Conversation interaction method, electronic equipment and computer readable storage medium

Abstract

The invention provides a dialogue interaction method, electronic equipment and a computer readable storage medium, and relates to the technical field of man-machine interaction. Historical dialogue records are screened, and the associated historical dialogue records are selected to perform inquiry rewriting on inquiry information; according to the method, the influence of historical dialogue records of irrelevant topics on the inquiry information can be reduced, the accuracy of the inquiry information is improved, the accuracy of reply data corresponding to the inquiry information is improved, the accuracy of interaction is improved, meanwhile, the associated historical dialogue records are selected to carry out inquiry rewriting on the inquiry information, and the interaction efficiency is improved. The input of simplified inquiry information of a user can be supported in a multi-round interaction process.

Description

Technical Field

[0001] The present application relates to the field of human-computer interaction technology, and in particular to a conversation interaction method, an electronic device, and a computer-readable storage medium. Background Art

[0002] With the development of artificial intelligence technology, interactive systems have been applied in more and more electronic devices, such as mobile phones, tablet computers, intelligent robots, etc. Interactive systems provide users with a way to interact with machines. The dialogue between the interactive system and the user can be divided into single-round dialogue and multi-round dialogue. For single-round dialogue, it is usually in the form of one question and one answer. The user asks a question or makes a request, and the system recognizes the user's intention and makes an answer or performs a specific operation. For multi-round dialogue, it is usually in the form of multiple questions and multiple answers. Summary of the invention

[0003] The present application provides a conversation interaction method, an electronic device, and a computer-readable storage medium, which can improve the accuracy of conversation interaction.

[0004] In order to achieve the above objectives, this application adopts the following technical solutions:

[0005] In a first aspect, a conversation interaction method is provided, which is applied to a first device, and the method includes: obtaining first inquiry information and a target historical conversation record, where the target historical conversation record is a historical conversation record associated with the first inquiry information in multiple rounds of historical conversation records; rewriting the inquiry according to the target historical conversation record and the first inquiry information to obtain second inquiry information; and obtaining reply data of the second inquiry information according to the second inquiry information.

[0006] Exemplarily, the rewriting of the first inquiry information may include subject completion and pronoun replacement.

[0007] Exemplarily, a search engine may be called to perform information query according to the second query information to obtain query data, and the query data may be sorted to obtain reply data for the second query information.

[0008] Among them, information organization of query data may include content extraction and integration of the extracted content. In this way, by extracting and integrating relevant content, on the one hand, duplicate content can be eliminated and the sentence can be ensured to be smooth, and on the other hand, users can also obtain accurate query results.

[0009] Exemplarily, the reply data of the second inquiry information includes the source address of each content, so that when the user needs to query more detailed information, he can query the relevant web page according to the source address.

[0010] The solution provided in the first aspect above improves the accuracy of the inquiry information by obtaining historical conversation records associated with the inquiry information and rewriting the inquiry based on the associated historical conversation records and the inquiry information, thereby improving the accuracy of the response data corresponding to the inquiry information and improving the accuracy of the interaction.

[0011] As a possible implementation, the target historical conversation record is a historical conversation record in multiple rounds of historical conversation records whose similarity with the first inquiry information meets a set threshold. In this way, by screening the historical conversation records according to the similarity, the impact of irrelevant historical conversation records on subsequent processes can be reduced, and the accuracy of the conversation interaction can be improved.

[0012] As a possible implementation, after obtaining the reply data of the second inquiry information according to the second inquiry information, the method further includes: sending the first inquiry result to the second device based on the reply data of the second inquiry information; or displaying the first inquiry result through the first result interface, the first inquiry result including the reply data of the second inquiry information. In this way, the user can obtain the inquiry result corresponding to the first inquiry information and realize the dialogue interaction.

[0013] As a possible implementation mode, after sending the first inquiry result to the second device, the method also includes: receiving an inquiry instruction for the first inquiry information from the second device, and obtaining reply data of the first inquiry information according to the first inquiry information; based on the reply data of the first inquiry information, sending a second inquiry result to the second device; after displaying the first inquiry result through the first result interface, the method also includes: in response to the inquiry operation for the first inquiry information, obtaining reply data of the first inquiry information according to the first inquiry information; and displaying the second inquiry result through the second result interface, the second inquiry result including the reply data of the first inquiry information.

[0014] Exemplarily, the first result interface may include a query control for the first query information, and a query information only for the first query information is initiated through the query control. In this way, by providing the user with a query control for the first query information while displaying the first query result, the content required by the user can be obtained more accurately, providing the user with a more accurate and efficient interactive experience.

[0015] As a possible implementation, before obtaining the first query information and the target historical conversation record, the method further includes: responding to a conversation fusion search instruction for a target historical conversation record in multiple rounds of historical conversation records, performing a query recommendation prediction based on the target historical conversation record to obtain predicted query information, wherein the target historical conversation record is a historical conversation record selected from multiple rounds of historical conversation records in response to a user's conversation selection operation; obtaining the first query information includes: obtaining the first query information obtained according to the predicted query information. In this way, by providing the user with a conversation record selection function, the user can be supported to arbitrarily select multiple conversation history records for fusion search, thereby realizing user conversation interaction across topics, records, and time, and improving the flexibility and diversity of conversation interaction.

[0016] As a possible implementation, obtaining the first query information obtained according to the predicted query information includes: sending the predicted query information to the second device, receiving the first query information obtained according to the predicted query information from the second device; or, displaying the predicted query information and the query input bar through the information display interface; and obtaining the first query information in response to a query operation on the information display interface for the predicted query information or in response to the query information input into the query input bar. In this way, based on the historical conversation records selected by the user, the query information that the user may be interested in is predicted, and the user can select the query information according to the interest point and search, thereby realizing a multi-round interactive experience across topics, records, and time.

[0017] As a possibility, after obtaining the reply data of the second query information according to the second query information, the method further includes: performing query recommendation prediction according to the second query information and the target historical dialogue record to obtain recommended query information; the first query result includes the recommended query information and the reply data of the second query information. In this way, by providing the user with the query prediction function, the user's participation and operability in the interaction process can be improved, the user's unnecessary input can be reduced, and more possibilities can be expanded for the dialogue interaction system.

[0018] In a second aspect, a conversation interaction method is provided, which is applied to a second device, and the method includes: receiving a first inquiry information input by a user; sending the first inquiry information and a target historical conversation record to the first device; the target historical conversation record is a historical conversation record associated with the first inquiry information in multiple rounds of historical conversation records; receiving a first inquiry result from the first device, the first inquiry result includes reply data of a second inquiry information, and the second inquiry information is the inquiry information obtained after the first device inquires and rewrites the first inquiry information according to the target historical conversation record; and displaying the first inquiry result through a first result interface.

[0019] The solution provided in the second aspect above improves the accuracy of the inquiry information by obtaining historical conversation records associated with the inquiry information and rewriting the inquiry based on the associated historical conversation records and the inquiry information, thereby improving the accuracy of the response data corresponding to the inquiry information and improving the accuracy of the interaction.

[0020] As a possible implementation, the target historical conversation record is a historical conversation record in multiple rounds of historical conversation records whose similarity with the first inquiry information meets a set threshold. In this way, by screening the historical conversation records according to the similarity, the impact of irrelevant historical conversation records on subsequent processes can be reduced, and the accuracy of the conversation interaction can be improved.

[0021] As a possible implementation, after displaying the first inquiry result through the first result interface, the method also includes: in response to the user's inquiry operation on the first inquiry information, sending an inquiry instruction for the first inquiry information to the first device; receiving a second inquiry result from the first device, the second inquiry result including reply data of the first inquiry information; and displaying the second inquiry result through the second result interface.

[0022] Exemplarily, the first result interface displayed by the second device may include a query control for the first query information, and a query information only for the first query information is initiated through the query control. In this way, by providing the user with a query control for the first query information while displaying the first query result, the content required by the user can be obtained more accurately, providing the user with a more accurate and efficient interactive experience.

[0023] As a possible implementation, before receiving the first query information input by the user, the method further includes: displaying a conversation record interface, the conversation record interface including multiple rounds of historical conversation records; in response to the user's conversation selection operation on the conversation record interface, obtaining a target historical conversation record in the multiple rounds of historical conversation records; in response to the user's conversation fusion search operation, sending a conversation fusion search instruction for the target historical conversation record to the first device; receiving predicted query information obtained from the first device based on the target historical conversation record; displaying the predicted query information through the information display interface, the information display interface also including a query input bar; the first query information is one of the predicted query information or the query information input through the query input bar. In this way, by providing the user with the function of selecting a conversation record, it is possible to support the user to arbitrarily select multiple conversation history records for fusion search, realize the user's conversation interaction across topics, records, and time, and improve the flexibility and diversity of conversation interaction.

[0024] As a possible implementation, the query result returned by the first device also includes recommended query information corresponding to the query information, and the first query information is one of the recommended query information included in the query result of the previous round of dialogue. In this way, by providing the user with a query prediction function, the user's participation and operability in the interaction process can be improved, unnecessary user input can be reduced, and more possibilities can be expanded for the dialogue interaction system.

[0025] As a possible implementation, before sending the first query information and the target historical conversation record to the first device, the method further includes: importing multiple rounds of historical conversation records in response to an import operation of the user. In this way, by providing a conversation record import function, discontinuous conversation interaction of the user can be achieved.

[0026] As a possible implementation, after displaying the first result interface, the method further includes: in response to the user's export operation, exporting multiple rounds of historical conversation records and the current round of conversation records, wherein the current round of conversation records includes the first inquiry information and the first inquiry result. In this way, by providing the function of exporting conversation records, it is convenient for users to review and archive historical conversation records offline, and it is convenient for users to trace back historical conversation records.

[0027] In a third aspect, an electronic device is provided, comprising: a display screen for interface display; a transceiver for sending and receiving radio signals; a memory for storing computer program instructions; and a processor for executing computer program instructions to support the electronic device to implement the method described in any possible implementation manner of the first aspect or the method described in any possible implementation manner of the second aspect.

[0028] In a fourth aspect, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processing circuit, the method described in any possible implementation of the first aspect or the method described in any possible implementation of the second aspect is implemented.

[0029] In a fifth aspect, a dialogue interaction system is provided, comprising a first device and a second device, wherein the first device comprises a first module for implementing the method described in any possible implementation manner of the first aspect, and the second device comprises a second module for implementing the method described in any possible implementation manner of the second aspect, and the first module and the second module are trained using the same model base.

[0030] In a sixth aspect, a chip system is provided, which includes a processing circuit and a storage medium, in which computer program instructions are stored; when the computer program instructions are executed by the processing circuit, the method described in any possible implementation manner in the first aspect or the method described in any possible implementation manner in the second aspect is implemented.

[0031] In a seventh aspect, a computer program product comprising instructions is provided, which, when executed on a computer, enables the computer to execute a method as described in any possible implementation of the first aspect or a method as described in any possible implementation of the second aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 A schematic diagram of the structure of a terminal device provided in an embodiment of the present application;

[0033] Figure 2 A schematic diagram of a functional module of a terminal device provided in an embodiment of the present application;

[0034] Figure 3 A schematic diagram of the architecture of a dialog interaction system provided in an embodiment of the present application;

[0035] Figure 4 One of the flowcharts of the dialogue interaction method provided in the embodiment of the present application;

[0036] Figure 5 A schematic diagram of an input interface for inquiry information provided in an embodiment of the present application;

[0037] Figure 6 One of the schematic diagrams of the result interface provided in the embodiment of the present application;

[0038] Figure 7 The second schematic diagram of the result interface provided in the embodiment of the present application;

[0039] Figure 8 The third schematic diagram of the result interface provided in the embodiment of the present application;

[0040] Fig. 9 A fourth schematic diagram of a result interface provided in an embodiment of the present application;

[0041] Fig.10 A fifth schematic diagram of a result interface provided in an embodiment of the present application;

[0042] Fig.11 The second flowchart of the dialogue interaction method provided in the embodiment of the present application;

[0043] Fig.12 One of the interaction timing diagrams of the dialogue interaction method provided in the embodiment of the present application;

[0044] Fig.13 One of the interface interaction diagrams of a multi-round dialogue provided in an embodiment of the present application;

[0045] Fig.14 One of the interface switching diagrams provided in the embodiment of the present application;

[0046] Fig.15 The third flowchart of the dialogue interaction method provided in the embodiment of the present application;

[0047] Fig.16 The second interface switching diagram provided in the embodiment of the present application;

[0048] Fig.17 The third interface switching diagram provided in the embodiment of the present application;

[0049] Fig.18 An information display interface diagram provided in an embodiment of the present application;

[0050] Fig.19 The second interactive timing diagram of the dialog interaction method provided in the embodiment of the present application;

[0051] Fig. 20 The second diagram of the interface interaction of multiple rounds of dialogue provided in the embodiment of the present application;

[0052] Fig.21 The sixth schematic diagram of the result interface provided in the embodiment of the present application;

[0053] Fig. 22 A schematic diagram of exporting historical conversation records provided in an embodiment of the present application;

[0054] Fig.23 A schematic diagram of importing historical conversation records provided in an embodiment of the present application;

[0055] Fig.24 A fourth flowchart of the dialog interaction method provided in an embodiment of the present application;

[0056] Fig.25 The fifth flowchart of the dialogue interaction method provided in the embodiment of the present application. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0058] The terms "including" and "having" and any variations thereof mentioned in the description of the embodiments of the present application are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or devices.

[0059] In the following, the terms "first", "second", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", etc. may explicitly or implicitly include one or more of the features.

[0060] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0061] In the description of the embodiments of the present application, unless otherwise specified, the meaning of "multiple" refers to two or more. The "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone.

[0062] As described in the background technology, with the development of artificial intelligence technology, interactive systems have been applied to more and more electronic devices, such as mobile phones, tablet computers, intelligent robots, etc. The interactive system provides users with an interactive way to communicate with machines, where the dialogue between the interactive system and the user can be divided into single-round dialogue and multi-round dialogue. For a single-round dialogue, it is usually in the form of one question and one answer, where the user asks a question or makes a request, and the system recognizes the user's intention and makes an answer or performs a specific operation. For a multi-round dialogue, it is usually in the form of multiple questions and multiple answers.

[0063] At present, for multi-round interaction systems, most of them use the following two methods for interaction. The first is to capture context information through a multi-information embedding mechanism and generate responses based on the context information. For example, you can first obtain multi-round dialogue history information, pre-process the multi-round dialogue history information, and pass the pre-processed multi-round dialogue history information through a multi-dimensional word-level encoder based on a multi-information embedding mechanism to obtain a sentence vector. The sentence vector is input into the discourse-level encoder to obtain a discourse-level encoding vector, and the discourse-level encoding vector is decoded by a decoder to generate a response. The second is to realize multi-round interaction by back-calculating the historical dialogue through multiple cascade functions. For example, according to the user's demand information, call the target function corresponding to the user's demand, and then obtain the cascade association function of the target function and the trigger expression of the cascade association function.

[0064] For the first method, the system architecture is a cumulative combination of multiple technical points, including but not limited to vector encoding, decoding, attention mechanism, vector feature fusion, etc. It is relatively complex and has a high technical cost. In addition, when implementing multi-round conversations, continuous multi-round dialogue history information is used, and its preprocessing is limited to the character level. Therefore, if there are multiple irrelevant topics in the multi-round dialogue history information, it will cause inaccurate responses.

[0065] For the second method, when using multiple cascade functions to back-calculate historical conversations, once the user inputs a question that is unrelated to the historical conversation, the fixed cascade function cannot accurately determine whether the current question is related to the historical conversation, resulting in inaccurate results for multiple rounds of conversation generation.

[0066] That is, the current multi-round dialogues are mostly based on the same dialogue topic. When multiple different dialogue topics appear during the interaction, the accuracy of the interaction is poor, resulting in irrelevant answers.

[0067] In addition, the current multi-round dialogue interaction system has relatively simple functions. If multiple functions are required, different strategies need to be designed for different functions, which requires strong expert knowledge and high manpower and time costs. As a result, the current multi-round dialogue interaction system has simple functions, poor scalability, slow architecture and function iteration, and the user operation experience cannot be quickly improved.

[0068] At the same time, the current multi-round dialogue interaction system has a single form of interaction. Each round will force users to input manually, and cannot perform good context linkage. It is impossible to further ask questions and interact with content across topics, records, and time. At the same time, when users ask further questions in the current multi-round dialogue interaction system, it is easy for "subject missing" and "unclear reference" to appear for further questions, resulting in low accuracy of interaction.

[0069] Based on the above problems, an embodiment of the present application provides a conversation interaction method, which can improve the accuracy of the inquiry information by obtaining historical conversation records associated with the inquiry information, and rewriting the inquiry based on the associated historical conversation records and the inquiry information, thereby improving the accuracy of the reply data corresponding to the inquiry information and improving the accuracy of the interaction.

[0070] In addition, the conversation interaction method provided in the embodiment of the present application can support users to independently select conversation history, provide multiple rounds of interaction across records, topics, and time, with diverse interaction forms and high flexibility.

[0071] In some embodiments, the conversation interaction method can be applied to a terminal device. Among them, the terminal device can also be called an access terminal, a user unit, a user station, a mobile station, a mobile station, a remote station, a remote terminal, a mobile device, a user terminal, a terminal, a wireless communication device, a user agent or a user device. The terminal device in the embodiment of the present application can be a mobile phone, a tablet computer, a computer with a wireless transceiver function, a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal in industrial control, a wireless terminal in self-driving, a wireless terminal in remote medical, a wireless terminal in smart grid, a wireless terminal in transportation safety, a wireless terminal in smart city, a wireless terminal in smart home, a vehicle-mounted terminal, an RSU with terminal function, etc. The terminal device of the present application may also be a vehicle-mounted module, vehicle-mounted module, vehicle-mounted component, vehicle-mounted chip or vehicle-mounted unit that is built into the vehicle as one or more components or units. The vehicle can implement the dialogue interaction method provided by the present application through the built-in vehicle-mounted module, vehicle-mounted module, vehicle-mounted component, vehicle-mounted chip or vehicle-mounted unit.

[0072] For easier understanding, see Figure 1 , Figure 1 FIG. 1 shows a schematic diagram of the structure of a terminal device. Figure 1 As shown, the terminal device 100 may include a processor 110, an external memory interface 120, an internal memory 130, an antenna 141, an antenna 142, a mobile communication module 150, a wireless communication module 160, a sensor module 170, a display screen 180, and the like.

[0073] It is understood that the structure illustrated in the embodiment of the present application does not constitute a specific limitation on the terminal device 100. In other embodiments of the present application, the terminal device 100 may include more or fewer components than shown in the figure, or combine some components, or split some components, or arrange the components differently. The components shown in the figure may be implemented in hardware, software, or a combination of software and hardware.

[0074] The processor 110 may include one or more processing units, for example, the processor 110 may include an application processor (AP), a modem processor, a graphics processor (GPU), an image signal processor (ISP), a controller, a memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), etc. Different processing units may be independent devices or integrated into one or more processors.

[0075] The controller may be the nerve center and command center of the terminal device 100. The controller may generate an operation control signal according to the instruction operation code and the timing signal to complete the control of fetching and executing instructions.

[0076] The processor 110 may also be provided with a memory for storing instructions and data. In some embodiments, the memory in the processor 110 is a cache memory. The memory may store instructions or data that the processor 110 has just used or cyclically used. If the processor 110 needs to use the instruction or data again, it may be directly called from the memory. This avoids repeated access, reduces the waiting time of the processor 110, and thus improves the efficiency of the system.

[0077] The wireless communication function of the terminal device 100 can be implemented through the antenna 141, the antenna 142, the mobile communication module 150, the wireless communication module 160, the modem processor and the baseband processor.

[0078] Antenna 141 and antenna 142 are used to transmit and receive electromagnetic wave signals. Each antenna in terminal device 100 can be used to cover a single or multiple communication frequency bands. Different antennas can also be reused to improve the utilization of antennas. For example, antenna 141 can be reused as a diversity antenna for a wireless local area network. In some other embodiments, the antenna can be used in combination with a tuning switch.

[0079] The mobile communication module 150 can provide solutions for wireless communications including 2G / 3G / 4G / 5G applied to the terminal device 100. The mobile communication module 150 may include at least one filter, a switch, a power amplifier, a low noise amplifier (LNA), etc. The mobile communication module 150 can receive electromagnetic waves from the antenna 141, and filter, amplify, and process the received electromagnetic waves, and transmit them to the modulation and demodulation processor for demodulation. The mobile communication module 150 can also amplify the signal modulated by the modulation and demodulation processor, and convert it into electromagnetic waves for radiation through the antenna 141. In some embodiments, at least some of the functional modules of the mobile communication module 150 can be set in the processor 110. In some embodiments, at least some of the functional modules of the mobile communication module 150 can be set in the same device as at least some of the modules of the processor 110.

[0080] The wireless communication module 160 can provide wireless communication solutions including wireless local area networks (WLAN) (such as wireless fidelity (Wi-Fi) networks), bluetooth (BT), global navigation satellite system (GNSS), frequency modulation (FM), near field communication (NFC), infrared (IR), etc., which are applied to the terminal device 100. The wireless communication module 160 can be one or more devices integrating at least one communication processing module. The wireless communication module 160 receives electromagnetic waves via the antenna 142, modulates the frequency of the electromagnetic wave signal and performs filtering, and sends the processed signal to the processor 110. The wireless communication module 160 can also receive the signal to be sent from the processor 110, modulate the frequency of the signal, amplify it, and convert it into electromagnetic waves for radiation through the antenna 142.

[0081] In some embodiments, the antenna 141 of the terminal device 100 is coupled to the mobile communication module 150, and the antenna 142 is coupled to the wireless communication module 160, so that the terminal device 100 can communicate with the network and other devices through wireless communication technology. The wireless communication technology may include global system for mobile communications (GSM), general packet radio service (GPRS), code division multiple access (CDMA), wideband code division multiple access (WCDMA), time-division code division multiple access (TD-SCDMA), long term evolution (LTE), BT, GNSS, WLAN, NFC, FM, and / or IR technology. The GNSS may include a global positioning system (GPS), a global navigation satellite system (GLONASS), a Beidou navigation satellite system (BDS), a quasi-zenith satellite system (QZSS) and / or a satellite based augmentation system (SBAS).

[0082] The terminal device 100 implements the display function through a GPU, a display screen 180, and an application processor.

[0083] The display screen 180 is used to display images, videos, etc. In some embodiments, the terminal device 100 may include 1 or N display screens 180 , where N is a positive integer greater than 1.

[0084] The GPU and the display screen 180 may be used to support the terminal device 100 in displaying an interactive interface, wherein the interactive interface displays content for interacting with a user.

[0085] The external memory interface 120 can be used to connect an external memory card, such as a Micro SD card, to expand the storage capacity of the terminal device 100. The external memory card communicates with the processor 110 through the external memory interface 120 to implement a data storage function, such as storing music, video and other files in the external memory card.

[0086] The internal memory 130 can be used to store computer executable program codes, which include instructions. The processor 110 executes various functional applications and data processing of the terminal device 100 by running the instructions stored in the internal memory 130. The internal memory 130 may include a program storage area and a data storage area. Among them, the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc. The data storage area may store data created during the use of the terminal device 100 (such as audio data, a phone book, etc.), etc. In addition, the internal memory 130 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, a universal flash storage (UFS), etc.

[0087] Among them, the sensor module 170 may include a touch sensor 170B, a pressure sensor 170A, and the like.

[0088] The pressure sensor 170A is used to sense the pressure signal and can convert the pressure signal into an electrical signal. In some embodiments, the pressure sensor 170A can be set on the display screen 180. There are many types of pressure sensors 170A, such as resistive pressure sensors, inductive pressure sensors, capacitive pressure sensors, etc. The capacitive pressure sensor can be a parallel plate including at least two conductive materials. When a force acts on the pressure sensor 170A, the capacitance between the electrodes changes. The terminal device 100 determines the intensity of the pressure according to the change in capacitance. When a touch operation acts on the display screen 180, the terminal device 100 detects the touch operation intensity according to the pressure sensor 170A. The terminal device 100 can also calculate the touch position according to the detection signal of the pressure sensor 170A. In some embodiments, touch operations acting on the same touch position but with different touch operation intensities can correspond to different operation instructions. For example: when a touch operation with a touch operation intensity less than the first pressure threshold acts on the short message application icon, an instruction to view the short message is executed. When a touch operation with a touch operation intensity greater than or equal to the first pressure threshold acts on the short message application icon, an instruction to create a new short message is executed.

[0089] The touch sensor 170B is also called a "touch panel". The touch sensor 170B can be set on the display screen 180, and the touch sensor 170B and the display screen 180 form a touch screen, also called a "touch screen". The touch sensor 170B is used to detect touch operations acting on or near it. The touch sensor 170B can pass the detected touch operation to the application processor to determine the type of touch event and execute the instruction corresponding to the touch event type, so as to achieve interaction with the user.

[0090] Please refer to Figure 2 In order to facilitate user interaction and improve the accuracy and flexibility of interaction, the terminal device 100 provided in the embodiment of the present application may include the following software function modules: an interaction module, a data processing module, a completion and rewriting module, a question and answer generation module, and a recommendation module. Among them, the interaction module is used to perceive the query information input by the user and pass the query information to the data processing module, or to perceive the historical dialogue record selected by the user and pass the historical dialogue record to the data processing module. The data processing module is used to process the query information, determine the historical dialogue record associated with the user's query information, and integrate the query information and the associated historical dialogue record. The completion and rewriting module is used to rewrite the query information based on the associated historical dialogue record, including operations such as subject completion of the query information and elimination of the reference of the query information. The question and answer generation module is used to perform information query according to the query information, obtain the reply data corresponding to the query information, and display the reply data to the user. The recommendation module is used to perform query recommendation prediction according to the query information and the associated historical dialogue record, obtain the predicted query information, and display the predicted query information to the user.

[0091] In an embodiment of the present application, the data processing module includes a data processing model, the completion and rewriting module includes a completion and rewriting model, the question and answer generation module includes a question and answer generation model, and the recommendation module includes a recommendation model. Among them, the completion and rewriting model, the question and answer generation model, and the recommendation model can be obtained by training the same model base using scenario data of different tasks. Among them, the model base can be a generative pre-training (GPT) model, a pre-trained language model (Generative Language Model), etc.

[0092] In the embodiment of the present application, the GPT model base can be trained with scenario data of different tasks to obtain the corresponding model. For example, the GPT model base can be trained with scenario data of data processing to obtain a data processing model; the GPT model base can be trained with scenario data of completion and rewriting to obtain a completion and rewriting model; the GPT model base can be trained with scenario data of question and answer generation to obtain a question and answer generation model; the GPT model base can be trained with scenario data of inquiry recommendation to obtain a recommendation model.

[0093] It can be understood that the scenario data for data processing refers to the data used in training data processing, which may include sample query information and sample historical conversation records associated with the sample query information. By using the scenario data for data processing to train the GPT model base, the obtained data processing model can obtain the historical conversation records associated with the input query information from the user's historical conversation records after obtaining the query information input by the user.

[0094] The scenario data for completion and rewriting refers to the data used in training completion and rewriting, which may include sample query information, sample historical conversation records, and sample rewritten query information. By using the scenario data for completion and rewriting to train the GPT model base, the obtained completion and rewriting model can rewrite the query information input by the user based on the associated historical conversation records to obtain the rewritten query information.

[0095] The scenario data generated by question and answer refers to the data used when training question and answer generation, which may include sample query information, sample query data, and sample reply data. By using the scenario data generated by question and answer to train the GPT model base, the obtained question and answer generation model can organize the query data corresponding to the query information to obtain reply data. In an embodiment of the present application, the query data is organized mainly to extract the main content from the query data, generate reply data, and annotate the content source address for each content in the reply data. For example, for query information A, after performing information query, query data including web page A and web page B can be obtained, the content in web page A is extracted to obtain query result A, the content in web page B is extracted to obtain query result B, and then based on query result A and query result B, reply data is generated. The reply data also includes the source address of query result A, that is, the web page address of web page A, and includes the source address of query result B, that is, the web page address of web page B.

[0096] The scenario data for query recommendation refers to the data used in training the query information recommendation, which may include sample history records, sample query information, and sample recommendation query information. By using the scenario data for query recommendation, the GPT model base is trained to obtain a recommendation model, which can make query recommendation predictions based on query information and historical conversation records to obtain predicted query information.

[0097] Considering the limitation of computing power, in some embodiments, the conversational interaction method can also be applied to the interactive system, such as Figure 3 As shown, the interactive system provided in the embodiment of the present application may include a terminal device 100 and a server 200, and a communication connection may be established between the terminal device 100 and the server 200.

[0098] The server 200 may be a cloud server. The cloud server may run one or more application services, and the terminal device 100 may connect to the cloud server via the Internet and use one or more application services on the cloud server.

[0099] It can be understood that a cloud server can be a single server or a server cluster consisting of multiple servers.

[0100] Among them, the terminal device 100 may include an interaction module and a data processing module, and the server 200 may include a completion and rewriting module, a question and answer generation module, and a recommendation module. The terminal device 100 may sense the query information input by the user or the historical conversation record selected by the user through the interaction module, process the query information through the data processing module, determine the historical conversation record associated with the user's query information, and integrate the query information and the associated historical conversation record, and send the integrated query information and the associated historical conversation record to the server 200. The server 200 rewrites the query information on the associated historical conversation record through the completion and rewriting module, performs information query on the rewritten query information through the question and answer generation module, obtains the reply data corresponding to the query information, sends the reply data to the terminal device 100, performs query recommendation prediction on the query information and the associated historical conversation record through the recommendation module, obtains the predicted query information, and sends the predicted query information to the terminal device 100.

[0101] It can be understood that after receiving the reply data and the predicted inquiry information, the terminal device 100 can display the reply data and the predicted inquiry information to achieve dialogue interaction.

[0102] In some implementations, the terminal device 100 may include an interaction module, and the server 200 may include a data processing module, a completion and rewriting module, a question and answer generation module, and a recommendation module. It is understandable that the functions of the modules included in the terminal device 100 and the server 200 can refer to the above description, and will not be repeated here.

[0103] Compared with the existing dialogue interaction system, the embodiment of the present application uses data from different scenarios to train the same model base to obtain a data processing module, a completion and rewriting module, a question and answer generation module, and a recommendation module. The architecture is simple and clear, with strong scalability. The implementation and iteration of functional modules are simple and convenient, and it can integrate a variety of functions to enhance the interactive experience.

[0104] The following is a detailed description of the conversation interaction method provided by the embodiment of the present application by taking a conversation interaction system as an example, wherein the conversation interaction system includes a first device and a second device, the first device may be a server, and the second device may be a terminal device, and the first device includes a completion and rewriting module, a question and answer generation module, and a recommendation module, and the second device includes an interaction module and a data processing module. Figure 4 , Figure 4 The flowchart of the dialogue interaction method provided in the embodiment of the present application is shown in FIG. Figure 4 As shown, the dialogue interaction method provided in the embodiment of the present application may include steps S301 to S306.

[0105] S301: The second device receives first inquiry information input by a user.

[0106] For example, Figure 5 As shown, the second device can provide an interface to the user, which includes a query information input field and a query control, and the user can enter the first query information to be queried in the query information input field. After the user enters the first query information in the query information input field, clicks the query control, and the second device responds to the user's click operation to obtain the first query information input by the user.

[0107] In some content browsing scenarios, the interface may be the content interface browsed by the user, and a query information input bar is set in the content interface. When the user needs to query the content of interest during the browsing process, the corresponding query information can be entered in the query information input bar. In some advertising recommendation scenarios, the interface may be an advertising playback interface, and a query information input bar is set in the advertising playback interface. When the user needs to query the advertising product during the advertising process, the corresponding query information can be entered in the query information input bar. The specific settings can be based on business needs, and this application does not make specific restrictions.

[0108] In some implementations, the first query information may also be input by a user on another device and transmitted to the second device via the other device.

[0109] It can be understood that the first inquiry information may refer to the information that the user needs to query, which can be specifically determined based on the user's input, and this application does not impose any specific limitation.

[0110] It can be understood that the second device can detect the user's operation through the interaction module to obtain the first inquiry information input by the user.

[0111] S302: The second device sends the first inquiry information and the target historical conversation record to the first device.

[0112] The target historical conversation record refers to the historical conversation record associated with the first inquiry information.

[0113] In an embodiment of the present application, the second device can store the user's historical conversation records. After obtaining the first inquiry information, the data processing module can be used to process the first inquiry information and the historical conversation records to obtain the target historical conversation records associated with the first inquiry information from the historical conversation records.

[0114] It can be understood that each round of historical conversation records includes at least one inquiry information and the corresponding answer data of the inquiry information. For example, when the inquiry information input by the user is "how is the weather in the next 15 days", and the feedback answer data is "the weather in the next 15 days is XXXX", the corresponding conversation record includes the content of "how is the weather in the next 15 days and the weather in the next 15 days is XXXX".

[0115] In the embodiment of the present application, when determining the target historical conversation record associated with the first inquiry information, the similarity calculation can be performed on the first inquiry information and the stored multiple rounds of historical conversation records respectively to obtain the similarity value between the first inquiry information and each round of historical conversation records. After obtaining the similarity value between the first inquiry information and each round of historical conversation records, the historical conversation record whose similarity value meets the set threshold is determined as the target historical conversation record, that is, the target historical conversation record is the historical conversation record whose similarity with the first inquiry information in the multiple rounds of historical conversation records meets the set threshold. The set threshold can be set according to actual needs, and the embodiment of the present application does not impose specific restrictions.

[0116] It can be understood that since the data processing module includes a data processing model, and the data processing model is obtained by training the GPT model base using the scenario data of data processing, therefore, after the first inquiry information and multiple rounds of historical conversation records are input into the data processing model, the data processing model can perform feature processing on the first inquiry information and multiple rounds of historical conversation records to obtain the similarity value between the first inquiry information and the stored multiple rounds of historical conversation records, and output the target historical conversation record associated with the first inquiry information according to the similarity value.

[0117] Optionally, when determining the similarity value between the first inquiry information and multiple rounds of stored historical conversation records, the key information of the first inquiry information and the key information of each round of historical conversation records can be determined first, and then the key information of the first inquiry information and the key information of each round of historical conversation records can be matched to obtain the matching degree between the key information of the first inquiry information and the key information of each round of historical conversation records. Based on the matching degree between the key information of the first inquiry information and the key information of each round of historical conversation records, the similarity between the first inquiry information and each round of historical conversation records can be obtained.

[0118] For example, when the key information of the first inquiry information Q includes key information A1, key information A2, key information A3, and key information A4. If the key information of the historical conversation record L1 includes key information A1 and key information A5, the key information of the historical conversation record L2 includes key information A1, key information A2, key information A3, and key information A6, and the key information of the historical conversation record L3 includes key information A3 and key information A5, and the matching degree is set to the number of matched key information / the total number of key information of the first inquiry information, then the matching degree of the key information of the first inquiry information Q and the key information of the historical conversation record L1 is 1 / 4, the matching degree of the key information of the first inquiry information Q and the key information of the historical conversation record L2 is 3 / 4, and the matching degree of the key information of the first inquiry information Q and the key information of the historical conversation record L3 is 1 / 4. Then, the similarity between the first inquiry information Q and the historical conversation record L1 can be 1 / 4, the similarity between the first inquiry information Q and the historical conversation record L2 can be 3 / 4, and the similarity between the first inquiry information Q and the historical conversation record L3 can be 1 / 4.

[0119] Among them, the key information may be the topic label of the conversation. For example, for a conversation about a mobile phone, the key information may include technology, electronic products, production date, production place, etc. For another example, for a conversation about a person, the key information may include person, occupation, nationality, region, etc.

[0120] After obtaining the first inquiry information and the target historical conversation record corresponding to the first inquiry information, the first inquiry information and the target historical conversation record may be sent to the first device.

[0121] It is understandable that when the second device does not store multiple rounds of historical conversation records of the user, that is, the first inquiry information is the inquiry information of the first conversation, only the first inquiry information can be sent to the first device, and the target historical conversation record can be empty.

[0122] The embodiment of the present application can reduce the impact of irrelevant historical conversation records on subsequent processes (such as rewriting of inquiry information and inquiry recommendation) by screening historical conversation records, and can improve the accuracy of conversation interaction.

[0123] S303: The first device rewrites the first inquiry information according to the target historical conversation record to obtain second inquiry information.

[0124] The second device may send an information search instruction to the first device according to the first query information and the target historical conversation record. The information search instruction includes the first query information and the target historical conversation record, and the information search instruction is used to instruct the completion rewriting module in the first device to rewrite the first query information according to the target historical conversation record, so as to perform information query based on the rewritten first query information.

[0125] Exemplarily, after receiving the information search instruction, the first device may respond to the information search instruction by inputting the target historical conversation record and the first inquiry information into a completion and rewriting model in the completion and rewriting module, and the completion and rewriting model may perform inquiry rewriting.

[0126] In some embodiments, the second device may also directly input the first query information and the target historical conversation record into the completion and rewriting model in the completion and rewriting module.

[0127] It can be understood that since the completion and rewriting model is obtained by training the GPT model base through the scenario-based data of completion and rewriting, after the first query information and the target historical conversation record are input into the completion and rewriting model, the completion and rewriting model can rewrite the first query information based on the target historical conversation record and output the rewritten query information.

[0128] Considering that in the scenario of multiple rounds of dialogue history, users usually continue the next round of dialogue based on the previous dialogue, and in the next round of dialogue, users often omit the subject or use pronouns, etc., resulting in incomplete information expressed by the inquiry information of the next round of dialogue, which in turn causes the dialogue to answer questions irrelevantly. Therefore, in the embodiment of the present application, the rewriting of the first inquiry information can include subject completion, pronoun replacement, etc.

[0129] Among them, subject completion refers to completing the subject of the first query information when the first query information lacks a subject. Exemplarily, the subject information can be determined based on the target historical conversation record, and then the subject information can be supplemented to the first query information. For example, the target historical conversation record is "What is A's age", and the first query information is "What are the works". After the target historical conversation record and the first query information are input into the completion rewriting model, the completion rewriting model analyzes the target historical conversation record and the first query information, and then the subject information "A" can be obtained from the target historical conversation record, and then the subject information "A" is supplemented to the first query information to obtain "What are A's works", that is, the second query information is "What are A's works". For another example, the target historical conversation record includes "How old is A" and "B's hometown", and the first inquiry information is "Have you ever had any cooperation experience?" After the target historical conversation record and the first inquiry information are input into the completion and rewriting model, the completion and rewriting model analyzes the target historical conversation record and the first inquiry information, and then obtains the subject information "A" and "B" from the target historical conversation record, and then supplements the subject information "A" and "B" to the first inquiry information to obtain "Have A and B ever had any cooperation experience?", that is, the second inquiry information is "Have A and B ever had any cooperation experience?"

[0130] Pronoun replacement refers to replacing the pronouns in the first query information when a pronoun appears in the first query information. Exemplarily, the subject information can be determined based on the target historical conversation record, and the pronouns in the first query information can be replaced according to the subject information. For example, the target historical conversation record is "What is A's age", and the first query information is "What are his works". After the target historical conversation record and the first query information are input into the completion and rewriting model, the completion and rewriting model analyzes the target historical conversation record and the first query information, and then the subject information "A" can be obtained from the target historical conversation record. The pronoun in the first query information is "he", and then the subject information is used to replace the pronoun in the first query information to obtain "What are A's works", that is, the second query information is "What are A's works". For another example, the target historical conversation record includes "How old is A" and "B's hometown", and the first inquiry information is "Do they have any cooperation experience?" After the target historical conversation record and the first inquiry information are input into the completion and rewriting model, the completion and rewriting model analyzes the target historical conversation record and the first inquiry information, and then obtains the subject information "A" and "B" from the target historical conversation record. The pronoun in the first inquiry information is "they", and then replaces the subject information "A" and "B" with the pronoun in the first inquiry information to obtain "Do A and B have any cooperation experience?", that is, the second inquiry information is "Do A and B have any cooperation experience?"

[0131] It can be understood that subject completion and pronoun replacement are examples of rewriting the first query information. In some embodiments, the rewriting of the first query information may also include other methods. For example, it may also include replacing typos in the first query information, etc. Specifically, the GPT model base may be trained with corresponding scenario data according to needs. It can be understood that the completion and rewriting model obtained by training can rewrite the first query information as needed.

[0132] The embodiment of the present application can improve the completeness and accuracy of the inquiry information by rewriting the inquiry information, thereby improving the accuracy of the dialogue interaction.

[0133] S304: The first device obtains response data of the second inquiry information according to the second inquiry information.

[0134] Among them, after obtaining the second query information, the first device can call the search engine, perform information query according to the second query information, obtain query data, and then input the query data into the question and answer generation model in the question and answer generation module, and the question and answer generation model will organize the information and output the reply data of the second query information.

[0135] Taking into account that when performing information query on the second inquiry information, the query data obtained is web page data, and the web page data may contain some irrelevant content, in order to accurately hit the content that the user wants to know, in an embodiment of the present application, the query data is sorted, including content extraction on the web page data, so as to extract content related to the second inquiry information from the web page data, and generate reply data, that is, the reply data includes content related to the second inquiry information in the web page data.

[0136] Considering that in actual applications, after performing information query, there may be multiple web page data. In order to facilitate users to obtain accurate query results, in the embodiment of the present application, the query data is sorted, and the web page content of each web page data can also be integrated. Among them, when integrating and sorting the web page content of each web page data, the content of each web page data can be extracted first to extract the content related to the second query information from each web page data, and then the related content is integrated to obtain the reply data. In this way, by integrating the related content, on the one hand, duplicate content can be eliminated and the sentence can be ensured to be smooth, and on the other hand, the user can also obtain accurate query results.

[0137] In order to facilitate users to understand the source of reply data and query more detailed information, in an embodiment of the present application, the generated reply data also includes the source address of each content, that is, the web address of the web page where the content is located. In this way, when the user needs to query more detailed information, he can query the relevant web page based on the source address.

[0138] It can be understood that since the question and answer generation model is obtained by training the GPT model base through the scenario data generated by question and answer, after the second query information and the query data of the second query information are input into the question and answer generation model, the question and answer generation model can perform content extraction, integration and other operations on the query data, and output the reply data of the second query information.

[0139] It is understandable that content extraction and integration are examples of information organization of query data. In some embodiments, information organization of query data may also include other methods, for example, it may also include relevance sorting of the relevant content of each web page data, and integrating the relevant content of each web page data according to the relevance sorting. Specifically, the GPT model base may be trained with corresponding scenario data according to needs. It is understandable that the trained question-answer generation model can organize the query data according to needs.

[0140] The embodiment of the present application organizes the query data of the inquiry information through the question-answer generation model, which can eliminate redundant information and accurately hit the content that the user wants to know. By providing the source address, it is convenient for the user to query detailed information and improve the user experience.

[0141] S305: The first device sends the first inquiry result to the second device based on the reply data of the second inquiry information.

[0142] S306: The second device displays the first inquiry result through the first result interface.

[0143] After obtaining the reply data of the second inquiry information, the first device can send the first inquiry result to the second device, and the first inquiry result includes the reply data of the second inquiry information. After receiving the first inquiry result, the second device displays the first inquiry result to the user through the first result interface, so that the user can obtain the inquiry result corresponding to the first inquiry information and realize dialogue interaction.

[0144] In the embodiment of the present application, the first result interface includes an inquiry information display area, an inquiry result display area, and an address information display area. The inquiry information display area is used to display the inquiry information, the inquiry result display area is used to display the reply data, and the address information display area is used to display the source address of the content in the reply data. Figure 6 As shown, the first inquiry information input by the user is "the weather in the next 15 days", and the first inquiry result includes the reply data of the second inquiry information "the weather in the next 15 days in place A". Figure 6The inquiry information display area displays the first inquiry information, namely, "the weather in the next 15 days", and the inquiry result display area displays the reply data of the second inquiry information, namely, the reply data of "the weather in place A in the next 15 days". The content in the inquiry result display area is marked with numbers such as ①, ②, etc., where numbers such as ①, ②, etc. are the address identifiers of the source address corresponding to the content, and the source address corresponding to each address identifier is displayed in the address information display area.

[0145] In order to improve the user's participation and operability in the interaction process, reduce unnecessary user input, and expand more possibilities for the dialogue interaction system, in the embodiment of the present application, a query prediction function can also be provided to the user, that is, in the embodiment of the present application, the query result returned by the first device can also include recommended query information corresponding to the query information. Based on this, the method provided in the embodiment of the present application, after obtaining the reply data of the second query information according to the second query information, can also include:

[0146] The first device performs a query recommendation prediction based on the second query information and the target historical conversation record to obtain recommended query information.

[0147] Among them, after obtaining the second inquiry information and the target historical conversation record, the first device can input the second inquiry information and the target historical conversation record into the recommendation model in the recommendation module, and perform inquiry recommendation prediction on the target historical conversation record and the second inquiry information through the recommendation model to obtain recommended inquiry information.

[0148] It can be understood that since the recommendation model is obtained by training the GPT model base using the scenario data of query recommendation, after the second query information and the target historical conversation record are input into the recommendation model, the recommendation model can make query recommendation predictions based on the target historical conversation record and the second query information, and output the recommended query information.

[0149] Accordingly, the step of the first device sending the first inquiry result to the second device based on the reply data of the second inquiry information may include:

[0150] The first device sends the first inquiry result to the second device according to the recommended inquiry information and the reply data of the second inquiry information.

[0151] The first inquiry result may include recommended inquiry information and reply data of the second inquiry information.

[0152] It can be understood that after receiving the first query result, the second device can display the first query result to the user through the first result interface, so that the user can obtain the query result corresponding to the first query information and the recommended prediction information for the first query information, thereby realizing dialogue interaction.

[0153] In the embodiment of the present application, the first result interface may also include a recommended information display area. Figure 7 As shown, the first inquiry information input by the user is "the weather in the next 15 days", and the first inquiry result includes the reply data of the second inquiry information "the weather in the next 15 days in place A". Figure 7 The inquiry information display area displays the first inquiry information, namely, “the weather in the next 15 days”; the inquiry result display area displays the reply data of the second inquiry information, namely, “the weather in place A in the next 15 days”; the content in the inquiry result display area is marked with numbers such as ①, ②, etc., where numbers such as ①, ②, etc. are the address identifiers of the source addresses corresponding to the content; the source address corresponding to each address identifier is displayed in the address information display area; the recommended information display area displays recommended inquiry information such as “Clothing guide for place A in the next 15 days”, “Tourist attractions in place A”, etc.

[0154] After displaying the recommended query information, if the user needs to inquire about a certain recommended query information, just click on the recommended query information. The second device responds to the user's operation and obtains the target recommended query information clicked by the user. Then a new round of dialogue interaction can be carried out, and the target recommended query information and the target historical dialogue record associated with the target recommended query information are sent to the first device.

[0155] It can be understood that the target recommendation query information can be regarded as the new first query information, and the process of determining the target recommendation query information and the target historical conversation record associated with the target recommendation query information can refer to the process of determining the first query information and the target historical conversation record associated with the first query information. Accordingly, after the first device obtains the target recommendation query information and the target historical conversation record associated with the target recommendation query information, the process of processing the target recommendation query information and the target historical conversation record associated with the target recommendation query information can refer to the process of processing the first query information and the target historical conversation record associated with the first query information, which will not be described in detail here, and can refer to steps S301 to S306 for details.

[0156] It should be noted that if Figure 8 As shown, the first result interface may further include a query information input field. When the recommended query information does not contain the query information that the user is interested in, the user may also input the query information to be queried based on the query information input field.

[0157] It can be understood that the first inquiry information can be input by the user in the inquiry information input field, or can be one selected by the user from the recommended inquiry information included in the inquiry result of the previous round of dialogue.

[0158] The conversation interaction method provided in the embodiment of the present application uses historical conversation records and current inquiry information to infer the user's next round of inquiry information, forming a guided conversation, thereby improving the user's participation and operability in the interaction process, reducing the user's operations, and improving the user experience.

[0159] Considering that the displayed reply data is the reply data of the second inquiry information after rewriting the first inquiry information, in order to facilitate the user to judge the correctness of the reply data, in the embodiment of the present application, the first inquiry result can also include the second inquiry information. After receiving the first inquiry result, the second device can display the second inquiry information and the reply data of the second inquiry information through the first result interface. In this way, the user can know whether the displayed reply data is the result he wants to query. Fig. 9 As shown, Fig. 9 The inquiry information display area displays the second inquiry information, namely, “the weather in place A for the next 15 days”; the inquiry result display area displays the reply data of the second inquiry information, namely, the reply data of “the weather in place A for the next 15 days”; the content in the inquiry result display area is marked with numbers such as ①, ②, etc., where numbers such as ①, ②, etc. are address identifiers of the source addresses corresponding to the content; the source address corresponding to each address identifier is displayed in the address information display area; the recommended information display area displays recommended inquiry information such as “Clothing guide for place A for the next 15 days”, “Tourist attractions in place A”, etc.

[0160] Considering that in actual application, the reply data of the second inquiry information may not be the result desired by the user, in order to accurately obtain the content that the user wants to know, in the embodiment of the present application, the first result interface displayed by the second device may also include an inquiry control of the first inquiry information. Fig.10 As shown, in order to facilitate user operation, in some embodiments, the query control of the first query information can be displayed on the first result interface with an operation button of "search only".

[0161] When the reply data of the second inquiry information is not the result desired by the user, the user can operate the inquiry control of the first inquiry information to indicate that only the first inquiry information is to be inquired without rewriting. Fig.11 , after step S306, the dialogue interaction method may further include steps S307 to S310.

[0162] S307: In response to the user's inquiry operation on the first inquiry information, the second device sends an inquiry instruction for the first inquiry information to the first device.

[0163] Among them, when the user wants to inquire only about the first inquiry information, he can click the inquiry control of the first inquiry information on the first result interface. The second device responds to the user's inquiry operation on the first inquiry information and can send an inquiry instruction for the first inquiry information to the first device. The inquiry instruction is used to instruct the first device to only query information for the first inquiry information.

[0164] It can be understood that the second device can respond to the user's inquiry operation on the first inquiry information through the interaction module, and send an inquiry instruction for the first inquiry information to the first device.

[0165] S308: The first device obtains response data of the first inquiry information according to the first inquiry information.

[0166] Among them, when the first device receives the query instruction for the first query information from the second device, it can respond to the query instruction, call the search engine, perform information query according to the first query information, obtain the query data of the first query information, and then input the query data of the first query information into the question and answer generation model in the question and answer generation module, and the question and answer generation model organizes the information and outputs the reply data of the first query information. For details, please refer to step S304, which will not be repeated here.

[0167] S309: The first device sends a second inquiry result to the second device based on the reply data of the first inquiry information.

[0168] S310: The second device displays the second inquiry result through the second result interface.

[0169] After obtaining the reply data of the first inquiry information, the first device may send a second inquiry result to the second device. The second inquiry result includes the reply data of the first inquiry information.

[0170] Correspondingly, after obtaining the reply data of the first inquiry information, the first device may also input the first inquiry information and the target historical conversation record into the recommendation model in the recommendation module, and perform inquiry recommendation prediction on the target historical conversation record and the first inquiry information through the recommendation model to obtain the recommended inquiry information corresponding to the first inquiry information. Therefore, in the embodiment of the present application, the second inquiry result may also include the reply data of the first inquiry information and the recommended inquiry information corresponding to the first inquiry information.

[0171] After receiving the second inquiry result, the second device may display a second result interface, and show the second inquiry result to the user through the second result interface, so that the user can obtain the inquiry result corresponding to the first inquiry information and realize dialogue interaction.

[0172] It can be understood that the second result interface can also include an inquiry information display area, an inquiry result display area, an address information display area, and a recommended information display area. Take the first inquiry information input by the user as "Person A", and the first inquiry result includes the reply data of the first inquiry information "Person A" as an example for explanation. The inquiry information display area of ​​the second result interface displays the first inquiry information, namely "Person A", and the inquiry result display area displays the reply data of the first inquiry information, namely the relevant introduction of "Person A". The content in the inquiry result display area is marked with numbers such as ①, ②, etc., where ①, ②, etc. are the address identifiers of the source address corresponding to the content, and the source address corresponding to each address identifier is displayed in the address information display area, and the recommended information display area displays the recommended inquiry information "The cooperation history between Person A and Person B" and the like.

[0173] The embodiment of the present application provides the user with a query control for the first query information while displaying the first query result, so that the content required by the user can be obtained more accurately, and a more accurate and efficient interactive experience is provided to the user.

[0174] Please refer to Fig.12 , Fig.12 A schematic diagram of an interaction sequence of the dialog interaction method provided in the embodiment of the present application. Fig.12As shown, the second device receives the first query information input by the user through the interaction module, and then transmits the first query information and the collected historical conversation records of the user to the data processing module, processes the first query information and the historical conversation records through the data processing model in the data processing module, obtains the target historical conversation record associated with the first query information from the historical conversation record, and sends the target historical conversation record and the first query information to the completion and rewriting module of the first device, and the completion and rewriting model in the completion and rewriting module performs query rewriting on the first query information according to the target historical conversation record to obtain the second query information, and sends the second query information to the question and answer generation module. The question and answer generation module obtains the reply data of the second query information according to the second query information, that is, calls the search engine, performs information query according to the second query information, obtains the query data, and then inputs the query data into the question and answer generation model in the question and answer generation module, and the question and answer generation model performs information sorting and outputs the reply data of the second query information. After obtaining the reply data of the second query information, the second query information and the reply data of the second query information are returned to the second device. After obtaining the second query information, the completion and rewriting module may also send the second query information and the target historical conversation record to the recommendation module. The recommendation model in the recommendation module performs query recommendation prediction based on the second query information and the target historical conversation record, obtains the recommended query information, and returns the recommended query information to the second device. After receiving the second query information, the reply data of the second query information, and the recommended query information, the second device displays the second query information, the reply data of the second query information, and the recommended query information.

[0175] Please refer to Fig.13 , Fig.13 A schematic diagram of a multi-round dialogue interface interaction is shown. Fig.13 As shown in (a) in the figure, the query information displayed on the result interface of the second device is "Person A", the response data is "Related introduction of Person A", and the recommended query information includes "His works" and "His collaborative works with Person B". The user can select the recommended query information of interest to initiate a conversation or enter the query information to be queried in the query information input field. Fig.13 As shown in (b), the user selects the recommended query information "his works", the second device responds to the user's operation, obtains the query information "his works", and determines the associated target historical conversation record, and then sends the target historical conversation record and the query information to the first device. The first device rewrites the query according to the target historical conversation record and the query information, and obtains the rewritten query information "person A's works", obtains the reply data and the next round of recommended query information "awards received" and "his collaborative works with person B" according to the rewritten query information, and sends the rewritten query information, reply data and the next round of recommended query information to the second device. Fig.13 As shown in (c), after receiving the rewritten query information, reply data and the next round of recommended query information, the second device displays the rewritten query information, reply data and the next round of recommended query information, and displays the operation button "Search only his works". When the user operates the search only operation button, the second device responds to the user's operation and sends a query instruction only for "his works" to the first device. The first device responds to the query instruction and returns the reply data of "his works" and the corresponding recommended query information "Explanation of the works" and "Scenarios of use of his words" to the second device. Fig.13 As shown in (d), after receiving the reply data and the recommended query information of “his works”, the second device can display the reply data and the recommended query information of “his works”.

[0176] In order to facilitate viewing of reply data, in an embodiment of the present application, the second device may provide the user with a reply interface for displaying reply data. The reply interface is used to display all reply data of the inquiry information, and the result interface may only display part of the reply data of the inquiry information. When the user only views the reply data, the reply data can be viewed through the reply interface. When the user needs to view the source address of the content, recommended inquiry information, and other information, the reply data can be viewed through the result interface. For example, Fig.14 As shown, after receiving the query result returned by the first device, the second device can display a reply interface, display the reply data through the reply interface, and when the user needs to view information such as the source address, the first setting operation can be performed on the reply interface, and the second device responds to the user's first setting operation and displays the result interface. The first setting operation can be an operation such as sliding or clicking, which is not limited in the embodiment of the present application and can be set according to actual needs.

[0177] The conversation interaction method provided in the embodiment of the present application can reduce the impact of conversation records of irrelevant topics on the inquiry information by screening historical conversation records and selecting related historical conversation records to rewrite the inquiry information, thereby improving the accuracy of the inquiry information, thereby improving the accuracy of the reply data corresponding to the inquiry information and improving the accuracy of the interaction. At the same time, by selecting related historical conversation records to rewrite the inquiry information, it can support users to simplify the input of inquiry information during multiple rounds of interaction. For example, users do not need to specify the subject of the inquiry information and can use references. At the same time, by providing recommended inquiry information to users, users can conduct conversations by clicking on the recommended inquiry information, simplifying user input and operation processes, and greatly improving user experience.

[0178] In order to provide more interactive forms and facilitate cross-context linkage for users, in the embodiments of the present application, users can also be provided with the function of flexibly selecting conversation history records for fusion retrieval, so that users can actively conduct cross-topic, cross-record, and cross-context linkage interactions. Fig.15 The dialogue interaction method provided in the embodiment of the present application may include S401 to S411.

[0179] S401: The second device displays a conversation record interface.

[0180] In the process of the second device interacting with the user, the second device can store the user's historical conversation records. When the user needs to view the historical conversation records, the user can view the conversation records in the current interface. The second device responds to the user's operation and displays the conversation record interface to the user, which includes the user's multiple rounds of historical conversation records.

[0181] The current interface may be a result interface, a reply interface, etc. during the conversation interaction process, and the conversation record viewing operation may be sliding, clicking, etc., which may be set according to actual needs and is not limited in the embodiment of the present application. For different interfaces, the conversation record viewing operation may be different.

[0182] For example, Fig.16 As shown, taking the current interface as the reply interface during the conversation interaction process and the conversation record viewing operation as a swiping operation as an example, when the user needs to view the historical conversation records, he can swiping down on the current reply interface, and the second device responds to the operation and displays the conversation record interface.

[0183] In order to facilitate information viewing and increase the flexibility and operability of interaction, in some embodiments of the present application, after the second device displays the conversation record interface, the user performs a second setting operation on any historical conversation record in the conversation record interface, and the second device responds to the second setting operation and displays the result interface corresponding to the historical conversation record. The second setting operation can be set according to actual needs, such as click, long press, etc., and is not specifically limited.

[0184] It is understandable that when the second device displays the result interface, the user can jump to the conversation record interface by viewing the conversation record on the result interface.

[0185] S402: The second device obtains a target historical conversation record from multiple rounds of historical conversation records in response to a conversation selection operation of the user on the conversation record interface.

[0186] The conversation record interface includes a selection control. When the user needs to select a historical conversation record, the user clicks the selection control, and the second device responds to the user's operation and displays the conversation record selection interface. Fig.17 As shown, each historical conversation record in the conversation record selection interface has a selection box, and the user can select the historical conversation record through the selection box. It can be understood that when the user needs to select a historical conversation record, just select the selection box of the historical conversation record, and the second device responds to the user's operation to obtain the historical conversation record selected by the user. It can be understood that the historical conversation record selected by the user is the target historical conversation record.

[0187] It can be understood that in order to facilitate users to conduct cross-topic, cross-record, and cross-context linkage interactions, in an embodiment of the present application, the target historical conversation record may include at least two rounds of historical conversation records, so as to perform a fusion search on at least two rounds of historical conversation records.

[0188] S403: In response to the user's conversation fusion search operation, the second device sends a conversation fusion search instruction for the target historical conversation record to the first device.

[0189] Among them, Fig.17 As shown, the conversation record selection interface may also include a fusion search control. When the user needs to perform a cross-topic and cross-record fusion search, the user can click the fusion search control after selecting the target historical conversation record. When the second device detects that the user has clicked the fusion search control, it can respond to the user's conversation fusion search operation and send a conversation fusion search instruction for the target historical conversation record to the first device.

[0190] In an embodiment of the present application, the conversation fusion search instruction includes a target historical conversation record, and the conversation fusion search instruction is used to instruct the first device to perform a fusion search based on the target historical conversation record.

[0191] S404: The first device responds to the conversation fusion search instruction for the target historical conversation record, performs query recommendation prediction based on the target historical conversation record, obtains predicted query information, and sends the predicted query information to the second device.

[0192] Among them, after receiving the conversation fusion search instruction, the first device responds to the conversation fusion search instruction, inputs the target historical conversation record into the recommendation model in the recommendation module, and performs query recommendation prediction on the target historical conversation record through the recommendation model to obtain predicted query information.

[0193] It can be understood that since the recommendation model is obtained by training the GPT model base using the scenario data of query recommendation, after the target historical conversation record is input into the recommendation model, the recommendation model can make query recommendation predictions based on the target historical conversation record and output predicted query information.

[0194] After obtaining the prediction query information, the first device may send the prediction query information to the second device.

[0195] S405: The second device displays the predicted query information through the information display interface.

[0196] Among them, after receiving the prediction query information sent by the first device, the second device can display the prediction query information to the user through the information display interface.

[0197] like Fig.18 As shown, the information display interface includes a query information display area and a query input field. The query information display area is used for predictive query information sent by the second device, and the query input field is used to receive query information input by the user.

[0198] S406: The second device obtains first inquiry information in response to an inquiry operation on the information display interface for the predicted inquiry information or in response to inquiry information input into the inquiry input field.

[0199] Among them, after the information display interface is displayed, the user can select the prediction query information of interest from the prediction query information displayed on the information display interface according to needs, or enter the query information they want to inquire from the query input field.

[0200] It is understandable that when the user selects the predicted query information that the user is interested in, the predicted query information selected by the user can be regarded as the first query information. Correspondingly, when the user enters the query information in the query input field, the query information entered by the user can be regarded as the first query information.

[0201] S407: The second device sends the first inquiry information and the target historical conversation record to the first device.

[0202] After receiving the first query information, the second device can send the first query information and the target historical conversation record to the first device. It can be understood that the target historical conversation record in step S407 is the target historical conversation record selected by the user from the multiple rounds of historical conversation records.

[0203] S408: The first device rewrites the first inquiry information according to the target historical conversation record to obtain second inquiry information.

[0204] S409: The first device obtains response data of the second inquiry information according to the second inquiry information.

[0205] S410: The first device sends a first inquiry result to the second device based on reply data of the second inquiry information.

[0206] S411: The second device displays the first inquiry result through the first result interface.

[0207] The process of step S408 to step S411 may refer to step S303 to step S306, and will not be described in detail here.

[0208] The embodiments of the present application provide users with the function of selecting conversation records, so as to support users to arbitrarily select multiple conversation history records for fusion search, realize users' conversation interaction across topics, records and time, and improve the flexibility and diversity of conversation interaction.

[0209] Please refer to Fig.19 , Fig.19 Another interactive sequence diagram of the dialog interaction method provided in the embodiment of the present application. Fig.19 As shown, the second device receives the target historical conversation record selected by the user through the interaction module, and then transmits the target historical conversation record selected by the user to the data processing module. The data processing module integrates the target historical conversation record according to a preset format, and sends the integrated target historical conversation record to the recommendation module of the first device. The recommendation model in the recommendation module and the target historical conversation record perform query recommendation prediction to obtain predicted query information, and send the predicted query information to the second device. The second device displays the predicted query information through the information display interface, and responds to the user's query operation on the information display interface for the predicted query information or responds to the query information input into the query input field to obtain the first query information, and sends the first query information and the target historical conversation record to the completion and rewriting module of the first device. The completion and rewriting model in the completion and rewriting module performs query rewriting on the first query information according to the target historical conversation record to obtain the second query information, and sends the second query information to the question and answer generation module. The question-answer generation module obtains the reply data of the second query information according to the second query information, that is, calls the search engine, performs information query according to the second query information, obtains the query data, and then inputs the query data into the question-answer generation model in the question-answer generation module, and the question-answer generation model organizes the information and outputs the reply data of the second query information. After obtaining the reply data of the second query information, the second query information and the reply data of the second query information are returned to the second device. After obtaining the second query information, the completion and rewriting module can also send the second query information and the target historical conversation record to the recommendation module. The recommendation model in the recommendation module performs query recommendation prediction according to the second query information and the target historical conversation record, obtains the recommended query information, and returns the recommended query information to the second device. After receiving the second query information, the reply data of the second query information and the recommended query information, the second device displays the second query information, the reply data of the second query information and the recommended query information.

[0210] Please refer to Fig. 20 , Fig. 20A schematic diagram of a multi-round dialogue interface interaction is shown. Fig. 20 As shown in (a) of FIG. 1 , the second device displays the historical conversation record through the conversation record interface. The historical conversation record in the conversation record interface includes "Person 1", "Person 2", "Person 3" and a selection control. After the user clicks the selection control, the second device responds to the user's operation on the selection control and displays the conversation record selection interface, such as Fig. 20 As shown in (b), each historical conversation record in the conversation record selection interface has a selection box, and the conversation record selection interface also includes a fusion search control. The user selects "Person 1" and "Person 2" in the conversation record selection interface, and after completing the selection, clicks the fusion search control. The second device responds to the user's operation and sends the historical conversation record selected by the user to the first device. The first device performs query recommendation prediction based on the historical conversation record selected by the user, obtains predicted query information, and sends the predicted query information to the second device. After receiving the predicted query information, the second device displays the predicted query information through the information display interface, such as Fig. 20 As shown in (c), the information display interface includes the predicted query information "cooperative works", "awards received by each party", and "age difference". The user selects the query information "cooperative works" on the information display interface. In response to the user's operation, the second device sends the selected query information "cooperative works", historical conversation records "person 1", and historical conversation records "person 2" to the first device. The first device performs an information query and returns the query result. Fig. 20 As shown in (d), after the second device receives the query result returned by the first device, the conversation record interface of the second device adds the conversation record of this round, that is, the conversation record of "the work of cooperation between Person 1 and Person 2".

[0211] The embodiment of the present application can support users to flexibly select multiple historical conversation records, and predict the query information that the user may be interested in based on the historical conversation records selected by the user. The user can select the query information according to the interest point and search, so as to achieve a multi-round interactive experience across topics, records, and time. Compared with the conventional technology in which the user cannot independently control the interaction mode of the multi-round conversation information that he wants to combine, the embodiment of the present application can achieve free context linkage and accurately generate content that the user is interested in.

[0212] In order to facilitate the user to review the historical conversation records, in the embodiment of the present application, the second device can also provide the user with the function of importing and exporting the historical conversation records. Based on this, in the embodiment of the present application, after displaying the first result interface, the following steps can also be included:

[0213] (1) In response to the user's export operation, multiple rounds of historical conversation records and current round of conversation records are exported.

[0214] The conversation record of this round includes the first inquiry information and the first inquiry result. It can be understood that when the user only inquires about the first inquiry information, that is, after the user clicks the query control of searching only, the conversation record of this round includes the first inquiry information and the second inquiry result.

[0215] like Fig.21 As shown, in the embodiment of the present application, the result interface displayed by the second device may also include an import control and an export control. When the user needs to export the conversation record, the user may click the export control, and the second device responds to the user's operation to export and save the multi-round historical conversation records and the current conversation record. Fig. 22 As shown, Fig. 22 This is a schematic diagram for exporting historical conversation records. The exported schematic diagram includes the historical conversation records of person A, person B, and person C.

[0216] Accordingly, in the embodiment of the present application, before sending the first inquiry information and the target historical conversation record to the first device, the following steps may also be included:

[0217] (2) In response to the user's import operation, multiple rounds of historical conversation records are imported.

[0218] When the user needs to import the conversation record, he can also click the import control, and the second device responds to the user's operation to import multiple rounds of historical conversation records, such as Fig.23 As shown, Fig.23 This is a schematic diagram of importing historical conversation records. After the historical conversation records of person A are imported, the result interface of the second device can display the detailed information of the historical conversation records of person A.

[0219] In an optional implementation, after multiple historical conversation records are imported, the multiple historical conversation records can be scrolled or folded to display the result interface, which can be specifically set according to business needs.

[0220] The embodiment of the present application provides a conversation record export function, which can facilitate users to review and archive historical conversation records offline, and can facilitate users to trace back historical conversation records.

[0221] In an embodiment of the present application, after the user exports the conversation record, the user can also share the conversation record, that is, send the conversation record to a third device.

[0222] Accordingly, in the embodiment of the present application, the multi-round historical conversation records may also be shared from the third device to the second device. Therefore, in the embodiment of the present application, the multi-round historical conversation records may be the records stored by the second device during the current conversation interaction process, or the records shared by the third device to the second device and imported, or the records previously exported by the second device but imported during the current conversation interaction process.

[0223] The embodiment of the present application can realize non-continuous conversation interaction of users by providing a function of importing conversation records. For example, when a user performs multiple rounds of conversation interaction, the multiple rounds of conversation records can be exported and saved. In the next conversation, if the user wants to continue the conversation interaction based on the last time, the conversation record exported last time can be imported. It can be understood that after importing the conversation record, the second device realizes the conversation interaction with the user based on the imported conversation record. For example, based on the imported conversation record, the target historical conversation record associated with the inquiry information input by the user is determined.

[0224] The embodiment of the present application can greatly improve the user's discontinuous interaction experience by providing the function of importing and exporting conversation records. All conversation records can be exported and archived for easy tracing and use.

[0225] The conversation interaction method provided in the embodiment of the present application obtains historical conversation records associated with inquiry information, and rewrites inquiries based on the associated historical conversation records and inquiry information, so as to improve the accuracy of the inquiry information, thereby improving the accuracy of the reply data corresponding to the inquiry information and improving the accuracy of the interaction.

[0226] The embodiments of the present application can support users to flexibly select single or multiple historical conversation records, and predict inquiry information that the user may be interested in based on the historical conversation records selected by the user. The user can select inquiry information based on the points of interest and search for it, thereby achieving a multi-round interactive experience across topics, records, and time.

[0227] It is understandable that the conversation interaction method provided in the embodiment of the present application can also be applied to a terminal device alone. Fig.24 ,like Fig.24 As shown, the dialogue interaction method applied to a terminal device provided in an embodiment of the present application may include steps S501 to S506.

[0228] S501: Acquire first inquiry information input by a user and target historical conversation records.

[0229] The terminal device may obtain the first inquiry information input by the user by referring to step S301, which will not be described in detail here.

[0230] Among them, when the terminal device obtains the target historical conversation record, it can calculate the similarity value between each round of historical conversation records in multiple rounds of historical conversation records and the first inquiry information, and obtain the target historical conversation record in the multiple rounds of historical conversation records based on the similarity value between each round of historical conversation records in the multiple rounds of historical conversation records and the first inquiry information. The specific process can refer to step S302 and will not be repeated here.

[0231] S502: rewriting the inquiry according to the target historical conversation record and the first inquiry information to obtain second inquiry information.

[0232] S503: Obtain reply data of the second inquiry information according to the second inquiry information.

[0233] The process from step S502 to step S503 may refer to the process from step S303 to step S304, and will not be described in detail here.

[0234] S504: Displaying the first inquiry result through the first result interface.

[0235] After obtaining the reply data of the second inquiry information, the terminal device may display the first inquiry result through the first result interface. The specific process may refer to step S306.

[0236] It can be understood that the first inquiry result displayed by the terminal device may also include the reply data of the second inquiry information and the recommended inquiry information. The specific process is referred to the above description and will not be repeated here.

[0237] Accordingly, the reply data in the second inquiry information may not be the result desired by the user. In order to accurately obtain the content that the user wants to know, in the embodiment of the present application, the first result interface displayed by the terminal device may also include an inquiry control of the first inquiry information. Based on this, after step S504, the following steps may also be included:

[0238] S505: In response to the inquiry operation for the first inquiry information, obtaining reply data of the first inquiry information according to the first inquiry information.

[0239] Among them, when the user wants to inquire only about the first inquiry information, he can click the inquiry control of the first inquiry information on the first result interface, and the terminal device responds to the user's inquiry operation on the first inquiry information, calls the search engine, and performs information query according to the first inquiry information to obtain the query data of the first inquiry information, and then inputs the query data of the first inquiry information into the question and answer generation model in the question and answer generation module, and the question and answer generation model organizes the information and outputs the reply data of the first inquiry information.

[0240] S506: Displaying the second inquiry result through the second result interface.

[0241] The process of the terminal device displaying the second inquiry result through the second result interface can refer to step S310, which will not be described in detail here.

[0242] It can be understood that the second inquiry result may also include the reply data of the first inquiry information and the recommended inquiry information corresponding to the first inquiry information.

[0243] Understandably, in order to provide more interactive forms and facilitate cross-context linkage of users, the terminal device can also provide users with the flexibility to select a single or multiple conversation history records for fusion retrieval, so that users can actively conduct cross-topic, cross-record, and cross-context linkage interactions. Fig.25 The dialogue interaction method applied to the terminal device may also include S601 to S609.

[0244] S601: Displaying the conversation record interface.

[0245] In the process of the terminal device interacting with the user, the terminal device can store the user's historical conversation records. When the user needs to view the historical conversation records, the conversation record viewing operation can be performed on the current interface. The terminal device responds to the user's operation and displays the conversation record interface to the user. The conversation record interface includes multiple rounds of historical conversation records of the user. The specific process can refer to step S401, which is not repeated here.

[0246] S602: In response to a conversation selection operation of the user on the conversation record interface, a target historical conversation record in multiple rounds of historical conversation records is obtained.

[0247] The process of step S602 may refer to step S402 and will not be described in detail here.

[0248] S603: In response to a conversation fusion search instruction for a target historical conversation record in the multiple rounds of historical conversation records, query recommendation prediction is performed according to the target historical conversation record to obtain predicted query information.

[0249] The conversation record selection interface may also include a fusion search control. When the user needs to conduct a cross-topic and cross-record fusion search, after the user selects the target historical conversation record, the fusion search control can be operated to issue a conversation fusion search instruction to the terminal device. In response to the user's conversation fusion search instruction, the terminal device inputs the target historical conversation record into the recommendation model in the recommendation module, and performs a query recommendation prediction on the target historical conversation record through the recommendation model to obtain the predicted query information.

[0250] S604: Displaying the predicted inquiry information through the information display interface.

[0251] S605: In response to an inquiry operation on the information display interface for predicted inquiry information or in response to inquiry information input into the inquiry input field, first inquiry information is obtained.

[0252] Among them, the information display interface includes a query information display area and a query input bar. After the information display interface is displayed, the user can select the predicted query information of interest from the predicted query information displayed on the information display interface according to needs, or input the query information they want to inquire from the query input bar.

[0253] It can be understood that when the user selects the predicted query information of interest, the predicted query information selected by the user can be regarded as the first query information. Correspondingly, when the user enters the query information in the query input field, the query information entered by the user can be regarded as the first query information. For details, please refer to steps S405 to S406, which will not be described in detail here.

[0254] S606: Obtain the target's historical conversation records.

[0255] S607: rewriting the inquiry according to the target historical conversation record and the first inquiry information to obtain second inquiry information.

[0256] S608: Obtain reply data of the second inquiry information according to the second inquiry information.

[0257] S609: Displaying the first inquiry result through the first result interface.

[0258] The process of step S606 to step S609 may refer to step S501 to step S504, and will not be described in detail here.

[0259] The embodiments of the present application provide users with the function of selecting conversation records, thereby supporting users to arbitrarily select a single or multiple conversation history records for fusion search, thereby realizing user conversation interaction across topics, records, and time, and improving the flexibility and diversity of conversation interaction.

[0260] Based on the same inventive concept, an embodiment of the present application provides an electronic device, which may be a terminal device or a server in the above embodiment. The electronic device may specifically include a communication interface, a memory, a processor, and one or more computer programs. Among them, the communication interface, the memory, and the processor may be connected via one or more communication buses. Among them, the one or more computer programs are stored in the above memory and configured to be executed by the processor, and when the one or more computer programs are executed by the processor, the functions or steps in the above dialogue interaction method are implemented.

[0261] An embodiment of the present application also provides a computer-readable storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processing circuit, the functions or steps in the above-mentioned dialogue interaction method are implemented.

[0262] In addition, an embodiment of the present application may also provide a chip system, which includes a processing circuit and a storage medium, wherein the storage medium stores computer program instructions, and when the computer program instructions are executed by the processing circuit, the functions or steps in the above-mentioned dialogue interaction method are implemented.

[0263] In addition, an embodiment of the present application may also provide a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute the functions or steps in the above-mentioned dialogue interaction method.

[0264] In addition, an embodiment of the present application may also provide a communication system, which includes an auxiliary device and a terminal device, and the terminal device and the auxiliary device can be used to execute the functions or steps in the above-mentioned dialogue interaction method.

[0265] In addition, an embodiment of the present application may also provide an operating system, which, when running on a computer, enables the computer to execute the functions or steps in the above-mentioned dialogue interaction method.

[0266] Through the description of the above implementation methods, technical personnel in the relevant field can clearly understand that, for the convenience and conciseness of description, the specific working processes of the chip system, electronic device, computer-readable storage medium, computer program product containing instructions, communication system, and operating system described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0267] It is understandable that the steps of the method or algorithm described in conjunction with the embodiments of the present application can be implemented in a hardware manner, or can be implemented by a processor executing a software instruction. The software instruction can be composed of corresponding software modules, and the software module can be stored in a random access memory, a flash memory, an erasable programmable read-only memory, an electrically erasable programmable read-only memory, a register, a hard disk, a mobile hard disk, a read-only optical disk, or any other form of storage medium. An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and can write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an application specific integrated circuit (Application Specific Integrated Circuit, ASIC). In addition, the ASIC can be located in an electronic device. Of course, the processor and the storage medium can also be present in an electronic device as discrete components.

[0268] In an optional manner, when software is used for implementation, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is implemented in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a digital video disk (DVD)), or a semiconductor medium (e.g., a solid state disk (SSD)), etc.

[0269] The above is only a specific implementation of the embodiment of the present application, but the protection scope of the embodiment of the present application is not limited thereto, and any changes or replacements within the technical scope disclosed in the embodiment of the present application should be included in the protection scope of the embodiment of the present application. Therefore, the protection scope of the embodiment of the present application should be based on the protection scope of the claims.

Claims

1. A conversation interaction method, characterized in that: Applied to a first device, the method includes: Acquire first inquiry information and a target historical conversation record, wherein the target historical conversation record is a historical conversation record associated with the first inquiry information in multiple rounds of historical conversation records; Rewrite the query according to the target historical conversation record and the first query information to obtain second query information; Obtain response data of the second inquiry information according to the second inquiry information.

2. The method according to claim 1, characterized in that The target historical conversation record is a historical conversation record in the multiple rounds of historical conversation records whose similarity with the first inquiry information meets a set threshold.

3. The method according to claim 1, characterized in that After obtaining the reply data of the second inquiry information according to the second inquiry information, the method further includes: Based on the reply data of the second inquiry information, sending the first inquiry result to the second device; or, The first inquiry result is displayed through a first result interface, and the first inquiry result includes response data of the second inquiry information.

4. The method according to claim 3, characterized in that After sending the first query result to the second device, the method further includes: receiving an inquiry instruction for the first inquiry information from the second device, and obtaining reply data of the first inquiry information according to the first inquiry information; Sending a second inquiry result to the second device based on the reply data of the first inquiry information; After displaying the first inquiry result through the first result interface, the method further includes: In response to the inquiry operation for the first inquiry information, obtaining reply data of the first inquiry information according to the first inquiry information; The second inquiry result is displayed through the second result interface, and the second inquiry result includes the reply data of the first inquiry information.

5. The method according to claim 1, characterized in that Before obtaining the first inquiry information and the target historical conversation record, the method further includes: In response to a conversation fusion search instruction for a target historical conversation record in the multiple rounds of historical conversation records, query recommendation prediction is performed according to the target historical conversation record to obtain predicted query information, wherein the target historical conversation record is a historical conversation record selected from the multiple rounds of historical conversation records in response to a conversation selection operation of a user; The obtaining of the first inquiry information includes: Acquire first query information obtained according to the predicted query information.

6. The method according to claim 5, characterized in that The obtaining of first query information obtained according to the predicted query information includes: sending the prediction query information to a second device, and receiving the first query information obtained from the second device according to the prediction query information; or, Display the predicted inquiry information and the inquiry input field through the information display interface; In response to an inquiry operation on the information display interface for the predicted inquiry information or in response to inquiry information input into the inquiry input field, the first inquiry information is obtained.

7. The method according to any one of claims 3 to 6, characterized in that: After obtaining the reply data of the second inquiry information according to the second inquiry information, the method further includes: Inquiry recommendation prediction is performed according to the second inquiry information and the target historical conversation record to obtain recommended inquiry information; the first inquiry result includes the recommended inquiry information and reply data of the second inquiry information.

8. A method for dialogue interaction, characterized in that: Applied to the second device, the method includes: Receiving first inquiry information input by a user; Sending the first inquiry information and a target historical conversation record to a first device; the target historical conversation record is a historical conversation record associated with the first inquiry information in multiple rounds of historical conversation records; receiving a first inquiry result from the first device, the first inquiry result including reply data of a second inquiry information, the second inquiry information being inquiry information obtained after the first device inquires and rewrites the first inquiry information according to the target historical conversation record; The first inquiry result is displayed through a first result interface.

9. The method according to claim 8, characterized in that The target historical conversation record is a historical conversation record in the multiple rounds of historical conversation records whose similarity with the first inquiry information meets a set threshold.

10. The method according to claim 8, characterized in that After displaying the first inquiry result through the first result interface, the method further includes: In response to the user's query operation on the first query information, sending a query instruction for the first query information to the first device; receiving a second inquiry result from the first device, wherein the second inquiry result includes reply data to the first inquiry information; The second inquiry result is displayed through a second result interface.

11. The method according to claim 8, characterized in that Before receiving the first inquiry information input by the user, the method further includes: Displaying a conversation record interface, wherein the conversation record interface includes the multiple rounds of historical conversation records; In response to a dialogue selection operation of the user on the dialogue record interface, obtaining a target historical dialogue record in the multiple rounds of historical dialogue records; In response to the user's conversation fusion search operation, sending a conversation fusion search instruction for the target historical conversation record to the first device; Receiving predicted query information obtained from the first device according to the target historical conversation record; The predicted query information is displayed through an information display interface, and the information display interface also includes a query input field; the first query information is one of the predicted query information or query information input through the query input field.

12. The method according to any one of claims 8 to 10, characterized in that: The inquiry result returned by the first device also includes recommended inquiry information corresponding to the inquiry information, and the first inquiry information is one of the recommended inquiry information included in the inquiry result of the previous round of dialogue.

13. The method according to any one of claims 8 to 12, characterized in that: Before sending the first inquiry information and the target historical conversation record to the first device, the method further includes: In response to the user's import operation, the multiple rounds of historical conversation records are imported.

14. The method according to any one of claims 8 to 13, characterized in that: After displaying the first result interface, the method further includes: In response to the user's export operation, the multiple rounds of historical conversation records and the current round of conversation records are exported, and the current round of conversation records includes the first inquiry information and the first inquiry result.

15. An electronic device, characterized in that: The electronic device comprises: Display screen, used for interface display; A transceiver for sending and receiving radio signals; a memory for storing computer program instructions; A processor is used to execute the computer program instructions to support the electronic device to implement the method as described in any one of claims 1-7 or any one of claims 8-14.

16. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by the processing circuit, the method according to any one of claims 1 to 7 or the method according to any one of claims 8 to 14 is implemented.

17. A dialogue interaction system, characterized in that: The method comprises a first device and a second device, wherein the first device comprises a first module for implementing the method as described in any one of claims 1 to 7, and the second device comprises a second module for implementing the method as described in any one of claims 8 to 14, and the first module and the second module are trained using the same model base.

18. A chip system, characterized in that: The chip system includes a processing circuit and a storage medium, wherein the storage medium stores computer program instructions; when the computer program instructions are executed by the processing circuit, the method as described in any one of claims 1 to 7 or the method as described in any one of claims 8 to 14 are implemented.

19. A computer program product comprising instructions, characterized in that When the computer program product is executed on a computer, the computer is enabled to execute the method according to any one of claims 1 to 7 or the method according to any one of claims 8 to 14.

Images