DST-based multi-round dialogue optimization method and related device thereof

Abstract

The invention relates to a DST-based multi-round dialogue optimization method and a related device thereof, and the method comprises the steps: obtaining first dialogue information inputted by user equipment, carrying out the intention recognition of the first dialogue information, and obtaining intention information; filling at least one preset semantic slot according to the intention information, and storing attribute information of the preset semantic slot in a first table; the first table is updated according to the historical context information of the dialogue, a second table is obtained, and the second table comprises attribute information of at least one preset semantic slot; and inputting the attribute information of the preset semantic slot in the second table into a preset generation model to generate second dialogue information, and sending the second dialogue information to the user equipment. According to the scheme provided by the invention, the computing resource consumption can be reduced, the processing efficiency of the model in the dialogue state is improved, and a more stable dialogue result is output.

Description

Technical Field

[0001] The present application relates to the field of natural language processing technology, and in particular to a DST-based multi-round dialogue optimization method and related devices. Background Art

[0002] In recent years, with the rapid development of deep learning technology, multi-round dialogue systems based on large models have made breakthrough progress in the field of natural language processing. Through deep neural networks, especially large-scale pre-trained models, these systems can generate natural and fluent responses in a wide range of dialogue scenarios, significantly improving the quality of human-computer interaction.

[0003] In the related art, when processing a conversation, a large model usually takes the entire conversation history as input and generates a state list containing all slots in each conversation round, which results in high consumption of computing resources and low processing efficiency. Summary of the invention

[0004] In order to solve or partially solve the problems existing in the related technology, the present application provides a multi-round dialogue optimization method based on DST and its related devices, which can reduce the consumption of computing resources, improve the processing efficiency of the model in the dialogue state, and output more stable dialogue results.

[0005] The first aspect of the present application provides a multi-round dialogue optimization method based on DST, including obtaining first dialogue information input by a user device, performing intent recognition on the first dialogue information to obtain intent information; filling at least one preset semantic slot according to the intent information, and storing the attribute information of the preset semantic slot in a first table; updating the first table according to the historical context information of the dialogue to obtain a second table, wherein the second table contains the attribute information of at least one preset semantic slot; inputting the attribute information of the preset semantic slot in the second table into a preset generation model to generate second dialogue information, and sending the second dialogue information to the user device.

[0006] In combination with the first aspect, in a possible implementation of the first aspect, filling at least one preset semantic slot according to the intent information includes: determining a preset semantic slot matching the intent information from the preset semantic slots according to the intent information; filling the intent data corresponding to the intent information into the preset semantic slot matching the intent information;

[0007] In combination with the first aspect, in a possible implementation method of the first aspect, storing the attribute information of the preset semantic slot in a first table includes: generating a slot-value pair according to the slot name of the preset semantic slot and the intent data stored in the preset semantic slot; and storing the slot-value pair in the first table.

[0008] In combination with the first aspect, in a possible implementation of the first aspect, the updating operation on the first table according to the historical context information of the conversation to obtain the second table includes: determining the preset semantic slot in which the attribute information has changed as the first semantic slot according to the historical context information of the conversation; and updating the first table according to the number of the first semantic slots to obtain the second table, and the updating operation includes adding the slot-value pair or deleting the slot-value pair.

[0009] In combination with the first aspect, in a possible implementation of the first aspect, the updating operation on the first table according to the number of the first semantic slots includes: substituting the first dialogue information and the historical context information into a first preset function to obtain first text data, wherein the first text data includes attribute information and the number of the first semantic slots; inputting the first text data into a preset processing model to generate second text data; substituting the second text data into a second preset function to obtain operation information of the update operation, and updating the first table according to the operation information of the update operation.

[0010] In combination with the first aspect, in a possible implementation method of the first aspect, it also includes: determining whether the user has ended the conversation; if the conversation has not been ended, obtaining third conversation information input by the user device, and using the third conversation information as the first conversation information, and returning to the execution step: performing intent recognition on the first conversation information to obtain intent information.

[0011] The second aspect of the present application provides a multi-round dialogue optimization device based on DST, including an acquisition module, which is used to acquire first dialogue information input by a user device, perform intent recognition on the first dialogue information, and obtain intent information; a storage module, which is used to fill at least one preset semantic slot according to the intent information, and store the attribute information of the preset semantic slot in a first table; a processing module, which is used to update the first table according to the historical context information of the dialogue to obtain a second table, wherein the second table contains attribute information of at least one preset semantic slot; a generation module, which is used to input the attribute information of the preset semantic slot in the second table into a preset generation model, generate second dialogue information, and send the second dialogue information to the user device.

[0012] In combination with the second aspect, in a possible implementation of the second aspect, the storage module is also used to determine a preset semantic slot that matches the intent information from the preset semantic slots based on the intent information; and fill the intent data corresponding to the intent information into the preset semantic slot that matches the intent information.

[0013] In combination with the second aspect, in a possible implementation of the second aspect, the storage module is also used to generate a slot-value pair based on the slot name of the preset semantic slot and the intention data stored in the preset semantic slot; and store the slot-value pair in the first table.

[0014] In combination with the second aspect, in a possible implementation of the second aspect, the processing module is also used to determine the preset semantic slot in which the attribute information has changed as the first semantic slot based on the historical context information of the conversation; and according to the number of the first semantic slots, perform an update operation on the first table to obtain the second table, and the update operation includes adding the slot-value pair or deleting the slot-value pair.

[0015] In combination with the second aspect, in a possible implementation of the second aspect, the processing module is also used to substitute the first dialogue information and the historical context information into a first preset function to obtain first text data, wherein the first text data includes attribute information and quantity of the first semantic slot; input the first text data into a preset processing model to generate second text data; substitute the second text data into a second preset function to obtain operation information of the update operation, and update the first table according to the operation information of the update operation.

[0016] In combination with the second aspect, in a possible implementation of the second aspect, the processing module is also used to determine whether the user has ended the conversation. If the conversation is not ended, the third conversation information input by the user device is obtained, and the third conversation information is used as the first conversation information, and the execution step is returned: performing intent recognition on the first conversation information to obtain intent information.

[0017] A third aspect of the present application provides an electronic device, including:

[0018] Processor; and

[0019] The memory stores executable codes thereon, and when the executable codes are executed by the processor, the processor is caused to execute the method as described above.

[0020] A fourth aspect of the present application provides a computer-readable storage medium having executable code stored thereon. When the executable code is executed by a processor of an electronic device, the processor is caused to execute the method as described above.

[0021] A fifth aspect of the present application provides a computer program product, comprising a computer program / instruction, which implements the method described above when executed by a processor.

[0022] The technical solution provided by this application may have the following beneficial effects:

[0023] The present application discloses a multi-round dialogue optimization method based on DST and a related device, comprising: obtaining first dialogue information input by a user device, performing intent recognition on the first dialogue information, and obtaining intent information; filling at least one preset semantic slot according to the intent information, and storing the attribute information of the preset semantic slot in a first table; updating the first table according to the historical context information of the dialogue, and obtaining a second table, wherein the second table contains the attribute information of at least one preset semantic slot; inputting the attribute information of the preset semantic slot in the second table into a preset generation model, generating second dialogue information, and sending the second dialogue information to the user device, which can reduce the consumption of computing resources, improve the processing efficiency of the model in the dialogue state, and output a more stable dialogue result.

[0024] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The above and other objects, features and advantages of the present application will become more apparent by describing in more detail exemplary embodiments of the present application in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the present application.

[0026] Figure 1 It is a flowchart of a multi-round dialogue optimization method based on DST shown in an embodiment of the present application;

[0027] Figure 2 is a structural diagram of a multi-round dialogue optimization device based on DST shown in an embodiment of the present application;

[0028] Figure 3 It is a schematic diagram of the structure of an electronic device shown in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

[0030] The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The singular forms of "a", "a", "an" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in this article refers to and includes any or all possible combinations of one or more associated listed items.

[0031] It should be understood that although the terms "first", "second", "third", etc. may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0032] In response to the above problems, an embodiment of the present application provides a multi-round dialogue optimization method based on DST, which can reduce the consumption of computing resources, improve the processing efficiency of the model in the dialogue state, and output more stable dialogue results.

[0033] The technical solution of the embodiments of the present application is described in detail below with reference to the accompanying drawings.

[0034] Figure 1 It is a flowchart of a multi-round dialogue optimization method based on DST shown in an embodiment of the present application.

[0035] See also Figure 1 , a multi-round dialogue optimization method based on DST, comprising:

[0036] S110: Acquire first dialogue information input by a user device, perform intent recognition on the first dialogue information, and obtain intent information.

[0037] Specifically, the user can input the first dialogue information on the user device, and then perform intent recognition on the first dialogue information input by the user to obtain the user's intent information. The purpose of intent recognition is to understand the essential purpose of the user's request, such as checking the weather, booking a restaurant, or playing music. Through the dialogue state tracking model (such as LSTM neural network model, statistical model, etc.), the key information in the dialogue is extracted to identify potential intent needs. For example, natural language processing (NLP) technology and machine learning models, such as recurrent neural network (RNN), long short-term memory network (LSTM) or more advanced Transformer architecture, can be used to classify text or converted voice text to determine the user's intent category; semantic slots refer to key information units required to implement specific functions, such as "date", "number of people" and "time" in restaurant reservations.

[0038] Specifically, the execution entity and user device can be a smart phone, computer, smart speaker or any other terminal device that can interact with the user; the first conversation information can be data in various forms, such as text information entered by the user through the keyboard, voice commands issued by the user through the microphone, which are then converted into processable text form, or in some application scenarios, the user may upload images or videos as part of the conversation, and the system converts them into descriptive text; the received raw input may need to be preprocessed, such as speech recognition, image recognition or natural language processing, to convert it into a form that the system can understand and further process.

[0039] S120: Fill at least one preset semantic slot according to the intent information, and store attribute information of the preset semantic slot in a first table.

[0040] Specifically, the user's goal and intention can be extracted from the first dialogue information input by the user, and the relevant preset semantic slots can be obtained and filled into the preset semantic slots. This information can be formalized into a dialogue state for subsequent dialogue strategy selection and system action generation. This process DST (Dialogue State Tracking) is a key part of the task-based dialogue system. Its main purpose is to predict and maintain the current dialogue state based on the dialogue history information and obtain the user's real purpose based on the dialogue state; preset semantic slots indicate that these slots are pre-defined based on the expectation of possible user intentions and scenarios. For example, in a restaurant reservation system, possible preset semantic slots include "number of people", "date", "time" and "special requirements". In the system development stage, based on domain analysis and user needs, the designer defines a series of semantic slots, each of which corresponds to a type of information. After filling, the attribute information of the preset semantic slot can include the slot name of the semantic slot and the corresponding slot value. The slot name of the semantic slot can be composed of a domain and a slot name, separated by a hyphen (-), such as restaurant-area. The first table may be a two-column table, the first column being used to store slot names of semantic slots, and the second column being used to store corresponding slot values.

[0041] In a possible implementation, a preset semantic slot matching the intent information is determined from preset semantic slots according to the intent information; and intent data corresponding to the intent information is filled into the preset semantic slot matching the intent information.

[0042] Specifically, semantic slots are structured units used to store and organize key information in user requests in a dialogue system. Each slot usually represents a specific type of information, such as time, location, quantity, preference, etc. This information is essential to completing the user's request. Semantic slots help the system clearly understand what information is needed to meet the user's specific needs. They also serve as a bridge for interaction between the system and the user, guiding the user to gradually provide the missing information. At the beginning of the dialogue, the dialogue state and the first table are initialized, all preset semantic slots are set to specific initial values, and the intent information in the first dialogue information is obtained. From the identified intent information, specific entities or attribute values ​​can be extracted in the user dialogue, and the preset semantic slots of the attributes or categories related to the intent information are determined, and the intent data corresponding to the intent information is input into the semantic slot.

[0043] In a possible implementation, a slot-value pair is generated according to a slot name of a preset semantic slot and the intent data stored in the preset semantic slot; and the slot-value pair is stored in a first table.

[0044] Specifically, after the intent data corresponding to the intent information is filled into the preset semantic slot, a slot-value pair can be generated according to the slot name of the preset semantic slot and the intent data, and the slot-value pair can be stored in the first table.

[0045] S130: performing an update operation on the first table according to the historical context information of the conversation to obtain a second table, where the second table includes attribute information of at least one preset semantic slot.

[0046] Specifically, the historical context information includes previous conversation records, semantic slots, etc., which are used to ensure that the generated conversation information can take into account the information in the previous conversation to make the conversation more natural and coherent. According to the historical context information, the changed semantic slots can be obtained, and the slots can be added, deleted, and updated according to the historical context information, and the slot values ​​of the current state can be stored, and the first table can be updated, such as adding attribute information of the new semantic slot.

[0047] In one possible implementation, based on historical context information of the conversation, a preset semantic slot whose attribute information has changed is determined as a first semantic slot; based on the number of the first semantic slots, an update operation is performed on the first table to obtain a second table, and the update operation includes adding a slot-value pair or deleting a slot-value pair.

[0048] Specifically, based on the historical context information of the conversation, the number of preset semantic slots whose attribute information has changed can be obtained, and new slot-value pairs can be added to the first table or existing slot-value pairs can be deleted to obtain the second table.

[0049] In one possible implementation, the first conversation information and historical context information are substituted into a first preset function to obtain first text data, the first text data including attribute information and quantity of a first semantic slot; the first text data is input into a preset processing model to generate second text data; the second text data is substituted into a second preset function to obtain operation information of an update operation, and the first table is updated according to the operation information of the update operation.

[0050] Specifically, the preset processing model can be a sequence-to-sequence model, and the preset processing model can be used to take the entire conversation history as input to generate attribute information of all semantic slots. The model can receive the historical context information of the conversation and generate attribute information of all semantic slots through the first preset function τ c Convert the first dialogue information and the historical context information into the first text data τ c (C t ), so that the preset processing model can understand and process the first text data, and the sequence-to-sequence model is f θ By inputting the first text data into the preset processing model, the second text data τ can be generated s (O t ), the second text data is input into the second preset function τ s In the example, the operation information of the update operation can be obtained. The update operation can be to add a new slot-value pair or delete a slot-value pair in the first table. The whole process can be expressed as:

[0051] τ s (O t ) = f θ (τ c (C t ))

[0052] Among them, f θ represents the preset processing model, τ c (C t ) is the first text data, τ s (O t ) is the second text data. Further, the first text data and the second text data can be generated simultaneously, that is, the slot value of the preset semantic slot and the operation information of the update operation can be generated simultaneously through the preset processing model.

[0053] S140: Inputting the attribute information of the preset semantic slot in the second table into the preset generation model, generating second dialogue information, and sending the second dialogue information to the user device.

[0054] Specifically, the attribute information of the preset semantic slot in the second table can be linearized, each row in the second table can be converted into a slot-value pair, and then converted into a string using the template {slot}={value}. All slot-value pairs are connected together using a semicolon as a separator to form a continuous string, which is input into the preset generation model. The preset generation model can generate a second dialogue message and send it to the user device. Furthermore, the preset generation model receives the slot value and historical context information of the preset semantic slot from the second table as input, and outputs the second dialogue message after a series of complex calculations. The second dialogue message is a sentence of dialogue information generated in combination with the understanding of the user's intention, i.e., the second dialogue message.

[0055] In a possible implementation, determine whether the user ends the conversation. If the conversation is not ended, obtain third conversation information input by the user device, and use the third conversation information as the first conversation information, and return to the execution step: perform intent recognition on the first conversation information to obtain intent information.

[0056] Specifically, when the user has not ended the conversation, for example, the user has not clicked the end conversation button, the third conversation information input by the user from the user device can be received. After receiving the third conversation information, it is not regarded as an independent input, but as the starting point of a new round of conversation, that is, it is regarded as the first conversation information to continue to advance the conversation process.

[0057] In summary, by saving the entire process of the dialogue in the form of slot-value pairs in a table and storing them in a table structure, the dialogue state tracking task is formalized as a table operation task. This simplifies the representation and update process of the dialogue state, making the state update more intuitive and easy to manage. Compared with the traditional sequence-to-sequence method, this method reduces the input and output complexity of the model, thereby improving the overall efficiency of the system. Through the form of table operations, the preset generation model can be more resistant to noise in the training data, improve the stability and reliability of the preset generation model in practical applications, and reduce performance degradation caused by data quality issues.

[0058] The present application discloses a multi-round dialogue optimization method based on DST, comprising: obtaining first dialogue information input by a user device, performing intent recognition on the first dialogue information, and obtaining intent information; filling at least one preset semantic slot according to the intent information, and storing the attribute information of the preset semantic slot in a first table; updating the first table according to the historical context information of the dialogue, and obtaining a second table, wherein the second table contains the attribute information of at least one preset semantic slot; inputting the attribute information of the preset semantic slot in the second table into a preset generation model, generating second dialogue information, and sending the second dialogue information to the user device, which can reduce the consumption of computing resources, improve the processing efficiency of the model in the dialogue state, and output a more stable dialogue result.

[0059] Corresponding to the aforementioned application function implementation method embodiment, the present application also provides a DST-based multi-round dialogue optimization device, electronic device and corresponding embodiments.

[0060] Figure 2 It is a structural diagram of a multi-round dialogue optimization device based on DST shown in an embodiment of the present application.

[0061] See also Figure 2 , a multi-round dialogue optimization device 200 based on DST, comprising:

[0062] The acquisition module 210 is used to acquire first dialogue information input by a user device, perform intent recognition on the first dialogue information, and obtain intent information.

[0063] Specifically, the user can input the first conversation information on the user device, and then perform intent recognition on the first conversation information input by the user to obtain the user's intent information. The purpose of intent recognition is to understand the essential purpose of the user's request, such as checking the weather, booking a restaurant, or playing music.

[0064] The storage module 220 is used to fill at least one preset semantic slot according to the intention information, and store the attribute information of the preset semantic slot in the first table.

[0065] Specifically, the user's goal and intention can be extracted from the first dialogue information input by the user, and the relevant preset semantic slots can be obtained and filled into the preset semantic slots. This information can be formalized into a dialogue state for subsequent dialogue strategy selection and system action generation. This process DST (Dialogue State Tracking) is a key part of the task-based dialogue system. Its main purpose is to predict and maintain the current dialogue state based on the dialogue history information and obtain the user's real purpose based on the dialogue state; preset semantic slots indicate that these slots are pre-defined based on the expectation of possible user intentions and scenarios. For example, in a restaurant reservation system, possible preset semantic slots include "number of people", "date", "time" and "special requirements". In the system development stage, based on domain analysis and user needs, the designer defines a series of semantic slots, each of which corresponds to a type of information. After filling, the attribute information of the preset semantic slot can include the slot name of the semantic slot and the corresponding slot value. The slot name of the semantic slot can be composed of a domain and a slot name, separated by a hyphen (-), such as restaurant-area. The first table may be a two-column table, the first column being used to store slot names of semantic slots, and the second column being used to store corresponding slot values.

[0066] In a possible implementation, the storage module 220 is further used to determine a preset semantic slot that matches the intent information from the preset semantic slots according to the intent information; and fill the intent data corresponding to the intent information into the preset semantic slot that matches the intent information.

[0067] In a possible implementation, the storage module 220 is further used to generate a slot-value pair according to the slot name of the preset semantic slot and the intention data stored in the preset semantic slot; and store the slot-value pair in the first table.

[0068] The processing module 230 is used to update the first table according to the historical context information of the conversation to obtain a second table, where the second table includes attribute information of at least one preset semantic slot.

[0069] Specifically, the historical context information includes previous conversation records, semantic slots, etc., which are used to ensure that the generated conversation information can take into account the information in the previous conversation to make the conversation more natural and coherent. According to the historical context information, the changed semantic slots can be obtained, and the slots can be added, deleted, and updated according to the historical context information, and the slot values ​​of the current state can be stored, and the first table can be updated, such as adding attribute information of the new semantic slot.

[0070] In one possible implementation, the processing module 230 is also used to substitute the first conversation information and historical context information into a first preset function to obtain first text data, the first text data including attribute information and quantity of a first semantic slot; input the first text data into a preset processing model to generate second text data; substitute the second text data into a second preset function to obtain operation information of an update operation, and perform an update operation on the first table according to the operation information of the update operation.

[0071] In a possible implementation, the processing module 230 is also used to determine whether the user has ended the conversation. If the conversation has not been ended, the third conversation information input by the user device is obtained, and the third conversation information is used as the first conversation information, and the execution step is returned to: performing intent recognition on the first conversation information to obtain intent information.

[0072] The generation module 240 is used to input the attribute information of the preset semantic slot in the second table into the preset generation model, generate the second dialogue information, and send the second dialogue information to the user device.

[0073] Specifically, the attribute information of the preset semantic slot in the second table can be linearized, each row in the second table can be converted into a slot-value pair, and then converted into a string using the template {slot}={value}. All slot-value pairs are connected together using a semicolon as a separator to form a continuous string, which is input into the preset generation model. The preset generation model can generate a second dialogue message and send it to the user device. Furthermore, the preset generation model receives the slot value and historical context information of the preset semantic slot from the second table as input, and outputs the second dialogue message after a series of complex calculations. The second dialogue message is a sentence of dialogue information generated in combination with the understanding of the user's intention, i.e., the second dialogue message.

[0074] The present application discloses a multi-round dialogue optimization device based on DST, comprising: obtaining first dialogue information input by a user device, performing intent recognition on the first dialogue information, and obtaining intent information; filling at least one preset semantic slot according to the intent information, and storing the attribute information of the preset semantic slot in a first table; updating the first table according to the historical context information of the dialogue, and obtaining a second table, wherein the second table contains the attribute information of at least one preset semantic slot; inputting the attribute information of the preset semantic slot in the second table into a preset generation model, generating second dialogue information, and sending the second dialogue information to the user device, which can reduce the consumption of computing resources, improve the processing efficiency of the model in the dialogue state, and output a more stable dialogue result.

[0075] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated again here.

[0076] Figure 3 It is a schematic diagram of the structure of an electronic device shown in an embodiment of the present application.

[0077] See also Figure 3 , the electronic device 300 includes a memory 310 and a processor 320.

[0078] The processor 320 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0079] The memory 310 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. Among them, ROM can store static data or instructions required by the processor 320 or other modules of the computer. The permanent storage device may be a readable and writable storage device. The permanent storage device may be a non-volatile storage device that does not lose the stored instructions and data even after the computer is powered off. In some embodiments, the permanent storage device uses a large-capacity storage device (such as a magnetic or optical disk, flash memory) as a permanent storage device. In some other embodiments, the permanent storage device may be a removable storage device (such as a floppy disk, optical drive). The system memory may be a readable and writable storage device or a volatile readable and writable storage device, such as a dynamic random access memory. The system memory may store some or all instructions and data required by the processor at run time. In addition, the memory 310 may include any combination of computer-readable storage media, including various types of semiconductor storage chips (such as DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, the memory 310 may include a readable and / or writable removable storage device, such as a laser disc (CD), a read-only digital versatile disc (such as a DVD-ROM, a double-layer DVD-ROM), a read-only Blu-ray disc, an ultra-density optical disc, a flash memory card (such as an SD card, a mini SD card, a Micro-SD card, etc.), a magnetic floppy disk, etc. Computer-readable storage media do not include carrier waves and transient electronic signals transmitted wirelessly or wired.

[0080] The memory 310 stores executable codes, and when the executable codes are processed by the processor 320 , the processor 320 can execute part or all of the methods described above.

[0081] In addition, the method according to the present application may also be implemented as a computer program or a computer program product, which includes computer program code instructions for executing some or all of the steps in the above method of the present application.

[0082] Alternatively, the present application can also be implemented as a computer-readable storage medium (or non-transitory machine-readable storage medium or machine-readable storage medium) on which executable code (or computer program or computer instruction code) is stored. When the executable code (or computer program or computer instruction code) is executed by a processor of an electronic device (or server, etc.), the processor executes part or all of the steps of the above-mentioned method according to the present application.

[0083] The embodiments of the present application have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A multi-round dialogue optimization method based on DST, characterized in that: include: Acquire first dialogue information input by a user device, perform intent recognition on the first dialogue information, and obtain intent information; Filling at least one preset semantic slot according to the intent information, and storing attribute information of the preset semantic slot in a first table; Performing an update operation on the first table according to the historical context information of the conversation to obtain a second table, wherein the second table includes attribute information of at least one of the preset semantic slots; The attribute information of the preset semantic slot in the second table is input into a preset generation model to generate second dialogue information, and the second dialogue information is sent to the user device.

2. The method according to claim 1, characterized in that Filling at least one preset semantic slot according to the intent information includes: Determining, from the preset semantic slots, a preset semantic slot matching the intent information according to the intent information; Fill the intent data corresponding to the intent information into the preset semantic slot matching the intent information.

3. The method according to claim 2, characterized in that The storing the attribute information of the preset semantic slot in the first table includes: Generate a slot-value pair according to the slot name of the preset semantic slot and the intention data stored in the preset semantic slot; The slot-value pairs are stored in the first table.

4. The method according to claim 3, characterized in that The updating operation on the first table according to the historical context information of the conversation to obtain the second table includes: According to the historical context information of the conversation, determining the preset semantic slot in which the attribute information has changed as a first semantic slot; According to the number of the first semantic slots, an update operation is performed on the first table to obtain the second table, wherein the update operation includes adding the slot-value pairs or deleting the slot-value pairs.

5. The method according to claim 4, characterized in that The updating operation on the first table according to the number of the first semantic slots includes: Substituting the first dialogue information and the historical context information into a first preset function to obtain first text data, wherein the first text data includes attribute information and quantity of the first semantic slot; Inputting the first text data into a preset processing model to generate second text data; The second text data is substituted into a second preset function to obtain operation information of the update operation, and the first table is updated according to the operation information of the update operation.

6. The method according to claim 1, characterized in that Also includes: Determine whether the user ends the conversation. If the conversation is not ended, obtain the third conversation information input by the user device, and use the third conversation information as the first conversation information, and return to the execution step: perform intent recognition on the first conversation information to obtain intent information.

7. A multi-round dialogue optimization device based on DST, characterized in that: include: An acquisition module, used to acquire first dialogue information input by a user device, perform intent recognition on the first dialogue information, and obtain intent information; A storage module, configured to fill at least one preset semantic slot according to the intent information, and store attribute information of the preset semantic slot in a first table; A processing module, configured to update the first table according to the historical context information of the conversation to obtain a second table, wherein the second table includes attribute information of at least one of the preset semantic slots; A generation module is used to input the attribute information of the preset semantic slot in the second table into a preset generation model, generate second dialogue information, and send the second dialogue information to the user device.

8. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. An electronic device, characterized in that: include: processor; as well as A memory having executable codes stored thereon, which, when executed by the processor, causes the processor to execute the method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: An executable code is stored thereon, and when the executable code is executed by a processor of an electronic device, the processor is caused to execute the method as claimed in any one of claims 1 to 6.

Images