Dialogue generation method and device based on application tool, electronic device, and medium

Abstract

The embodiments of the present application provide a method and device, electronic device, and medium for generating a conversation based on an application tool, which belongs to the field of artificial intelligence technology and is applicable to the field of financial technology. The method includes: obtaining initial conversation data of a target object; performing intent recognition on the initial conversation data to obtain initial conversation intention data; performing task detection on the initial conversation data based on the initial conversation intention data to determine a target task label; obtaining tool attribute data of a candidate application tool, and selecting an application tool based on the initial conversation intention data, the target task label, and the tool attribute data to determine a target application tool; obtaining a preset process execution instruction of the target application tool; wherein the preset process execution instruction is used to instruct the target object to execute a preset execution process; and performing conversation generation based on the initial conversation intention data and the preset process execution instruction. The embodiments of the present application can improve the accuracy of conversation generation related to the task matched by the application tool.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology and is applicable to the field of financial technology, and in particular to a method and device for generating a dialogue based on an application tool, an electronic device, and a medium. Background Art

[0002] Application tools are designed within intelligent systems to help users complete tasks or provide services. They can be standalone applications or plug-ins integrated into other software. For example, in a fintech claims system, application tools could include claims information collection tools, claims information management tools, and claims resource settlement tools. These tools could be used by professionals such as insurance agents and brokers. Users can then use different application tools within the claims system to perform different tasks, such as collecting claims information and arranging claims resources, thereby improving task completion efficiency.

[0003] Currently, due to users' lack of proficiency in the use of application tools, and the frequent updates of application tools included in intelligent systems, it is difficult for users to quickly and accurately master the specific steps for using application tools, thus being unable to complete tasks such as claim information collection and claim resource settlement. Furthermore, while related technologies can provide users with answers to questions in intelligent dialogue scenarios within intelligent systems, they do not link dialogue generation with application tools, which can easily reduce the accuracy of dialogue generation related to the tasks matched by the application tools. Therefore, how to improve the accuracy of dialogue generation related to the tasks matched by application tools and avoid users' inability to complete tasks efficiently due to their lack of proficiency in the use of application tools has become a technical problem that needs to be solved urgently. Summary of the Invention

[0004] The main purpose of the embodiments of the present application is to propose a conversation generation method and device based on an application tool, an electronic device, and a medium, aiming to improve the accuracy of conversation generation related to the tasks matched by the application tool, and to avoid the user being unable to complete the task efficiently due to lack of proficiency in the use of the application tool.

[0005] To achieve the above objectives, a first aspect of an embodiment of the present application provides a method for generating a dialogue based on an application tool, the method comprising:

[0006] Acquiring initial conversation data of a target object in a target application, wherein the target application is provided with candidate application tools, each of the candidate application tools has a candidate function, and the candidate functions of any two candidate application tools are different;

[0007] Performing intent recognition on the initial conversation data to obtain initial conversation intention data;

[0008] Performing task detection on the initial conversation data based on the initial conversation intention data to determine a target task tag for the target object; wherein each target task tag is used to indicate a task function, and the task function is configured with a preset execution process;

[0009] Acquiring tool attribute data of the candidate application tool, and performing application tool selection on the candidate application tool based on the initial conversation intention data, the target task label, and the tool attribute data, to determine a target application tool, wherein the candidate function of the target application tool is the same as the task function, the candidate function is configured with a candidate execution process, and the candidate execution process of the target application tool is the same as the preset execution process;

[0010] Obtaining a preset process execution instruction of the target application tool; wherein the preset process execution instruction is used to instruct the target object to execute the preset execution process;

[0011] A dialogue is generated based on the initial dialogue intention data and the preset process execution instruction.

[0012] In some embodiments, generating a dialogue based on the initial dialogue intention data and the preset process execution instruction includes:

[0013] Determining candidate process nodes of the preset execution process;

[0014] Extracting node data from the preset process execution instruction based on the candidate process node to obtain candidate node data of the candidate process node, wherein the candidate node data includes a node processing intention and node input data, and the node processing intentions of different candidate process nodes are different;

[0015] Performing node selection on the candidate process nodes based on the initial conversation intention data and the node processing intention, determining a target process node, and using the node input data of the target process node as target input data, wherein the intention indicated by the initial conversation intention data is the same as the node processing intention;

[0016] Performing data type detection on the target input data to obtain input data type;

[0017] A dialogue generation is performed on the initial dialogue intention data, the target input data, and the input data type based on a preset dialogue generation model.

[0018] In some embodiments, the input data type includes a data upload type and a data selection type, wherein the data upload type is used to indicate uploading data of a preset data type pre-configured for the target process node, and the data selection type is used to indicate selecting data from preset task data configured for the target process node;

[0019] The performing dialogue generation on the initial dialogue intention data, the target input data, and the input data type based on a preset dialogue generation model includes:

[0020] If the input data type is the data upload type, obtaining node configuration parameters of the target process node, and performing a dialog generation on the initial dialog intention data, the target input data, and the node configuration parameters based on the preset dialog generation model to obtain a first generated dialog, wherein the first generated dialog is used to guide the target object to upload data of the preset data type, and the node configuration parameters are conditions that must be met for data uploaded at the target process node;

[0021] If the input data type is the data selection type, the preset task data of the target process node is obtained, and a dialogue is generated for the initial dialogue intention data, the target input data and the preset task data based on the preset dialogue generation model to obtain a second generated dialogue, which is used to guide the target object to perform data selection on the preset task data.

[0022] In some embodiments, the method further comprises:

[0023] Obtaining the response data fed back by the target object, wherein the response data is a response file uploaded by the target object or a response text input by the target object;

[0024] Sending the reply data to the calling interface of the target application tool, and obtaining data execution information of the target application tool on the reply data from the calling interface;

[0025] Determining a process execution state of the target task based on the data execution information;

[0026] Update the preset process execution instruction based on the process execution status to obtain the current process execution instruction;

[0027] A dialogue is generated based on the current process execution instruction, the initial dialogue intention data and the response data.

[0028] In some embodiments, selecting the candidate application tools based on the initial conversation intention data, the target task label, and the tool attribute data to determine the target application tool includes:

[0029] Performing multi-level intent recognition on the initial dialogue intention data and the target task label based on a preset multi-level attention hybrid model to obtain multi-level dialogue intention data;

[0030] Performing data encoding on the multi-level dialogue intention data to obtain a multi-level dialogue intention vector;

[0031] Performing data encoding on the tool attribute data to obtain a tool attribute vector;

[0032] Calculating similarity between the multi-level dialogue intention vector and the tool attribute vector to obtain vector similarity of the candidate application tool;

[0033] An application tool selection is performed on the candidate application tools based on the vector similarity to determine the target application tool.

[0034] In some embodiments, performing multi-level intent recognition on the initial dialogue intent data and the target task label based on a preset multi-level attention hybrid model to obtain multi-level dialogue intent data includes:

[0035] Extracting data features from the initial conversation intention data to obtain an initial conversation intention feature, wherein the initial conversation intention feature includes an initial conversation intention sub-feature, and the initial conversation intention sub-feature is provided with an initial interaction item, wherein the initial interaction item is used to indicate an item associated with a conversation interaction;

[0036] Performing label feature extraction on the target task label to obtain target task features;

[0037] Combining the sub-features of the initial conversation intention based on a preset number of tail items to obtain a combined conversation intention feature;

[0038] Performing attention calculation based on the target task feature and the combined dialogue intention feature to obtain a combined attention feature;

[0039] Perform a linear transformation on the combined attention features to obtain the multi-level dialogue intention data.

[0040] In some embodiments, performing task detection on the initial conversation data based on the initial conversation intention data to determine the target task label of the target object includes:

[0041] Extracting intention keywords from the initial conversation intention data to obtain conversation intention keywords;

[0042] Obtaining candidate task keywords associated with preset candidate task tags from a preset tag library;

[0043] Calculating similarity between the conversation intention keyword and the candidate task keyword to obtain keyword similarity;

[0044] The target task tag of the target object is determined from the candidate task tags based on the keyword similarity.

[0045] To achieve the above-mentioned objectives, a second aspect of an embodiment of the present application provides a conversation generation device based on an application tool, the device comprising:

[0046] A first acquisition module is configured to acquire initial conversation data of a target object in a target application, wherein the target application is provided with candidate application tools, each of the candidate application tools has a candidate function, and the candidate functions of any two candidate application tools are different;

[0047] an identification module, configured to perform intent identification on the initial conversation data to obtain initial conversation intention data;

[0048] a detection module, configured to perform task detection on the initial conversation data based on the initial conversation intention data, and determine a target task tag of the target object; wherein each target task tag is used to indicate a task function, and the task function is configured with a preset execution process;

[0049] a tool selection module, configured to obtain tool attribute data of the candidate application tools, and select an application tool for the candidate application tools based on the initial conversation intention data, the target task label, and the tool attribute data, to determine a target application tool, wherein the candidate function of the target application tool is the same as the task function, the candidate function is configured with a candidate execution process, and the candidate execution process of the target application tool is the same as the preset execution process;

[0050] A second acquisition module is configured to acquire a preset process execution instruction of the target application tool; wherein the preset process execution instruction is used to instruct the target object to execute the preset execution process;

[0051] A dialogue generation module is used to generate a dialogue based on the initial dialogue intention data and the preset process execution instruction.

[0052] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the application tool-based dialogue generation method described in the first aspect above.

[0053] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the application tool-based dialogue generation method described in the first aspect above.

[0054] The present application proposes a method and device for generating a conversation based on an application tool, an electronic device, and a medium. The method obtains initial conversation data of a target object in a target application, wherein the target application is provided with candidate application tools, each candidate application tool has a candidate function, and the candidate functions of any two candidate application tools are different; further, intent recognition is performed on the initial conversation data to obtain initial conversation intent data; and task detection is performed on the initial conversation data based on the initial conversation intent data to determine a target task label of the target object; wherein each target task label is used to indicate a task function, and the task function is configured with a preset execution process; further, tool attribute data of the candidate application tools is obtained, and application tool selection is performed on the candidate application tools based on the initial conversation intent data, to determine a target application tool, wherein the candidate function of the target application tool is the same as the task function, the candidate function is configured with a candidate execution process, and the candidate execution process of the target application tool is the same as the preset execution process; and a preset process execution instruction is obtained for the target application tool; wherein the preset process execution instruction is used to instruct the target object to execute the preset execution process; further, conversation generation is performed based on the initial conversation intent data and the preset process execution instruction. Compared to related techniques that require users to learn to use application tools in order to complete corresponding tasks, this application can link dialogue generation with application tools. That is, the target user can use dialogue to call the corresponding application tool and rely on the application tool's preset process execution instructions to continuously generate dialogue. This avoids tedious tool screening and manual operation processes, saves time, and reduces the risk of human error. Therefore, this application can effectively improve the accuracy of dialogue generation related to the task matched by the application tool, avoiding the user's lack of proficiency in the use of the application tool and the inability to complete the task efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flow chart of a method for generating a conversation based on an application tool provided in an embodiment of the present application;

[0056] Figure 2 yes Figure 1 A flowchart of step S130 in FIG.

[0057] Figure 3 yes Figure 1 A flowchart of step S140 in FIG.

[0058] Figure 4 yes Figure 3 A flowchart of step S310 in FIG.

[0059] Figure 5 yes Figure 1 A flowchart of step S160 in FIG.

[0060] Figure 6 is another flow chart of the method for generating a conversation based on an application tool provided in an embodiment of the present application;

[0061] Figure 7 This is a specific flow chart of the method for generating a conversation based on an application tool provided in an embodiment of the present application;

[0062] Figure 8 This is a structural diagram of a conversation generation device based on an application tool provided in an embodiment of the present application;

[0063] Figure 9 This is a hardware structure diagram of the electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0065] It should be noted that although the device schematics illustrate functional module divisions and the flowcharts illustrate logical sequences, in certain circumstances, the steps shown or described may be performed in a sequence that differs from the module divisions in the device or the sequence in the flowcharts. The terms "first," "second," and so on, in the specification, claims, and drawings, are used to distinguish similar items and are not necessarily used to describe a specific sequence or precedence.

[0066] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0067] First, let’s analyze some of the terms used in this application:

[0068] Artificial intelligence (AI) is a new technical discipline that studies and develops theories, methods, technologies, and application systems for simulating, extending, and expanding human intelligence. A branch of computer science, AI seeks to understand the essence of intelligence and produce new intelligent machines that can respond in a manner similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thinking. It also encompasses the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, to perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results.

[0069] Natural language processing (NLP): NLP uses computers to process, understand, and apply human languages ​​(such as Chinese and English). A branch of artificial intelligence, NLP is an interdisciplinary field between computer science and linguistics, often referred to as computational linguistics. Natural language processing encompasses grammatical analysis, semantic analysis, and discourse comprehension. Natural language processing is commonly used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intent recognition, information extraction and filtering, text classification and clustering, public opinion analysis, and opinion mining. It encompasses data mining, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research related to language processing, and linguistics research related to language computing.

[0070] Large Language Models (LLMs) are language models based on deep learning and natural language processing techniques. Pre-trained LLMs are applied to specific tasks or datasets, and model parameters are adjusted through supervised learning to adapt to the specific task. At their core, LLMs utilize deep learning methods, typically based on the Transformer architecture, which can capture the complexity and diversity of language.

[0071] Application tools are designed within intelligent systems to help users complete tasks or provide services. They can be standalone applications or plug-ins integrated into other software. For example, in a fintech claims system, application tools could include claims information collection tools, claims information management tools, and claims resource settlement tools. These tools could be used by professionals such as insurance agents and brokers. Users can then use different application tools within the claims system to perform different tasks, such as collecting claims information and arranging claims resources, thereby improving task completion efficiency.

[0072] Currently, understanding and using application tools requires a learning curve and operational thresholds. Although developers believe they have designed their products from the user's perspective as closely as possible, users are often unfamiliar with the tools and the tools included in intelligent systems are frequently updated, making it difficult for them to quickly and accurately master the specific steps for using them. This makes it difficult for them to complete tasks such as collecting claims information and settling claims resources. Furthermore, while related technologies can provide users with answers to questions in intelligent dialogue scenarios within intelligent systems, they do not link dialogue generation to the tools, which can easily reduce the accuracy of dialogue generation related to the tasks matched by the tools. Therefore, improving the accuracy of dialogue generation related to tasks matched by application tools, and preventing users from being unable to efficiently complete tasks due to their lack of familiarity with the tools, has become a pressing technical issue.

[0073] Based on this, the embodiments of the present application provide a method and device, electronic device, and medium for generating a conversation based on an application tool, aiming to improve the accuracy of conversation generation related to the task matched by the application tool, and to avoid the user being unskilled in using the application tool and unable to complete the task efficiently.

[0074] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0075] Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0076] The dialogue generation method based on application tools provided in the embodiment of the present application relates to the field of artificial intelligence technology. The dialogue generation method based on application tools provided in the embodiment of the present application can be applied to a terminal, can be applied to a server side, or can be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or as a server cluster or distributed system composed of multiple physical servers, or as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms; the software can be an application that implements the dialogue generation method based on application tools, etc., but is not limited to the above forms.

[0077] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments in which tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media, including storage devices.

[0078] It should be noted that in each specific embodiment of the present application, when it comes to the need to perform relevant processing based on data related to the identity or characteristics of the object, such as the object's historical data, object conversation data, and data uploaded by the object, the object's permission or consent will be obtained first, and the collection, use, and processing of such data will comply with relevant laws, regulations, and standards. In addition, when the embodiment of the present application needs to obtain the sensitive personal information of the object, the object's separate permission or consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the object's separate permission or consent, the necessary object-related data for the normal operation of the embodiment of the present application will be obtained.

[0079] See also Figure 1 , Figure 1This is an optional flow chart of the method for generating a conversation based on an application tool provided in an embodiment of the present application. In some embodiments, Figure 1 The method may include but is not limited to steps S110 to S160:

[0080] Step S110, obtaining the initial conversation data of the target object in the target application;

[0081] Step S120, performing intent recognition on the initial conversation data to obtain initial conversation intent data;

[0082] Step S130, performing task detection on the initial conversation data based on the initial conversation intention data to determine the target task label of the target object;

[0083] Step S140 , obtaining tool attribute data of candidate application tools, and selecting application tools for the candidate application tools based on the initial conversation intention data, the target task label, and the tool attribute data, to determine the target application tool;

[0084] Step S150, obtaining a preset process execution instruction of the target application tool;

[0085] Step S160: generating a dialogue based on the initial dialogue intention data and the preset process execution instruction.

[0086] In steps S110 to S160 of some embodiments, the embodiments of the present application integrate other complex application tools into the background of the system (such as the claims system) by retaining a simple intelligent dialogue part. In this way, the background call processing can be quickly performed according to the intention requirements of the operation object, so that the target product required by the operation object can be directly output. In this way, the tedious tool screening and manual operation process can be avoided, saving time and reducing the risk of human error. In addition, the intelligent interaction method of the present application can effectively improve the accuracy of dialogue generation related to the task matched by the application tool, and avoid the user's lack of proficiency in the use of the application tool and the inability to complete the task efficiently.

[0087] In step S110 of some embodiments, the target application refers to an application that can conduct intelligent dialogue, such as the application corresponding to the insurance company's claims management system, the application corresponding to the digital medical system, etc. The target application is provided with candidate application tools, each candidate application tool has a candidate function, and the candidate functions of any two candidate application tools are different. For example, in the application corresponding to the financial technology's claims management system, the application tools can be tools corresponding to different functions in the claims management system, such as a claims information collection tool, a claims information management tool, a claims resource settlement tool, etc. The target object can be an insurance agent, a broker, and other professionals. For another example, the candidate function corresponding to the claims information collection tool can be the policy opening function, and the candidate function corresponding to the claims information management tool can be the customer claims information query. At this time, when the insurance agent wants to open a policy for his or her client, he or she can trigger the control corresponding to the claims information collection tool in the application corresponding to the claims management system, and start to operate step by step based on the policy opening process pre-set in this tool.

[0088] It should be noted that the candidate functions included in the target application may also include marketing center, business opportunity filling, material submission, video interview, electronic signing, contract inquiry, order management, residual value assessment, etc., which can cover various business nodes in the entire business management process.

[0089] It's important to note that initial conversation data refers to the data generated by the target subject's conversational interaction with the intelligent robot within the target application's dialog window. In other words, initial conversation data captures the entire interaction between the target subject and the target application—not just a single input, but the context of the entire conversation. This allows for a better understanding of the target subject's input and the provision of an appropriate response. In practice, the target subject can open a dialog window within the target application and, through this dialog, enter data such as text or images into the system, enabling the target subject and the target application to communicate their task objectives or requirements through natural language interaction.

[0090] In step S120 of some embodiments, to understand the target's task objectives, such as electronic signing or contract inquiry, the present application may further perform intent recognition on the initial conversation data to obtain initial conversation intent data. This initial conversation intent data is used to characterize the target's goal or intent in the initial conversation data. The initial conversation intent data can be a label or a more detailed description of the intent, providing a basis for subsequent processing and response.

[0091] It should be noted that this application can use natural language processing technology, such as text classification and named entity recognition based on linguistic rules and machine learning models, to analyze and understand the initial dialogue data input by the target object to identify the intention behind it, thereby determining the task the target object wants to perform or the information it wants to obtain. For example, the target object inputs the initial dialogue data as "How do I know the company's A realization rate?" At this time, the target application can start the dialogue generation method based on the application tool to guide the target object to enter more information, thereby accurately identifying whether the target object has a simple search intention or a comparative analysis intention, and then calling the corresponding application tool to implement the function.

[0092] In step S130 of some embodiments, after obtaining the initial conversation intention data, the present application can perform task detection on the initial conversation data based on the initial conversation intention data to determine the specific task of the target object, that is, to determine the target task label. Each task is associated with a task label, which represents the task function, and each task label is configured with a preset execution process, that is, the steps required to complete the task. The preset execution process is a series of standardized steps designed to complete the task, ensuring that the task can be completed in the expected manner. For example, if the target object is a simple search intention, the task function corresponding to the target task label can be "contract query". At this time, the preset execution process corresponding to the task function may include: entering query information and query conditions, and then entering further screening conditions based on the returned data.

[0093] In some embodiments, see Figure 2 , Figure 2 This is an optional flowchart of step S130 provided in an embodiment of the present application. Step S130 may specifically include but is not limited to steps S210 to S240:

[0094] Step S210, extracting intention keywords from the initial conversation intention data to obtain conversation intention keywords;

[0095] Step S220: acquiring candidate task keywords associated with preset candidate task tags from a preset tag library;

[0096] Step S230, calculating the similarity between the conversation intention keywords and the candidate task keywords to obtain keyword similarity;

[0097] Step S240 : determining a target task tag of the target object from the candidate task tags based on the keyword similarity.

[0098] In step S210 of some embodiments, when determining the target task label, the present application may first extract intent keywords from the initial conversation intent data to extract conversation intent keywords that are more representative of the main needs or purposes of the target object.

[0099] In step S220 of some embodiments, the present application may further access a preset tag library, which contains candidate task keywords associated with each candidate task tag. These candidate task keywords are predefined and used to describe and distinguish different tasks.

[0100] In some embodiments, in step S230, similarity may be calculated between the conversation intent keywords and the candidate task keywords of the candidate task tags. Similarity calculation methods may include cosine similarity, Jaccard similarity, etc., to quantify the similarity between the conversation intent keywords and the candidate task keywords, thereby more accurately detecting the target task tag of the target object.

[0101] In step S240 of some embodiments, further, the keyword similarities of multiple candidate task keywords can be numerically compared, and the candidate task tag corresponding to the largest keyword similarity can be used as the target task tag. Alternatively, the present application can arrange the keyword similarities of multiple candidate task keywords in descending order, and filter out the most relevant target task tags based on a preset similarity threshold. That is to say, when performing tag matching based on a preset similarity threshold, if the matching result is empty, intent recognition can be re-performed or the preset tag library can be updated to improve the accuracy of task detection.

[0102] In the above embodiment, in order to more accurately match the intention of the target object with the tasks that the target application can perform, the present application can be implemented through keyword similarity calculation, thereby improving the target application's understanding of the target object's intention and providing more accurate services or responses.

[0103] In step S140 of some embodiments, the tool attribute data is used to characterize the characteristics and capabilities of the corresponding candidate application tool. The tool attribute data may include the functional description, performance indicators, user evaluation, compatibility information, etc. of the candidate application tool, without specific limitation. The target application tool is the tool that best matches the target task label selected from the candidate application tools based on the initial dialogue intention data, the target task label and the tool attribute data. The candidate function of the target application tool is the same as the task function, the candidate function is configured with a candidate execution process, and the candidate execution process of the target application tool is the same as the preset execution process.

[0104] It should be noted that this application can select the most suitable one from multiple candidate application tools to perform the task according to the task requirements of the target object, and this tool can complete the task in the expected manner. In addition, the process of determining the target application tool in this application is equivalent to conversation-based function recommendation, that is, the conversation-based function recommendation module can be started based on the intent recognition result (that is, the initial conversation intention data). This module evaluates the current conversation stage and context information, and selects a suitable tool from a series of available tools based on a multi-level reasoning strategy to locate a solution that better meets the needs of the end user.

[0105] In some embodiments, see Figure 3 , Figure 3 This is an optional flowchart of step S140 provided in an embodiment of the present application. Step S140 may specifically include but is not limited to steps S310 to S350:

[0106] Step S310: Perform multi-level intent recognition on the initial dialogue intent data and the target task label based on a preset multi-level attention hybrid model to obtain multi-level dialogue intent data;

[0107] Step S320, encoding the multi-level dialogue intention data to obtain a multi-level dialogue intention vector;

[0108] Step S330, encoding the tool attribute data to obtain a tool attribute vector;

[0109] Step S340 , performing similarity calculation on the multi-level dialogue intention vector and the tool attribute vector to obtain the vector similarity of the candidate application tool;

[0110] Step S350 : selecting an application tool from the candidate application tools based on the vector similarity to determine a target application tool.

[0111] In step S310 of some embodiments, the present application may adopt a multi-level reasoning strategy (i.e., a multi-level attention hybrid model) to help prune a large amount of search space, avoid local optimal solutions, and converge to a more satisfactory solution by considering the overall behavior trend of the user (i.e., considering the initial dialogue intent data and the target task label). Specifically, the present application may use a preset multi-level attention hybrid model to process the initial dialogue intent data and the target task label. This model can recognize and understand multiple levels of intent in a dialogue, such as surface intent, deep intent, and implicit intent. This model can extract richer and more detailed intent information to form multi-level dialogue intent data.

[0112] It should be noted that the multi-level nature of this application refers to understanding user intent at different levels of abstraction. Surface intent refers to needs or questions directly expressed by the target object. Deep intent refers to intentions that may not be directly expressed by the target object but can be inferred through context. Implicit intent refers to potential needs that may not be explicitly expressed by the target object but are inferred by the system through information such as the target object's behavior patterns and historical interactions.

[0113] It should be noted that in order to obtain these high-level concepts, two SBR-related inductive biases, namely local invariance and inherent priority, are used to reduce the search space. It is understandable that the multi-level attention hybrid model used in this application can be a model built based on the Atten-Mixer model, a large model, etc., without limitation.

[0114] In some embodiments, see Figure 4 , Figure 4 This is an optional flowchart of step S310 provided in an embodiment of the present application. Step S310 may specifically include but is not limited to steps S410 to S45:

[0115] Step S410: extracting data features from the initial conversation intention data to obtain initial conversation intention features;

[0116] Step S420: extracting label features from the target task label to obtain target task features;

[0117] Step S430, combining the initial conversation intention sub-features based on a preset number of tail items to obtain a combined conversation intention feature;

[0118] Step S440, performing attention calculation based on the target task feature and the combined dialogue intention feature to obtain a combined attention feature;

[0119] Step S450: Perform a linear transformation on the combined attention features to obtain multi-level dialogue intention data.

[0120] In step S410 of some embodiments, the present application may extract key features from the initial conversation intent data. These features are referred to as initial conversation intent features. These features are further broken down into smaller units, namely initial conversation intent sub-features. Each sub-feature is associated with a specific initial interaction item, which indicates a specific interaction point or topic in the conversation. In other words, an initial interaction item is used to indicate the items associated with a conversation interaction.

[0121] In step S420 of some embodiments, the present application further performs feature extraction on the target task label. These features can describe the attributes or characteristics of the task corresponding to the target task label.

[0122] In step S430 of some embodiments, the present application may further combine the initial conversation intent sub-features into a larger feature set, called a combined conversation intent feature, based on preset rules, such as a preset number of tail items. This helps capture more complex patterns and relationships in the conversation. In other words, the multi-level attention hybrid model can form groups based on different numbers of preset tail items. These groups are used to construct related high-level concepts, thereby generating group representations.

[0123] In steps S440 and S450 of some embodiments, the present application may further use the target task features and the combined conversation intent features to calculate attention weights to determine which parts of the conversation are more important for understanding the user's intent and completing the task. Finally, a linear transformation is performed on the combined attention features to generate multi-level conversation intent data. This transformation may involve weight adjustment, feature scaling, or other mathematical operations to convert the features into a form suitable for subsequent processing, such as input into a classifier or prediction model.

[0124] In the above embodiments, the present application converts the original dialogue intent data into structured, information-rich multi-level dialogue intent data through feature extraction, feature combination and attention calculation. These data can be used in more complex dialogue systems and machine learning tasks, helping to improve the system's understanding of user intentions and optimize the efficiency of dialogue management and task execution.

[0125] In step S320 of some embodiments, after identifying the multi-level conversation intent data, the present application may encode the multi-level conversation intent data into a vector form, namely a multi-level conversation intent vector. This encoding typically involves converting text data into a numerical representation so that it can be processed by a machine learning model. This process may include techniques such as word embedding (such as Word2Vec or BERT) to convert text into a fixed-length vector.

[0126] In step S330 of some embodiments, the present application may also encode the tool attribute data into a vector form, i.e., a tool attribute vector. This step involves converting the tool attributes (such as functional description, performance indicators, etc.) into numerical representations for subsequent similarity calculations.

[0127] In some embodiments, in steps S340 and S350, the present application may further calculate the similarity between the multi-level conversation intent vectors and the tool attribute vectors to quantify the degree of match between the conversation intent and the tool attributes. The most suitable candidate application tool is then selected based on the calculated vector similarity. This selection process may involve selecting the tool with the highest similarity or determining the most relevant tool based on a certain threshold. The target application tool ultimately determined is the tool that best matches the user's intent.

[0128] In the above embodiments, the present application achieves accurate selection of candidate application tools through multi-level intent recognition and vector similarity calculation to ensure that the system can provide the most relevant services or responses to users, thereby improving the accuracy of the system's understanding and response to user intentions.

[0129] In step S150 of some embodiments, after determining the target application tool, the present application can directly call the selected tool and guide the target user to enter the data required by the tool, that is, according to the needs of the tool, guide the target user to enter relevant information in the dialogue interface. Specifically, the preset process execution instruction of the target application tool can be first obtained, and the preset process execution instruction is used to instruct the target user to execute the preset execution process.

[0130] In step S160 of some embodiments, the present application may further use the target object's initial intention data and preset process execution instructions to generate a dialogue to guide the target object to complete the required task or provide further guidance.

[0131] It should be noted that the conversational interface of the target application of this application can support multimodal input, that is, the input data can be images, files, text, etc., without limitation. After the target object enters a piece of information, the instructions can be executed according to the updated preset process, guiding the target object to enter new data until the task function is completed.

[0132] In some embodiments, see Figure 5 , Figure 5 This is an optional flowchart of step S160 provided in an embodiment of the present application. Step S160 may specifically include but is not limited to steps S510 to S550:

[0133] Step S510, determining a candidate process node of a preset execution process;

[0134] Step S520 , extracting node data from the preset process execution instruction based on the candidate process node to obtain candidate node data of the candidate process node;

[0135] Step S530 , selecting candidate process nodes based on the initial conversation intention data and the node processing intention, determining a target process node, and using the node input data of the target process node as target input data;

[0136] Step S540, performing data type detection on the target input data to obtain the input data type;

[0137] Step S550 , performing dialogue generation on the initial dialogue intention data, target input data, and input data type based on a preset dialogue generation model.

[0138] In step S510 of some embodiments, the present application may first identify all candidate process nodes included in the preset execution process. These candidate process nodes represent various stages or steps in the preset execution process. Each candidate process node corresponds to a specific processing task or decision point. The preset process execution instruction is guidance information instructing the target object to execute the preset execution process.

[0139] In step S520 of some embodiments, the candidate node data includes node processing intentions and node input data, and different candidate process nodes have different node processing intentions. The present application can extract data related to each candidate process node from the preset process execution instructions, which include node processing intentions (i.e., the tasks or goals that the node needs to complete) and node input data (i.e., the content type, information format, etc. of the input information required to execute the node task). Different candidate process nodes may have different processing intentions, so their node input data will also be different.

[0140] In some embodiments, in step S530, the intent indicated by the initial conversation intent data is the same as the node processing intent. The present application can match the initial conversation intent data with the processing intents of candidate process nodes to determine which target process node is most suitable for processing the current conversation intent. The selection of this target process node must also comply with the node execution order of the preset execution process. Once a matching node is found, the node is determined as the target process node, and its corresponding node input data is selected as the target input data for subsequent processing.

[0141] In step S540 of some embodiments, the present application may analyze the determined target input data to identify the required input data type. The input data type may include different types of data such as text, numbers, dates, pictures, files, etc.

[0142] In step S550 of some embodiments, the present application may further utilize a preset dialogue generation model, combining the initial dialogue intent data, the target input data, and the identified input data type, to generate a dialogue response to the target object. This dialogue generation model may be based on machine learning or natural language processing technology, and can generate natural and appropriate responses based on the input data and context.

[0143] In some embodiments, the input data type includes a data upload type and a data selection type. The data upload type indicates uploading data of a preset data type pre-configured for the target process node, and the data selection type indicates selecting data from the preset task data configured for the target process node. In this case, the process of generating a dialogue based on the preset dialogue intent data, target input data, and input data type based on the preset dialogue generation model may specifically include:

[0144] If the input data type is a data upload type, the node configuration parameters of the target process node are obtained, and a dialog generation is performed based on the initial dialog intent data, the target input data, and the node configuration parameters based on the preset dialog generation model to obtain a first generated dialog. The first generated dialog is used to guide the target object to upload data of the preset data type. The node configuration parameters are the conditions that must be met for data uploaded to the target process node.

[0145] If the input data type is a data selection type, the preset task data of the target process node is obtained, and a dialogue is generated for the initial dialogue intention data, the target input data, and the preset task data based on the preset dialogue generation model to obtain a second generated dialogue. The second generated dialogue is used to guide the target object to perform data selection on the preset task data.

[0146] It should be noted that the data upload type is used to indicate that the target object needs to upload non-text data such as images and files within the dialog box of the conversation window. The preset data type refers to at least one type of image, file, etc. The data selection type is used to indicate that the target object needs to select output data from the preset task data configured for the target process node output from the historical conversation within the dialog box of the conversation window. In other words, if the input data type is the data upload type, a conversation is generated based on the preset conversation generation model based on the initial conversation intent data, the target input data, and the node configuration parameters to obtain a first generated conversation. The first generated conversation is used to guide the target object to upload data of the preset data type. The node configuration parameters are the conditions that must be met for data uploaded at the target process node. At this point, the target object can send images or files based on the preset data type to the dialog box. The target application can then directly upload this content to the target application tool in the background, and the automatic script will perform a series of processing of the tool.

[0147] If the input data type is data selection, a dialog generation process is performed based on the initial dialog intent data, the target input data, and the preset task data, generating a second generated dialog. This second generated dialog guides the target user to select the preset task data, and this second generated dialog includes the preset task data. The target user can then respond to the selected data in the dialog box without jumping to the target application tool to perform the operation.

[0148] In some embodiments, see Figure 6 , Figure 6 This is another optional flow chart of the method for generating a conversation based on an application tool provided in an embodiment of the present application. After step S160, the method for generating a conversation based on an application tool provided in an embodiment of the present application may further include steps S610 to S650:

[0149] Step S610, obtaining the response data fed back by the target object;

[0150] Step S620 , sending the reply data to the calling interface of the target application tool, and obtaining data execution information of the target application tool on the reply data from the calling interface;

[0151] Step S630, determining the process execution status of the target task based on the data execution information;

[0152] Step S640, updating the preset process execution instruction based on the process execution status to obtain the current process execution instruction;

[0153] Step S650: generate a dialogue based on the current process execution instruction, the initial dialogue intention data and the response data.

[0154] In steps S610 and S620 of some embodiments, when generating a dialogue, the present application can continuously update the generated dialogue according to the requirements of the task function in the target application tool. Specifically, the present application can obtain the reply data currently fed back by the target object in real time. The reply data is a reply file uploaded by the target object (such as PDF, Word document, etc.) or a reply text directly entered in the dialog box. Further, these reply data are sent to the calling interface of the target application tool. This interface can be an application programming interface (API), which allows the system to communicate and exchange data with external application tools. Through this interface, the system requests the target application tool to perform specific processing or analysis on the reply data.

[0155] In step S630 of some embodiments, after sending the reply data to the target application tool, the system receives the result information of the target application tool performing the operation on the reply data from the calling interface. This information may include processing status, analysis results or other feedback, and this data execution information is crucial for determining subsequent steps. Furthermore, the process execution status of the target task can be determined based on the data execution information received from the target application tool. This status may indicate whether the task has been completed, whether further operation is required, or whether there are errors or abnormal conditions, etc.

[0156] In step S640 of some embodiments, the present application may then update the preset process execution instructions based on the determined process execution status. These instructions guide the system on how to continue executing subsequent steps or adjust the process path based on the current execution status.

[0157] In some embodiments, step S650 may further generate a dialogue using the current process execution instruction, the target object's initial dialogue intent data, and the user's response data. This dialogue may confirm the user's response, provide further inquiries, or provide information required to complete the task, ensuring the coherence and relevance of the dialogue while maintaining user engagement.

[0158] In some specific embodiments, see Figure 7 , Figure 7 It is a specific flow chart of the dialogue generation method based on application tools provided in an embodiment of the present application. In actual application, the present application can first start a dialogue in the target application, and the target application can perform intent recognition on the initial dialogue data input by the target object to obtain initial dialogue intention data. Furthermore, the target application can perform multi-level intent recognition on the initial dialogue intention data and the target task label based on a multi-level attention hybrid model, thereby determining the target application tool. Furthermore, the generated dialogue can call the target application tool to guide the target object to input the data required by the tool. When the auxiliary process of the dialogue system ends, the output generated by the target application tool can be passed into the dialog box via the original environment.

[0159] The embodiment of the present application provides a method for generating dialogues based on application tools. The method describes the task objectives or requirements by having a dialogue with the system. The target object can complete the task without directly operating complex tools. This intelligent interaction method provides a more convenient, fast and personalized user experience. Secondly, the present application can use natural language processing technology to parse and understand user input and select the corresponding tool according to the intention. This process avoids the tedious tool screening and manual operation process, saves time and reduces the risk of human error. In addition, the function recommendation module based on the conversation of the present application adopts a multi-level reasoning strategy as a decision-making mechanism and considers contextual information to improve the quality of decision-making, which helps to provide more comprehensive and accurate matching of needs. Among them, the present application can use a multi-level attention hybrid model based on the Atten-Mixer model structure to construct a multi-level attention hybrid model, which not only processes single user input, but analyzes it in the context of the entire conversation. The model can understand the changes in user intentions over time and take into account historical interactions when generating responses. Combined with dynamic context fusion technology, this technology can update the conversation context in real time and integrate it into the current user intention recognition process. For example, customized word embedding layers are developed for professional terms in the insurance field. These embedding layers are pre-trained on large-scale insurance corpus to enhance the model's understanding of professional terms. And the weight distribution strategy in the attention mechanism is adjusted to make it focus more on key insurance information in user input to improve the accuracy of intent recognition. In addition, the present application can support multimodal input (such as text, pictures, files, etc.) and introduce local invariance reasoning in the model, which can maintain attention on key information during user intent recognition and maintain focus even in the face of complex dialogue processes. Furthermore, the present application can implement a strategy based on priority reasoning, which can dynamically adjust the order of recommendation tools according to the importance and urgency of user intent. Therefore, compared with the related technology that requires the user to learn to use application tools to complete the corresponding task, the present application can associate dialogue generation with application tools, that is, the target object can use dialogue to call the corresponding application tool, and rely on the preset process execution instructions of the application tool to continuously generate dialogue, avoiding the tedious tool screening and manual operation process, saving time and reducing the risk of human error, thereby effectively improving the accuracy of dialogue generation related to the task matched by the application tool, and avoiding the situation where the user is not proficient in the use of the application tool and cannot complete the task efficiently.

[0160] See also Figure 8 The present application also provides an application tool-based dialogue generation device, which can implement the above-mentioned application tool-based dialogue generation method. The device includes:

[0161] A first acquisition module 810 is configured to acquire initial conversation data of a target object in a target application, wherein the target application has candidate application tools, each candidate application tool has a candidate function, and the candidate functions of any two candidate application tools are different;

[0162] Identification module 820, used to perform intent recognition on the initial conversation data to obtain initial conversation intent data;

[0163] A detection module 830 is configured to perform task detection on the initial conversation data based on the initial conversation intention data, and determine a target task tag for the target object; wherein each target task tag is used to indicate a task function, and the task function is configured with a preset execution process;

[0164] A tool selection module 840 is configured to obtain tool attribute data of candidate application tools, and select an application tool from the candidate application tools based on the initial conversation intention data, the target task label, and the tool attribute data, to determine a target application tool, wherein the candidate function of the target application tool is the same as the task function, the candidate function is configured with a candidate execution process, and the candidate execution process of the target application tool is the same as the preset execution process;

[0165] The second acquisition module 850 is used to acquire a preset process execution instruction of the target application tool; wherein the preset process execution instruction is used to instruct the target object to execute a preset execution process;

[0166] The dialogue generation module 860 is used to generate a dialogue based on the initial dialogue intention data and the preset process execution instructions.

[0167] The specific implementation of the application tool-based conversation generation device is basically the same as the specific embodiment of the above-mentioned application tool-based conversation generation method, and will not be repeated here.

[0168] The present application also provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the aforementioned application tool-based conversation generation method. The electronic device can be any smart terminal, such as a tablet computer or an in-vehicle computer.

[0169] See also Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is shown. The electronic device includes:

[0170] The processor 910 may be implemented as a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is configured to execute relevant programs to implement the technical solutions provided in the embodiments of the present application.

[0171] The memory 920 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 920 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 920 and is called by the processor 910 to execute the application tool-based dialogue generation method of the embodiments of this application.

[0172] Input / output interface 930, used to implement information input and output;

[0173] Communication interface 940, used to implement communication interaction between this device and other devices, which can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.);

[0174] bus 950 , which transmits information between various components of the device (e.g., processor 910 , memory 920 , input / output interface 930 , and communication interface 940 );

[0175] The processor 910 , the memory 920 , the input / output interface 930 , and the communication interface 940 are connected to each other in communication within the device via a bus 950 .

[0176] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned application tool-based dialogue generation method.

[0177] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0178] The embodiments described in the embodiments of this application are intended to more clearly illustrate the technical solutions of the embodiments of this application and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0179] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0180] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0181] Those skilled in the art will appreciate that all or some of the steps in the methods, systems, and functional modules / units in the devices disclosed above may be implemented as software, firmware, hardware, or appropriate combinations thereof.

[0182] The terms "first", "second", "third", "fourth", etc. (if any) in the specification 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 data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0183] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0184] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the above-mentioned units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0185] The units described above as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0186] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0187] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes multiple instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: various media that can store programs, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0188] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. A method for generating a dialogue based on an application tool, characterized in that: The method comprises: Acquiring initial conversation data of a target object in a target application, wherein the target application is provided with candidate application tools, each of the candidate application tools has a candidate function, and the candidate functions of any two candidate application tools are different; Performing intent recognition on the initial conversation data to obtain initial conversation intention data; Performing task detection on the initial conversation data based on the initial conversation intention data to determine a target task tag for the target object; wherein each target task tag is used to indicate a task function, and the task function is configured with a preset execution process; Acquiring tool attribute data of the candidate application tool, and performing application tool selection on the candidate application tool based on the initial conversation intention data, the target task label, and the tool attribute data, to determine a target application tool, wherein the candidate function of the target application tool is the same as the task function, the candidate function is configured with a candidate execution process, and the candidate execution process of the target application tool is the same as the preset execution process; Obtaining a preset process execution instruction of the target application tool; wherein the preset process execution instruction is used to instruct the target object to execute the preset execution process; A dialogue is generated based on the initial dialogue intention data and the preset process execution instruction.

2. The method according to claim 1, characterized in that The generating of the dialogue based on the initial dialogue intention data and the preset process execution instruction includes: Determining candidate process nodes of the preset execution process; Extracting node data from the preset process execution instruction based on the candidate process node to obtain candidate node data of the candidate process node, wherein the candidate node data includes a node processing intention and node input data, and the node processing intentions of different candidate process nodes are different; Selecting a node from the candidate process node based on the initial conversation intention data and the node processing intention, determining a target process node, and using the node input data of the target process node as target input data, wherein the intention indicated by the initial conversation intention data is the same as the node processing intention; Performing data type detection on the target input data to obtain input data type; A dialogue generation is performed on the initial dialogue intention data, the target input data, and the input data type based on a preset dialogue generation model.

3. The method according to claim 2, characterized in that The input data type includes a data upload type and a data selection type, wherein the data upload type is used to indicate uploading data of a preset data type pre-configured for the target process node, and the data selection type is used to indicate selecting data from the preset task data configured for the target process node; The performing dialogue generation on the initial dialogue intention data, the target input data, and the input data type based on a preset dialogue generation model includes: If the input data type is the data upload type, obtaining node configuration parameters of the target process node, and performing a dialog generation on the initial dialog intention data, the target input data, and the node configuration parameters based on the preset dialog generation model to obtain a first generated dialog, wherein the first generated dialog is used to guide the target object to upload data of the preset data type, and the node configuration parameters are conditions that must be met for data uploaded at the target process node; If the input data type is the data selection type, the preset task data of the target process node is obtained, and a dialogue is generated for the initial dialogue intention data, the target input data and the preset task data based on the preset dialogue generation model to obtain a second generated dialogue, which is used to guide the target object to perform data selection on the preset task data.

4. The method according to claim 1, wherein The method further comprises: Obtaining the response data fed back by the target object, wherein the response data is a response file uploaded by the target object or a response text input by the target object; Sending the reply data to the calling interface of the target application tool, and obtaining data execution information of the target application tool on the reply data from the calling interface; Determining a process execution state of the target task based on the data execution information; Update the preset process execution instruction based on the process execution status to obtain the current process execution instruction; A dialogue is generated based on the current process execution instruction, the initial dialogue intention data and the response data.

5. The method according to claim 1, wherein The selecting the candidate application tools based on the initial conversation intention data, the target task label, and the tool attribute data to determine the target application tool includes: Performing multi-level intent recognition on the initial dialogue intention data and the target task label based on a preset multi-level attention hybrid model to obtain multi-level dialogue intention data; Performing data encoding on the multi-level dialogue intention data to obtain a multi-level dialogue intention vector; Performing data encoding on the tool attribute data to obtain a tool attribute vector; Calculating similarity between the multi-level dialogue intention vector and the tool attribute vector to obtain vector similarity of the candidate application tool; An application tool selection is performed on the candidate application tools based on the vector similarity to determine the target application tool.

6. The method according to claim 5, characterized in that The multi-level intention recognition is performed on the initial dialogue intention data and the target task label based on the preset multi-level attention hybrid model to obtain multi-level dialogue intention data, including: Extracting data features from the initial conversation intention data to obtain an initial conversation intention feature, wherein the initial conversation intention feature includes an initial conversation intention sub-feature, and the initial conversation intention sub-feature is provided with an initial interaction item, wherein the initial interaction item is used to indicate an item associated with a conversation interaction; Performing label feature extraction on the target task label to obtain target task features; Combining the sub-features of the initial conversation intention based on a preset number of tail items to obtain a combined conversation intention feature; Performing attention calculation based on the target task feature and the combined dialogue intention feature to obtain a combined attention feature; Perform a linear transformation on the combined attention features to obtain the multi-level dialogue intention data.

7. The method according to any one of claims 1 to 6, characterized in that The performing task detection on the initial conversation data based on the initial conversation intention data to determine the target task label of the target object includes: Extracting intention keywords from the initial conversation intention data to obtain conversation intention keywords; Obtaining candidate task keywords associated with preset candidate task tags from a preset tag library; Calculating similarity between the conversation intention keyword and the candidate task keyword to obtain keyword similarity; The target task tag of the target object is determined from the candidate task tags based on the keyword similarity.

8. A conversation generation device based on application tools, characterized in that: The device comprises: A first acquisition module is configured to acquire initial conversation data of a target object in a target application, wherein the target application is provided with candidate application tools, each of the candidate application tools has a candidate function, and the candidate functions of any two candidate application tools are different; an identification module, configured to perform intent identification on the initial conversation data to obtain initial conversation intention data; a detection module, configured to perform task detection on the initial conversation data based on the initial conversation intention data, and determine a target task tag of the target object; wherein each target task tag is used to indicate a task function, and the task function is configured with a preset execution process; a tool selection module, configured to obtain tool attribute data of the candidate application tools, and select an application tool for the candidate application tools based on the initial conversation intention data, the target task label, and the tool attribute data, to determine a target application tool, wherein the candidate function of the target application tool is the same as the task function, the candidate function is configured with a candidate execution process, and the candidate execution process of the target application tool is the same as the preset execution process; A second acquisition module is configured to acquire a preset process execution instruction of the target application tool; wherein the preset process execution instruction is used to instruct the target object to execute the preset execution process; A dialogue generation module is used to generate a dialogue based on the initial dialogue intention data and the preset process execution instruction.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images