants to generate an insurance proposal for the company’s products for the customer;

[0087] •Example:

[0088] o Proposal preparation or preparation of proposals;

[0089] Please help my client create a proposal.

[0090] oPlease prepare a proposal;

[0091] o Help me create a proposal and set the premium at x dollars;

[0092] Please help me create a proposal for the client, Zhang San.

[0093] o Please generate a proposal for Zhang San to meet with Younian;

[0094] o Please generate a proposal for a 30-year-old customer to meet the best years.

[0095] ## Output format

[0096] - `[{"Intent Type": "Enter the intent identifier here (the intent identifier is the code in the matching "System Intent List", and forgery is not allowed)", "Score": "Enter the matching score between the user's current question and the system intent here"}]`.

[0097] In addition, in some specific implementations, to further improve the accuracy of intent recognition, the intent recognition prompt word further includes: fourth prompt content, the fourth prompt content is used to indicate intended restricted content, and the large language model identifies whether the conversation information contains the user intent based on the intent recognition rules indicated by the intent recognition prompt word, and obtains intent recognition result data, which also includes:

[0098] When the large language model identifies at least two intent identifiers, it selects one intent identifier from all intent identifiers based on the intent restriction content.

[0099] Specifically, the fourth prompt may include the following:

[0100] If the system intent is matched, the degree of match between the user's current question and the matched system intent will be scored (the score range is 0-100, and only when the match degree is very high and there is almost no deviation can it be scored above 85 points).

[0101] If the user's current question involves multiple intentions, please select the one that best matches as the main intention for judgment.

[0102] In all other cases, `[{"Intent Type": "UNDEFINED_000", "Rating": "0"}]` is returned.

[0103] It is understandable that those skilled in the art can adaptively adjust and select the content of the above-mentioned intention recognition prompt words as needed, which will not be elaborated here.

[0104] In some embodiments, the parameter extraction prompt includes: first parameter indication content, second parameter indication content, and third parameter indication content, wherein the first parameter indication content is used to indicate the parameter extraction range and parameter output format, the second parameter indication content is used to indicate the parameter extraction type, and the third parameter indication content is used to indicate the parameter extraction location. The parameter extraction prompt corresponding to the target intent identifier and the conversation information are input into the large language model, so that the large language model extracts the proposal parameters associated with the target intent identifier from the conversation information according to the parameter extraction rule indicated by the parameter extraction prompt, including:

[0105] The large language model searches for parameters in the session information according to the extraction scope, and identifies target parameters according to the parameter extraction type. When target parameters are repeated, the target parameters at the parameter extraction position are used as proposal parameters and output according to the parameter output format.

[0106] Specifically, the content of the parameter extraction prompt word can be as follows:

[0107] The first parameter indicates the content:

[0108] As a professional information extraction and format conversion expert, he is good at accurately identifying and extracting the required parameters from user questions and previous conversations, and outputting them in the specified format.

[0109] The second parameter indicates the content:

[0110] Carefully read the parameters that need to be extracted, accurately identify and extract the required parameters from the user questions and the answers in the previous round.

[0111] The third parameter indicates the content:

[0112] Parameters are extracted from the user's question first. If the same parameters were extracted in the previous round of Q&A, the parameters in the user's question will prevail.

[0113] The parameters to be extracted can be set differently depending on the target intent identifier. Necessary parameters may include: user's name, insurance type name, age or birthday, etc.

[0114] Other non-essential parameters such as: customer gender, social security, disability certificate, premium, insured amount, payment period, payment type

[0115] , protection type, protection period, pension collection type, pension start age, dividend collection method, etc. can be selected and added according to user needs.

[0116] In other embodiments, to achieve more accurate and reliable acquisition of proposal-related content, the above-mentioned acquisition of proposal-related content based on the target intent identifier and proposal parameters includes:

[0117] The target intent identifier and proposal parameters are used as search keys, and the backend interface is called to filter out the search values ​​that match the search keys from the database as the proposal-related content.

[0118] Specifically, various proposal-related contents, intent identifiers, and proposal parameters may be stored in a database in the background system, thereby facilitating subsequent search for proposal-related contents based on target intent identifiers and proposal parameters.

[0119] As a specific application of the above proposal generation method, refer to Figure 3 As shown, the following processing procedures may be specifically included:

[0120] The salesperson opens the proposal generation software in the terminal and enters the chat interface.

[0121] Then enter the corresponding question by voice or text, for example: Q: Can you generate a proposal for me?

[0122] Intent recognition is performed through a large language model to determine whether the proposal meets the intent.

[0123] Based on the large language model's identification of the intention to generate a proposal, the large language model is used to extract the corresponding proposal parameters based on history and conversation.

[0124] Then, the completeness of the parameters is judged. If it is determined that the parameters are missing, modifications and parameter extraction are carried out through multiple rounds of interactive communication to improve the parameters of the proposal.

[0125] Then request the backend interface to obtain the related content generated by the proposal.

[0126] Finally, a large language model is used to generate a personalized proposal that is rich in content, well-structured, and tailored to user needs according to the professional insurance proposal format and logic, and then displayed to the user for confirmation on the front end of the application.

[0127] The above describes a proposal generation method provided by an embodiment of the present application. The following describes an apparatus for executing the above proposal generation method.

[0128] See also Figure 4 , Figure 4This is a schematic diagram of the structure of a proposal generation device provided in an embodiment of the present application. Figure 4 As shown, the proposal generating device includes:

[0129] The session information acquisition module 401 is used to acquire the user's session information;

[0130] Operation intention recognition module 402 is used to input the conversation information and the intention recognition prompt word into the large language model, so that the large language model can identify whether the conversation information contains the user intention according to the intention recognition rules indicated by the intention recognition prompt word, and obtain the intention recognition result data;

[0131] The suggestion parameter acquisition module 403 is configured to input the parameter extraction prompt word corresponding to the target intent identifier and the session information into the large language model, so that the large language model extracts the suggestion parameters associated with the target intent identifier from the session information according to the parameter extraction rule indicated by the parameter extraction prompt word. The target intent identifier is the intent identifier corresponding to the user intent when the intent recognition result data indicates that the session information contains the user intent.

[0132] The suggestion content acquisition module 404 is used to acquire suggestion-related content based on the target intent identifier and suggestion parameters; and

[0133] The suggestion content integration module 405 is used to input the suggestion related content, suggestion parameters and suggestion generation prompt words corresponding to the target intention identifier into the large language model, so that the large language model integrates the suggestion related content according to the suggestion generation rules indicated by the suggestion generation prompt words to obtain the suggestion.

[0134] In a possible implementation, the method further includes: a proposal parameter supplementing module configured to, before obtaining proposal-related content based on the target intent identifier and the proposal parameters:

[0135] Whether the proposal parameters are complete is determined based on the requirement parameters corresponding to the target intent identifier, and when it is determined that the proposal parameters are incomplete, at least one round of parameter supplementation dialogue is initiated to the user, so that the user can supplement the parameters according to the prompts of the parameter supplementation dialogue.

[0136] In one possible implementation, the intent recognition prompt word includes: a first prompt content, a second prompt content, and a third prompt content, the first prompt content is used to indicate the task description content, the second prompt content is used to indicate the schematic example, and the third prompt content is used to indicate the result output format. The large language model in the operation intention recognition module 402 identifies whether the session information contains the user intent according to the intent recognition rule indicated by the intent recognition prompt word, and the process of obtaining the intent recognition result data includes:

[0137] The large language model identifies the similarity between the user's current intent in the conversation information based on the intent examples, and outputs a similarity score and the highest-scoring intent identifier based on the task description content;

[0138] The large language model outputs a similarity score and the highest-scoring intent identifier based on the result output format.

[0139] In one possible implementation, the intent recognition prompt further includes: fourth prompt content, the fourth prompt content is used to indicate intended restricted content, and the operation intention recognition module 402 large language model identifies whether the session information contains the user intent based on the intent recognition rule indicated by the intent recognition prompt word, and the process of obtaining intent recognition result data further includes:

[0140] When the large language model identifies at least two intent identifiers, it selects one intent identifier from all intent identifiers based on the intent restriction content.

[0141] In one possible implementation, the parameter extraction prompt includes: first parameter indication content, second parameter indication content, and third parameter indication content. The first parameter indication content is used to indicate the parameter extraction range and parameter output format, the second parameter indication content is used to indicate the parameter extraction type, and the third parameter indication content is used to indicate the parameter extraction location. The process of the recommended parameter acquisition module 403 inputting the parameter extraction prompt corresponding to the target intent identifier and the session information into the large language model, and causing the large language model to extract the recommendation parameters associated with the target intent identifier from the session information according to the parameter extraction rule indicated by the parameter extraction prompt includes:

[0142] The large language model searches for parameters in the session information according to the extraction scope, and identifies the target parameters according to the parameter extraction type. When target parameters are repeated, the target parameters at the parameter extraction position are used as proposal parameters and output according to the parameter output format.

[0143] In one possible implementation, the process of obtaining the content associated with the proposal by the suggestion content acquisition module 404 based on the target intent identifier and the proposal parameters includes:

[0144] The target intent identifier and proposal parameters are used as search keys, and the backend interface is called to filter out the search values ​​that match the search keys from the database as the proposal-related content.

[0145] An electronic device is also provided in an embodiment of the present application. Figure 5 , which shows a schematic diagram of the structure of an electronic device suitable for implementing the embodiments of the present application. The electronic device in the embodiments of the present application may include, but is not limited to, fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 5 The electronic device shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0146] like Figure 5 As shown, the electronic device may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes based on programs stored in a read-only memory (ROM) 502 or programs loaded from a storage device 508 into a random access memory (RAM) 503. When the electronic device is powered on, the RAM 503 also stores various programs and data required for the operation of the electronic device. The processing device 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0147] Typically, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a memory card, a hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Figure 5 The electronic device is shown with various devices, but it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed instead.

[0148] An embodiment of the present application further provides a computer program product including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any one of the proposal generation methods provided in the embodiments of the present application.

[0149] An embodiment of the present application further provides a computer-readable storage medium, which carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any one of the proposal generation methods provided in the embodiments of the present application.

[0150] It should also be noted that the device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the device embodiments provided in this application, the connection relationship between the modules indicates that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines.

[0151] Through the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course can also be implemented by special hardware including application-specific integrated circuits, special CPUs, special memories, special components, etc. In general, all functions performed by computer programs can be easily implemented with corresponding hardware, and the specific hardware structures used to implement the same function can also be diverse, such as analog circuits, digital circuits or special circuits, etc. However, for the present application, software program implementation is a better implementation method in most cases. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer's floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk or optical disk, etc., and includes a number of instructions to enable a computer device (which can be a personal computer, training equipment, or network equipment, etc.) to execute the methods described in each embodiment of the present application.

[0152] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware, or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0153] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, a computer, a training device or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website, a computer, a training device or a data center. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integrations. The available medium can be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).

Claims

1. A method for generating a proposal, characterized in that: include: Get the user's session information; Inputting the conversation information and the intent recognition prompt word into a large language model, causing the large language model to identify whether the conversation information contains the user intent according to the intent recognition rule indicated by the intent recognition prompt word, and obtaining intent recognition result data; Inputting a parameter extraction prompt corresponding to a target intent identifier and the session information into the large language model, causing the large language model to extract a proposal parameter associated with the target intent identifier from the session information according to a parameter extraction rule indicated by the parameter extraction prompt, wherein the target intent identifier is an intent identifier corresponding to the user intent when the intent recognition result data indicates that the session information contains a user intent; Acquiring proposal-related content based on the target intent identifier and the proposal parameters; The proposal-related content, the proposal parameters, and the proposal generation prompt word corresponding to the target intent identifier are input into the large language model, so that the large language model integrates the proposal-related content according to the proposal generation rule indicated by the proposal generation prompt word to obtain the proposal.

2. The method for generating a proposal according to claim 1, wherein: Before obtaining the proposal-related content based on the target intent identifier and the proposal parameters, the method further includes: Determine whether the proposal parameters are complete based on the requirement parameters corresponding to the target intent identifier, and when it is determined that the proposal parameters are incomplete, initiate at least one round of parameter supplementation dialogue to the user, so that the user supplements the parameters according to the prompts of the parameter supplementation dialogue.

3. The method for generating a proposal according to claim 1, wherein: The intent recognition prompt includes: a first prompt content, a second prompt content, and a third prompt content, wherein the first prompt content is used to indicate a task description content, the second prompt content is used to indicate a schematic example, and the third prompt content is used to indicate a result output format. The large language model identifies whether the session information contains user intent based on the intent recognition rules indicated by the intent recognition prompt, and obtains intent recognition result data, including: The large language model performs similarity recognition on the user's current intention in the session information based on the intention example, and outputs a similarity score and an identifier of the intention with the highest score based on the task description content; The large language model outputs the similarity score and the highest-scoring intent identifier according to the result output format.

4. The method for generating a proposal according to claim 3, wherein: The intention recognition prompt further includes: a fourth prompt content, the fourth prompt content being used to indicate intended restricted content; the large language model identifying whether the conversation information contains the user intention based on the intention recognition rule indicated by the intention recognition prompt, and obtaining intention recognition result data, further including: When the large language model identifies at least two intent identifiers, one intent identifier is selected from all intent identifiers according to the intent restriction content.

5. The method for generating a proposal according to claim 1, wherein: The parameter extraction prompt includes: first parameter indication content, second parameter indication content, and third parameter indication content, wherein the first parameter indication content is used to indicate a parameter extraction range and a parameter output format, the second parameter indication content is used to indicate a parameter extraction type, and the third parameter indication content is used to indicate a parameter extraction location. Inputting the parameter extraction prompt corresponding to the target intent identifier and the session information into the large language model, so that the large language model extracts the proposal parameters associated with the target intent identifier from the session information according to the parameter extraction rule indicated by the parameter extraction prompt, includes: The large language model searches for parameters in the session information according to the extraction range, and identifies target parameters according to the parameter extraction type. When the target parameters are repeated, the target parameters at the parameter extraction position are used as the proposal parameters and output according to the parameter output format.

6. The method for generating a proposal according to any one of claims 1 to 5, characterized in that: The acquiring of proposal-related content based on the target intent identifier and the proposal parameters includes: The target intention identifier and the proposal parameter are used as search keys, and a backend interface is called to filter out search values ​​matching the search keys from a database as the proposal-related content.

7. A proposal generating device, characterized in that: include: Session information acquisition module, used to obtain user session information; an operation intention recognition module, configured to input the session information and the intention recognition prompt word into a large language model, causing the large language model to recognize whether the session information contains the user intention based on the intention recognition rules indicated by the intention recognition prompt word, and obtain intention recognition result data; a suggestion parameter acquisition module, configured to input a parameter extraction prompt corresponding to a target intent identifier and the session information into the large language model, causing the large language model to extract suggestion parameters associated with the target intent identifier from the session information according to the parameter extraction rule indicated by the parameter extraction prompt, wherein the target intent identifier is an intent identifier corresponding to the user intent when the intent recognition result data indicates that the session information contains a user intent; A suggestion content acquisition module, configured to acquire suggestion-related content based on the target intent identifier and the suggestion parameters; as well as, The suggestion content integration module is used to input the suggestion-related content, the suggestion parameters, and the suggestion generation prompt word corresponding to the target intent identifier into the large language model, so that the large language model integrates the suggestion-related content according to the suggestion generation rule indicated by the suggestion generation prompt word to obtain the suggestion.

8. A computer program product, characterized in that The method comprises computer-readable instructions, which, when executed on an electronic device, enable the electronic device to implement the proposal generating method according to any one of claims 1 to 6.

9. An electronic device, characterized in that: comprising at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is configured to execute the computer program so as to enable the electronic device to implement the proposal generating method according to any one of claims 1 to 6.

10. A computer storage medium, characterized in that The storage medium carries one or more computer programs, and when the one or more computer programs are executed by an electronic device, the electronic device can implement the proposal generation method according to any one of claims 1 to 6.

Images