Dialogue Management Method, Device, Computer Equipment and Storage Medium

Abstract

An embodiment of the present invention discloses a dialogue management method, device, computer device, and storage medium. The method includes: receiving a dialogue voice sent by a user based on a preset first dialogue state; switching the first dialogue state to a second dialogue state according to the dialogue voice; generating recommended information requested by the user in the second dialogue state and obtaining evaluation feedback information of the user on the second dialogue state; switching the second dialogue state according to the evaluation feedback information to obtain a third dialogue state for executing the user's next round of dialogue. Based on artificial intelligence technology, the present invention accurately locates the user's needs through this method, greatly improving the user experience.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a dialogue management method, device, computer equipment and storage medium. Background Art

[0002] In recent years, with the rapid development of computer Internet technology, various fields of technology have begun to focus on artificial intelligence technology, leading to a wave of artificial intelligence research and opening a new chapter in the era of artificial intelligence. Among them, intelligent dialogue management systems, as an important form of artificial intelligence, have played a role in all aspects of people's lives, bringing great convenience to mankind. Traditional dialogue management systems usually adopt methods based on templates and manual rules, but such systems not only require the design of cumbersome manual rules and fixed language templates, but also only analyze user usage data, resulting in the inability of users to accurately locate user needs from the dialogue management system when interacting with the dialogue management system. Summary of the invention

[0003] The embodiments of the present invention provide a dialogue management method, apparatus, computer equipment and storage medium, which are used to solve the technical problem in the prior art that user needs cannot be accurately located in the dialogue management system.

[0004] In a first aspect, an embodiment of the present invention provides a dialog management method, which includes:

[0005] Receiving a conversation voice sent by a user based on a preset first conversation state;

[0006] Switching the first dialogue state to a second dialogue state according to the dialogue voice;

[0007] generating, in the second dialogue state, recommendation information requested by the user and obtaining evaluation feedback information of the user on the second dialogue state;

[0008] The second dialogue state is switched according to the evaluation feedback information to obtain a third dialogue state for executing the next round of dialogue with the user.

[0009] In a second aspect, an embodiment of the present invention provides a dialog management device, which includes:

[0010] A receiving unit, configured to receive a conversation voice sent by a user based on a preset first conversation state;

[0011] A first switching unit, configured to switch the first dialogue state to a second dialogue state according to the dialogue voice;

[0012] A first generation unit, configured to generate the recommended information requested by the user in the second dialogue state and obtain the evaluation feedback information of the user on the second dialogue state;

[0013] A second switching unit, configured to switch the second dialogue state according to the evaluation feedback information to obtain a third dialogue state for executing the next round of dialogue of the user.

[0014] In a third aspect, an embodiment of the present invention further provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the dialogue management method described in the first aspect above is implemented.

[0015] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor is caused to execute the dialogue management method described in the first aspect above.

[0016] An embodiment of the present invention provides a dialogue management method, device, computer device, and storage medium. By constructing an evaluation feedback mechanism of a user in a dialogue management system, when the user interacts with the dialogue management system for one round, the dialogue management system pre-switches the dialogue state through the dialogue voice sent by the user and generates recommended information in the switched dialogue state. At the same time, according to the evaluation feedback information of the user on the switched dialogue state, the switched dialogue state is further switched to facilitate the user's next round of dialogue, thereby realizing accurate positioning of the user's needs and improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a schematic flowchart of the dialogue management method provided by an embodiment of the present invention;

[0019] Figure 2 It is a schematic flowchart of the dialogue management method provided by an embodiment of the present invention;

[0020] Figure 3 It is another schematic flowchart of the dialogue management method provided by an embodiment of the present invention;

[0021] Figure 4Another flowchart of the dialogue management method provided by the embodiment of the present invention;

[0022] Figure 5 Another flowchart of the dialogue management method provided by the embodiment of the present invention;

[0023] Figure 6 Another flowchart of the dialogue management method provided by the embodiment of the present invention;

[0024] Figure 7 Another flowchart of the dialogue management method provided by the embodiment of the present invention;

[0025] Figure 8 Schematic block diagram of the dialogue management device provided by the embodiment of the present invention;

[0026] Figure 9 Schematic block diagram of the computer device provided by the embodiment of the present invention. Detailed implementation manners

[0027] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0028] It should be understood that when used in this specification and the appended claims, the terms "comprises" and "comprising" indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0029] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in this specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" are intended to include the plural forms.

[0030] It should be further understood that the term " / and / or" used in this specification of the present invention and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0031] Please refer to Figure 1 , Figure 1Schematic flowchart of the dialogue management method provided by the embodiments of the present invention. The dialogue management method of the embodiments of the present invention is applied to a terminal device, and the method is executed through a dialogue management system installed in the terminal device. Among them, the terminal device is a terminal device capable of collecting voice and processing voice, such as a liquid crystal TV, a smart screen, a desktop computer, a laptop computer, a tablet computer, or a mobile phone, etc.

[0032] The following provides a detailed description of the dialogue management method.

[0033] As Figure 1 shown, the method includes the following steps S110 to S140.

[0034] S110. Receive the dialogue voice sent by the user based on a preset first dialogue state.

[0035] Specifically, the first dialogue state is the dialogue state of the dialogue management system of the terminal device before the user sends the dialogue voice to the terminal device. The first dialogue state can be either the dialogue state of the dialogue management system of the terminal device before the user first sends the dialogue voice to the terminal device, or the dialogue state of the dialogue management system of the terminal device before the user sends the dialogue voice to the terminal device non-first. After the terminal device receives the dialogue voice sent by the user based on the first dialogue state of its dialogue management system, the first dialogue state can be switched to a dialogue state for serving the dialogue voice in the dialogue management system through the dialogue voice, so as to meet the needs when the user sends the dialogue voice. Among them, the dialogue state S in the dialogue management system of the terminal device = {Domain, Intent, Slot}, the dialogue voice or request sent by the user to the dialogue management system is called Query, and Domain, Intent, and Slot together form the dialogue state of the dialogue management system. Domain is the event-related domain corresponding to Query, Intent is the user's intention, and Slot is the slot information in Query.

[0036] S120. Switch the first dialogue state to a second dialogue state according to the dialogue voice.

[0037] Specifically, the conversation voice is the voice information sent by the user to the terminal device when interacting with the conversation management system in the terminal device. After receiving the conversation voice sent by the user, the terminal device converts the conversation voice into a speech text of the conversation voice, and then performs natural language processing on the text. Then, the conversation management system can switch the first conversation state to the second conversation state, and finally make an action response to the conversation voice sent by the user in the second conversation state, that is, push the recommended information of the conversation voice to the user. Among them, the first conversation state and the second conversation state can be the corresponding conversation states in the existing database or rule template, or can be newly created conversation states in the conversation management system, which are not specifically limited here.

[0038] In other embodiments of the invention, such as Figure 2 shown, step S120 includes sub-steps S121, S122, and S123.

[0039] S121. Generate the speech text of the conversation voice;

[0040] S122. Input the speech text into a pre-trained language model to obtain the semantic information of the conversation voice;

[0041] S123. Switch the first conversation state to the second conversation state according to the semantic information.

[0042] In this embodiment, the second conversation state is obtained by updating the first conversation state. When the conversation management system is in the second conversation state, it can perform an action response to the conversation voice in the second conversation state. The language model is pre-trained and used for natural language processing. After receiving the conversation voice sent by the user, the conversation voice is converted into a speech text of the conversation voice in the terminal device, and then the speech text is input into the language model for a series of processing such as entity named entity recognition, and the semantic information of the conversation voice can be obtained. Finally, through the semantic information of the conversation voice, the first conversation state can be switched to the second conversation state by the conversation management system, and an action response to the conversation voice sent by the user can be made in the second conversation state.

[0043] In addition, when the conversation voice is the voice information sent by the user to the terminal device for the first time, the Domain, Intent, and Slot in the first conversation state are all different from those in the second conversation state; when the conversation voice is the voice information sent by the user to the terminal device not for the first time, only the Slot in the first conversation state is different from the Slot in the second conversation state, while the Domain and Intent in the first conversation state and the second conversation state are the same.

[0044] In other embodiments of the invention, as Figure 3 shown, step S123 includes sub-steps S1231, S1232, and S1233.

[0045] S1231. Obtain multiple semantic tags of the semantic information;

[0046] S1232. Screen the multiple semantic tags according to the evaluation feedback information of the first conversation state to obtain the screened semantic tags;

[0047] S1233. Switch the first conversation state to the second conversation state according to the screened semantic tags.

[0048] Specifically, the semantic information of the conversation voice is represented by semantic tags. After the language model processes the speech text of the conversation voice, it outputs multiple semantic tags of the speech text. Among them, when determining the semantic information of the conversation voice, usually, multiple output semantic tags are screened according to their probabilities, and the semantic tag with the highest probability is used as the semantic information of the conversation voice. However, this method still ignores the user's current actual intention, resulting in a large error in the recommended information finally pushed to the user. Therefore, when screening the multiple semantic tags output by the language model in this embodiment, the evaluation feedback information of the user on the first conversation state is added to screen the multiple semantic tags of the semantic information, thereby improving the generalization, timeliness, and accuracy of information pushing to the user. For example, when the evaluation feedback information of the first conversation state is negative evaluation information, when selecting the semantic tag with the highest probability as the semantic information of the conversation voice, it is very easy to miss the user's current true intention. At this time, the two semantic tags with the highest probabilities need to be retained as the semantic information of the conversation voice, and the conversation management system is updated through the evaluation feedback information of the first conversation state so that the updated conversation management system switches the first conversation state to the second conversation state according to this semantic information and realizes the complete recommendation of the information required by the user to the user.

[0049] S130. Generate the recommended information requested by the user in the second dialogue state and obtain the evaluation feedback information of the user on the second dialogue state.

[0050] Specifically, the recommended information is data information that the user requests to obtain from the dialogue management system in the terminal device by sending dialogue voice to the terminal device. This data information can be programs, movies, TV dramas, etc. that the user watches. After the user receives the recommended information, the dialogue management system of the terminal device needs to collect the evaluation feedback information of the user on the current state service in the second dialogue state. This evaluation feedback information can be either the satisfaction degree of the user with the current recommended information or the satisfaction degree of the user with the current dialogue management system service, and no specific limitation is made here.

[0051] In other embodiments of the invention, such as Figure 4 shown, step S130 includes sub-steps S131 and S132.

[0052] S131. Obtain the duration for which the user browses the recommended information;

[0053] S132. Generate the evaluation feedback information according to the duration.

[0054] Specifically, the evaluation feedback information of the user is usually through an evaluation form pushed by the dialogue management system to the user. The user evaluates the satisfaction degree of the current dialogue state through the evaluation form to update the dialogue management system. However, this method usually causes interference to the user. Therefore, in this embodiment, the duration for which the user browses the recommended information is obtained in advance to determine whether to push the evaluation form to the user. When the duration for which the user browses the recommended information exceeds a preset threshold, it indicates that the dialogue management system has pushed accurate recommended information to the user. Furthermore, it can be determined that the user is relatively satisfied with the current recommended information. At this time, there is no need to push the evaluation form to the user, and it is defaulted that the user is relatively satisfied with the current service. Therefore, the evaluation feedback information can be directly generated.

[0055] In other embodiments of the invention, such as Figure 5 shown, step S132 includes sub-steps S1321 and S1322.

[0056] S1321. Mark a preset evaluation form according to the duration to obtain multiple evaluation labels of the evaluation feedback information;

[0057] S1322. Generate the evaluation feedback information according to the multiple evaluation labels.

[0058] In this embodiment, the evaluation form is used by the user to evaluate whether they are satisfied with the second dialogue state. The evaluation form is marked by judging whether the duration exceeds a preset threshold. If the duration exceeds the preset threshold, there is no need to push the evaluation form to the user and the evaluation reply in the evaluation form is directly completed, and then each reply is marked to obtain multiple evaluation tags of the evaluation feedback information. If the duration does not exceed the preset threshold, the evaluation form is pushed to the user when the user browses the recommended information, and the evaluation form is marked according to the user's reply to obtain multiple evaluation tags of the evaluation feedback information. Among them, when the dialogue management system pushes the evaluation form to the user and the user ignores the evaluation form, the dialogue management system automatically completes the evaluation reply in the evaluation form, and then each reply is marked to obtain multiple evaluation tags of the evaluation feedback information. For example, when a question statement in the evaluation form is "Is the problem solved / Is the task completed", if the duration does not exceed the preset threshold, the question statement is directly marked with a positive tag; if the user ignores the evaluation form, the question statement is directly marked with a negative tag.

[0059] S140. Switch the second dialogue state according to the evaluation feedback information to obtain a third dialogue state for executing the next round of dialogue of the user.

[0060] Specifically, the third dialogue state is the dialogue state required for the dialogue management system of the terminal device to provide the user with the next round of recommendation services. The difference between the second dialogue state and the third dialogue state is only that the Slots in the dialogue state are different, that is, the slot information in the two dialogue states is different. The slot information in the third dialogue state has richer word segmentation information supplemented by the dialogue management system, such as more accurate movie information. When the terminal device collects the evaluation feedback information of the user on the second dialogue state, the dialogue management system where the second dialogue state is located is iteratively updated through the evaluation feedback information to optimize the dialogue management system, so that when the dialogue management system switches the second dialogue state to the third dialogue state, the slot information in the third dialogue state is more abundant, and further enables the dialogue management system to more accurately locate the user's needs when having the next round of dialogue with the user in the third dialogue state.

[0061] In other inventive embodiments, as Figure 6 shown, step S140 includes sub-steps S141 and S142.

[0062] S141. Obtain a reward signal for switching the second dialogue state to the third dialogue state according to the evaluation feedback information.

[0063] S142. Switch the second dialogue state to the third dialogue state according to the reward signal.

[0064] In this embodiment, the reward signal is used to optimize the dialogue management system so that after the dialogue management system switches the second dialogue state to the third dialogue state, the dialogue management system can accurately locate the user's needs according to the third dialogue state. The dialogue management system is iteratively updated based on reinforcement learning and generates a reward signal for the reinforcement learning of the dialogue management system according to the evaluation feedback information of each dialogue state. Through the reward signal, the iterative update of the dialogue management system can be achieved. Among them, the reward function of the reward signal generated in the dialogue management system can be designed according to specific applications, and no specific limitation is made in this embodiment. For example, the reward function can be:

[0065]

[0066] wherein, R i is the reward signal generated in different dialogue states, and E i is the evaluation label of the evaluation feedback information in different dialogue states.

[0067] In other embodiments of the invention, as Figure 7 shown, step S141 includes sub-steps S1411 and S1412.

[0068] S1411. Input the evaluation feedback information into a pre-trained scoring model to obtain the score of the second dialogue state;

[0069] S1412. Obtain the reward signal according to the score of the second dialogue state.

[0070] In this embodiment, the evaluation feedback information includes multiple evaluation labels, and the scoring model is pre-trained and used to perform a linear weighted calculation on the multiple evaluation labels of the evaluation feedback information to obtain the score of the evaluation feedback information, and then obtain the score of the second dialogue state. Then, this score is input into the reward function for calculation, and the reward signal for optimizing the dialogue management system can be obtained. After the dialogue management system is optimized according to this reward information, the second dialogue state can be switched to the third dialogue state for executing the user's next round of dialogue, and the recommended information required by the user can be pushed to the user more accurately through the third dialogue state.

[0071] In the dialogue management method provided by the embodiments of the present invention, a dialogue voice sent by a user is received based on a preset first dialogue state; the first dialogue state is switched to a second dialogue state according to the dialogue voice; recommended information requested by the user is generated in the second dialogue state, and evaluation feedback information of the user on the second dialogue state is obtained; the second dialogue state is switched according to the evaluation feedback information to obtain a third dialogue state for executing the next round of dialogue of the user. By constructing an evaluation feedback mechanism for the user in the dialogue management system, the present invention enables the dialogue management system to pre-switch the dialogue state through the dialogue voice sent by the user and generate recommended information in the switched dialogue state when the user interacts with the dialogue management system for one round. At the same time, the switched dialogue state is further switched according to the evaluation feedback information of the user on the switched dialogue state to facilitate the user's next round of dialogue, so as to realize state transition in multi-round dialogue, service action selection of the dialogue management system, and response to user requests. At the same time, it can also identify, define, and update in a timely manner Queries that have not been processed in the database or template and the language behavior of the user, improving the generalization of the dialogue management system, and then adapting to user needs in a timely manner and accurately positioning the user's needs.

[0072] An embodiment of the present invention further provides a dialogue management device 100, which is used to execute any embodiment of the foregoing dialogue management method.

[0073] Specifically, please refer to Figure 8 , Figure 8 which is a schematic block diagram of the dialogue management device 100 provided by the embodiments of the present invention.

[0074] As Figure 8 shown, the dialogue management device 100 includes: a receiving unit 110, a first switching unit 120, a first generating unit 130, and a second switching unit 140.

[0075] The receiving unit 110 is configured to receive a dialogue voice sent by a user based on a preset first dialogue state.

[0076] The first switching unit 120 is configured to switch the first dialogue state to a second dialogue state according to the dialogue voice.

[0077] In another embodiment, the first switching unit 120 includes: a second generating unit, a first input unit, and a third switching unit.

[0078] A second generation unit for generating a speech text of the conversation speech; a first input unit for inputting the speech text into a pre-trained language model to obtain semantic information of the conversation speech; a third switching unit for switching the first conversation state to the second conversation state according to the semantic information.

[0079] In another embodiment, the third switching unit includes: a first acquisition unit, a screening unit, and a fourth switching unit.

[0080] The first acquisition unit is used to acquire a plurality of semantic tags of the semantic information; the screening unit is used to screen the plurality of semantic tags according to the evaluation feedback information of the first conversation state to obtain the screened semantic tags; the fourth switching unit is used to switch the first conversation state to the second conversation state according to the screened semantic tags.

[0081] The first generation unit 130 is used to generate recommendation information requested by the user in the second conversation state and obtain the evaluation feedback information of the user on the second conversation state.

[0082] In another embodiment, the first generation unit 130 includes: a second acquisition unit and a third generation unit.

[0083] The second acquisition unit is used to acquire the duration of the user browsing the recommendation information; the third generation unit is used to generate the evaluation feedback information according to the duration.

[0084] In another embodiment, the third generation unit includes: a marking unit and a fourth generation unit.

[0085] The marking unit is used to mark a preset evaluation form according to the duration to obtain a plurality of evaluation tags of the evaluation feedback information; the fourth generation unit is used to generate the evaluation feedback information according to the plurality of evaluation tags.

[0086] The second switching unit 140 is used to switch the second conversation state according to the evaluation feedback information to obtain a third conversation state for executing the next round of conversation of the user.

[0087] In another embodiment, the second switching unit 140 includes: a third acquisition unit and a fifth update unit.

[0088] The third acquisition unit is used to obtain a reward signal for switching the second conversation state to the third conversation state according to the evaluation feedback information; the fifth update unit is used to switch the second conversation state to the third conversation state according to the reward signal.

[0089] In another embodiment, the third acquisition unit includes: a second input unit and a fourth acquisition unit.

[0090] The second input unit is configured to input the evaluation feedback information into a pre-trained scoring model to obtain a score of the second dialogue state; the fourth acquisition unit is configured to obtain the reward signal according to the score of the second dialogue state.

[0091] The dialogue management device 100 provided by the embodiments of the present invention is configured to execute the above-mentioned steps of receiving the dialogue voice sent by the user based on the preset first dialogue state; switching the first dialogue state to a second dialogue state according to the dialogue voice; generating the recommended information requested by the user in the second dialogue state and obtaining the evaluation feedback information of the user on the second dialogue state; and switching the second dialogue state according to the evaluation feedback information to obtain a third dialogue state for executing the next round of dialogue of the user.

[0092] It should be noted that those skilled in the art can clearly understand that the specific implementation processes of the above-mentioned dialogue management device 100 and each unit can refer to the corresponding descriptions in the foregoing method embodiments. For the sake of convenience and brevity of description, they will not be repeated here.

[0093] The above-mentioned dialogue management device can be implemented in the form of a computer program, and this computer program can run on a computer device as shown in Figure 9 shown.

[0094] Please refer to Figure 9 , Figure 9 which is a schematic block diagram of a computer device provided by an embodiment of the present application. The computer device 500 may be a terminal. Among them, the terminal may be an electronic device with a communication function such as a smart phone, a tablet computer, a notebook computer, a desktop computer, a personal digital assistant, and a wearable device.

[0095] Referring to Figure 9 , the computer device 500 includes a processor 502, a memory, and a network interface 505 connected through a system bus 501. Among them, the memory may include a non-volatile storage medium 503 and an internal memory 504.

[0096] The non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032. The computer program 5032 includes program instructions, and when the program instructions are executed, the processor 502 can execute a dialogue management method.

[0097] The processor 502 is configured to provide computing and control capabilities to support the operation of the entire computer device 500.

[0098] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can be caused to execute a dialogue management method.

[0099] The network interface 505 is used for network communication with other devices. Those skilled in the art can understand that Figure 9 the structure shown in is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device 500 to which the solution of this application is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0100] Among them, the processor 502 is used to run the computer program 5032 stored in the memory to implement the following steps: receiving the dialogue voice sent by the user based on a preset first dialogue state; switching the first dialogue state to a second dialogue state according to the dialogue voice; generating the recommended information requested by the user in the second dialogue state and obtaining the evaluation feedback information of the user on the second dialogue state; switching the second dialogue state according to the evaluation feedback information to obtain a third dialogue state for executing the user's next round of dialogue.

[0101] In one embodiment, when the processor 502 implements switching the first dialogue state to the second dialogue state according to the dialogue voice, the following steps are specifically implemented: generating the speech text of the dialogue voice; inputting the speech text into a pre-trained language model to obtain the semantic information of the dialogue voice; switching the first dialogue state to the second dialogue state according to the semantic information.

[0102] In one embodiment, when the processor 502 implements switching the first dialogue state to the second dialogue state according to the semantic information, the following steps are specifically implemented: obtaining multiple semantic tags of the semantic information; screening the multiple semantic tags according to the evaluation feedback information of the first dialogue state to obtain the screened semantic tags; switching the first dialogue state to the second dialogue state according to the screened semantic tags.

[0103] In one embodiment, when the processor 502 implements obtaining the evaluation feedback information of the user on the second dialogue state, the following steps are specifically implemented: obtaining the duration of the user browsing the recommended information; generating the evaluation feedback information according to the duration.

[0104] In one embodiment, when the processor 502 implements generating the evaluation feedback information according to the duration, the specific implementation steps are as follows: annotate a preset evaluation form according to the duration to obtain multiple evaluation labels of the evaluation feedback information; generate the evaluation feedback information according to the multiple evaluation labels.

[0105] In one embodiment, when the processor 502 implements switching the second dialogue state according to the evaluation feedback information to obtain a third dialogue state for executing the next round of dialogue of the user, the specific implementation steps are as follows: obtain a reward signal for switching the second dialogue state to the third dialogue state according to the evaluation feedback information; switch the second dialogue state to the third dialogue state according to the reward signal.

[0106] In one embodiment, when the processor 502 implements obtaining the reward signal for switching the second dialogue state to the third dialogue state according to the evaluation feedback information, the specific implementation steps are as follows: input the evaluation feedback information into a pre-trained scoring model to obtain a score of the second dialogue state; obtain the reward signal according to the score of the second dialogue state.

[0107] It should be understood that in the embodiments of the present application, the processor 502 may be a central processing unit (CPU), and this processor 502 may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0108] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program includes program instructions, and the computer program can be stored in a storage medium, and the storage medium is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0109] Therefore, the present invention also provides a storage medium. The storage medium can be a computer-readable storage medium. The storage medium stores a computer program, wherein the computer program includes program instructions. When the program instructions are executed by a processor, the processor performs the following steps: receiving the dialogue voice sent by the user based on a preset first dialogue state; switching the first dialogue state to a second dialogue state according to the dialogue voice; generating the recommended information requested by the user in the second dialogue state and obtaining the evaluation feedback information of the user on the second dialogue state; switching the second dialogue state according to the evaluation feedback information to obtain a third dialogue state for executing the next round of dialogue of the user.

[0110] In one embodiment, when the processor executes the program instructions to implement switching the first dialogue state to the second dialogue state according to the dialogue voice, the following steps are specifically implemented: generating a speech text of the dialogue voice; inputting the speech text into a pre-trained language model to obtain semantic information of the dialogue voice; switching the first dialogue state to the second dialogue state according to the semantic information.

[0111] In one embodiment, when the processor executes the program instructions to implement switching the first dialogue state to the second dialogue state according to the semantic information, the following steps are specifically implemented: obtaining a plurality of semantic tags of the semantic information; screening the plurality of semantic tags according to the evaluation feedback information of the first dialogue state to obtain the screened semantic tags; switching the first dialogue state to the second dialogue state according to the screened semantic tags.

[0112] In one embodiment, when the processor executes the program instructions to implement obtaining the evaluation feedback information of the user on the second dialogue state, the following steps are specifically implemented: obtaining the duration of the user browsing the recommended information; generating the evaluation feedback information according to the duration.

[0113] In one embodiment, when the processor executes the program instructions to implement generating the evaluation feedback information according to the duration, the following steps are specifically implemented: annotating a preset evaluation form according to the duration to obtain a plurality of evaluation tags of the evaluation feedback information; generating the evaluation feedback information according to the plurality of evaluation tags.

[0114] In one embodiment, when the processor executes the program instructions to implement switching the second dialogue state according to the evaluation feedback information to obtain a third dialogue state for the user's next round of dialogue, the specific implementation is as follows: obtaining a reward signal for switching the second dialogue state to the third dialogue state according to the evaluation feedback information; switching the second dialogue state to the third dialogue state according to the reward signal.

[0115] In one embodiment, when the processor executes the program instructions to implement obtaining a reward signal for switching the second dialogue state to the third dialogue state according to the evaluation feedback information, the specific implementation is as follows: inputting the evaluation feedback information into a pre-trained scoring model to obtain a score of the second dialogue state; obtaining the reward signal according to the score of the second dialogue state.

[0116] The storage medium may be a variety of computer-readable storage media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disc that can store program codes.

[0117] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the components and steps of the examples have been generally described according to their functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0118] In several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of each unit is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0119] The steps in the method embodiments of the present invention can be adjusted, combined, and deleted according to actual needs. The units in the device embodiments of the present invention can be combined, divided, and deleted according to actual needs. In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0120] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention.

[0121] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A dialogue management method, characterized in that, including: receiving the dialogue voice sent by the user based on a preset first dialogue state, where the dialogue state includes: an event-related domain corresponding to the dialogue voice, a user intention, and slot information in the dialogue voice; switching the first dialogue state to a second dialogue state according to the dialogue voice, where when the dialogue voice is the voice information sent by the user for the non-first time, the slot information in the first dialogue state is different from that in the second dialogue state, but the event-related domain and the user intention in the first dialogue state and the second dialogue state are the same; generating recommended information requested by the user in the second dialogue state and obtaining evaluation feedback information of the user on the second dialogue state; switching the second dialogue state according to the evaluation feedback information to obtain a third dialogue state for executing the next round of dialogue of the user, where the slot information in the second dialogue state is different from that in the third dialogue state, and the slot information in the third dialogue state includes word segmentation information supplemented by a dialogue management system.

2. The dialogue management method according to claim 1, characterized in that The switching of the first dialogue state to the second dialogue state according to the dialogue voice includes: generating a speech text of the dialogue voice; inputting the speech text into a pre-trained language model to obtain semantic information of the dialogue voice; switching the first dialogue state to the second dialogue state according to the semantic information.

3. The dialogue management method according to claim 2, wherein The switching of the first dialogue state to the second dialogue state according to the semantic information includes: obtaining multiple semantic tags of the semantic information; screening the multiple semantic tags according to the evaluation feedback information of the first dialogue state to obtain screened semantic tags; switching the first dialogue state to the second dialogue state according to the screened semantic tags.

4. The dialogue management method according to claim 1, characterized in that, The obtaining of the evaluation feedback information of the user on the second dialogue state includes: obtaining the duration for which the user browses the recommended information; generating the evaluation feedback information according to the duration.

5. The dialogue management method according to claim 4, characterized in that, The generating of the evaluation feedback information according to the duration includes: annotating a preset evaluation form according to the duration to obtain multiple evaluation tags of the evaluation feedback information; generating the evaluation feedback information according to the multiple evaluation tags.

6. The dialogue management method according to claim 1, characterized in that, The switching of the second dialogue state according to the evaluation feedback information to obtain a third dialogue state for executing the next round of dialogue of the user includes: obtaining a reward signal for switching the second dialogue state to the third dialogue state according to the evaluation feedback information; switching the second dialogue state to the third dialogue state according to the reward signal.

7. The dialogue management method according to claim 6, wherein The obtaining of the reward signal for switching the second dialogue state to the third dialogue state according to the evaluation feedback information includes: inputting the evaluation feedback information into a pre-trained scoring model to obtain a score of the second dialogue state; obtaining the reward signal according to the score of the second dialogue state.

8. A dialogue management device, characterized in that including: A receiving unit, configured to receive a conversation voice sent by a user based on a preset first conversation state, where the conversation state includes: an event-related domain corresponding to the conversation voice, a user intention, and slot information in the conversation voice; A first switching unit, configured to switch the first conversation state to a second conversation state according to the conversation voice, where when the conversation voice is voice information sent by the user for non-first time, the slot information in the first conversation state is different from that in the second conversation state, but the event-related domain and the user intention in the first conversation state and the second conversation state are the same; A first generating unit, configured to generate recommended information requested by the user in the second conversation state and obtain evaluation feedback information of the user on the second conversation state; A second switching unit, configured to switch the second conversation state according to the evaluation feedback information to obtain a third conversation state for executing the next round of conversation of the user, where the slot information in the second conversation state is different from that in the third conversation state, and the slot information in the third conversation state includes word segmentation information supplemented by a conversation management system.

9. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, it implements the conversation management method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the conversation management method according to any one of claims 1 to 7.

Images