Active dialogue method and device of dialogue robot, electronic equipment and storage medium

Abstract

This invention discloses a method, apparatus, electronic device, and storage medium for an active dialogue robot. The method includes: detecting dialogue parameters of the dialogue object during a dialogue between the dialogue robot and the dialogue object; when the dialogue parameters meet pre-set active dialogue conditions, generating active dialogue content based on the attribute information of the user account simulated by the dialogue robot; and playing the active dialogue content based on an active dialogue mode independent of the current dialogue state of the dialogue object. The dialogue robot of this invention can initiate active dialogue, achieve full-duplex human-machine dialogue, and the dialogue content is controllable, also achieving scalable dialogue control.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of conversational robot technology, and more specifically, to a method, apparatus, electronic device and storage medium for an active conversational robot. Background Technology

[0002] A chatbot is a combination of hardware and computer programs that enable natural dialogue with users. In a chatbot's dialogue, the user's voice information undergoes the following processes: (1) Automatic Speech Recognition (ASR) processing: After preprocessing such as sentence segmentation, the speech is converted into text for machine understanding; (2) Natural Language Understanding (NLU): Semantic information in the text is extracted, such as emotion, sentence structure (interrogative or declarative), intent, and slots; (3) Dialogue Manager (DM): Response strategy is given by combining conversation history, current semantic information, and relevant data; (4) Natural Language Generation (NLG): Response strategy is converted into natural language text by combining contextual information; (5) Text-to-Speech (TTS): Natural language text is converted into speech signal and the speech signal is played.

[0003] However, most current chatbots primarily provide passive answers and do not proactively ask questions. Summary of the Invention

[0004] The present invention provides a method, apparatus, electronic device, and storage medium for active dialogue of a chatbot.

[0005] The technical solution of the embodiments of the present invention is as follows:

[0006] A method for proactive dialogue in a chatbot, comprising:

[0007] Detect the dialogue parameters of the dialogue object during the dialogue process between the chatbot and the dialogue object;

[0008] When the dialogue parameters meet the preset active dialogue conditions, active dialogue content is generated based on the attribute information of the user account simulated by the chatbot.

[0009] The active dialogue content is played based on an active dialogue mode that is independent of the current dialogue state of the dialogue object.

[0010] In an exemplary implementation, the attribute information includes interest information;

[0011] When the dialogue parameters meet the preset active dialogue conditions, the generated active dialogue content includes at least one of the following:

[0012] When the duration of continuous silence of the dialogue object is greater than or equal to a preset silence threshold, the proactive dialogue content is generated based on a question template associated with the interest information.

[0013] When the continuous speaking time of the dialogue object is greater than or equal to a preset speaking time threshold, the active dialogue content is generated based on the question template associated with the interest information.

[0014] When the content of the dialogue object matches the interest information, the proactive dialogue content is generated based on the question template associated with the interest information.

[0015] In an exemplary implementation, generating the proactive dialogue content based on a question template associated with the interest information includes:

[0016] Determine the context information during the dialogue process;

[0017] Based on the context information, determine the interest factors that the dialogue partner has already described;

[0018] Remove the interest factors from the interest information to obtain the remaining interest information;

[0019] The proactive dialogue content is generated based on the question template associated with the remaining interest information.

[0020] In an exemplary embodiment, before playing the active dialogue content based on an active dialogue mode independent of the current dialogue state of the dialogue object, the method further includes:

[0021] Speech recognition processing is performed on the speech of the dialogue partners during the dialogue process to generate the text content of the speech;

[0022] Perform semantic recognition on the text content;

[0023] Based on the semantic recognition results, the response content of the chatbot is determined;

[0024] When the dialogue parameters do not meet the conditions for proactive dialogue, or when the dialogue parameters meet the conditions for proactive dialogue but the number of times the chatbot initiates proactive dialogue exceeds a preset threshold, the proactive dialogue content will be updated to the response content.

[0025] In the exemplary implementation, it also includes:

[0026] Detect the silence signal fragments of the dialogue object during the dialogue process;

[0027] The N consecutive mute signals are divided into segments and sent to the server in a concentrated manner, where N is a pre-set positive integer value;

[0028] On the server side, the continuous silence duration is determined based on the total duration of the N silence signal segments.

[0029] At least one of N, the mute threshold, and the time length of the mute signal segment is adjustable.

[0030] An active dialogue device for a chatbot, comprising:

[0031] The detection module is used to detect the dialogue parameters of the dialogue object during the dialogue process between the chatbot and the dialogue object;

[0032] The generation module is used to generate active dialogue content based on the attribute information of the user account simulated by the chatbot when the dialogue parameters meet the preset active dialogue conditions.

[0033] The playback module is used to play the active dialogue content based on an active dialogue mode that is independent of the current dialogue state of the dialogue object.

[0034] In an exemplary implementation, the attribute information includes interest information;

[0035] The generation module is configured to perform at least one of the following:

[0036] When the duration of continuous silence of the dialogue object is greater than or equal to a preset silence threshold, the proactive dialogue content is generated based on a question template associated with the interest information.

[0037] When the continuous speaking time of the dialogue object is greater than or equal to a preset speaking time threshold, the active dialogue content is generated based on the question template associated with the interest information.

[0038] When the content of the dialogue object matches the interest information, the proactive dialogue content is generated based on the question template associated with the interest information.

[0039] In an exemplary implementation, the generation module is configured to determine contextual information during the dialogue process; determine the interest factors already discussed by the dialogue object based on the contextual information; remove the interest factors from the interest information to obtain the remaining interest information; and generate the proactive dialogue content based on a question template associated with the remaining interest information.

[0040] In an exemplary embodiment, before the playback module plays the active dialogue content based on an active dialogue mode independent of the current dialogue state of the dialogue object, the generation module performs speech recognition processing on the speech of the dialogue object during the dialogue process to generate text content of the speech; performs semantic recognition on the text content; determines the response content of the chatbot based on the semantic recognition result; and updates the active dialogue content to the response content when the dialogue parameters do not meet the active dialogue conditions, or when the dialogue parameters meet the active dialogue conditions and the number of times the chatbot initiates active dialogue exceeds a preset active dialogue number threshold.

[0041] In an exemplary embodiment, a detection module is used to detect mute signal segments of the dialogue object during the dialogue process; to send N consecutive mute signal segments to the server, where N is a preset positive integer value; wherein on the server, the continuous mute duration is determined based on the total duration of the N mute signal segments; wherein at least one of N, the mute threshold value, and the duration of the mute signal segments is adjustable.

[0042] A computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, can implement the steps of the active dialogue method of a chatbot as described in any of the preceding claims.

[0043] An electronic device comprising:

[0044] processor;

[0045] Memory for storing the executable instructions of the processor;

[0046] The processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the active dialogue method of the chatbot as described in any of the preceding claims.

[0047] A computer program product includes computer instructions that, when executed by a processor, implement the proactive dialogue method of a chatbot as described in any of the preceding claims.

[0048] As can be seen from the above technical solution, in the embodiments of the present invention, the dialogue parameters of the dialogue object are detected during the dialogue process between the chatbot and the dialogue object; when the dialogue parameters meet the preset active dialogue conditions, active dialogue content is generated based on the attribute information of the user account simulated by the chatbot; and the active dialogue content is played based on an active dialogue mode independent of the current dialogue state of the dialogue object. Therefore, the chatbot of the embodiments of the present invention can utilize the dialogue parameters of the dialogue object to initiate active dialogue independent of the current dialogue state of the dialogue object, realizing full-duplex human-machine dialogue and more accurately simulating the user communicating with the chat object.

[0049] Moreover, the embodiments of the present invention generate proactive dialogue content based on question templates associated with interest information, thereby making the dialogue content controllable and reducing the risk of use.

[0050] Furthermore, since N, the silence threshold, and the duration of the silence signal segmentation are all adjustable, this embodiment of the invention also achieves scalable dialogue control. For example, when the server-side pressure is high, N can be increased to reduce the request volume; conversely, N can be decreased to improve the sensitivity of the robot's proactive dialogue. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a schematic diagram of the human-computer dialogue mode.

[0053] Figure 2 This is an exemplary flowchart of the active dialogue method of the chatbot according to an embodiment of the present invention.

[0054] Figure 3 This is an exemplary schematic diagram of the dialogue processing logic of the chatbot according to an embodiment of the present invention.

[0055] Figure 4 This is an exemplary schematic diagram of a conversational robot participating in the training of brokers according to an embodiment of the present invention.

[0056] Figure 5 This is an exemplary structural diagram of the active dialogue device of the dialogue robot according to an embodiment of the present invention.

[0057] Figure 6 This is an exemplary structural diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0058] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings.

[0059] For the sake of brevity and intuitiveness, the following description uses several representative embodiments to illustrate the solution of the present invention. Numerous details in the embodiments are only used to aid in understanding the solution of the present invention. However, it is obvious that the technical solution of the present invention can be implemented without being limited to these details. To avoid unnecessarily obscuring the solution of the present invention, some embodiments are not described in detail, but only a framework is given. In the following text, "comprising" means "including but not limited to," and "according to..." means "at least according to..., but not limited to only according to...". Due to Chinese language habits, unless the quantity of a component is specifically indicated below, it means that the component can be one or more, or can be understood as at least one.

[0060] The terms "first," "second," "third," "fourth," etc. (if present) in the description, claims, and accompanying drawings of embodiments of the present invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented, for example, in orders other than those illustrated or described herein.

[0061] Figure 1 This is a schematic diagram of human-computer dialogue modes. Human-computer dialogue modes mainly include: simplex mode, half-duplex mode, and full-duplex mode.

[0062] In simplex mode: User 11 speaks, and robot 12 only records without responding; robot 12 acts like a tape recorder. In half-duplex mode: User 11 speaks, and robot 12 responds only after the user finishes speaking; robot 12 acts like a walkie-talkie. In full-duplex mode: User 11 and robot 12 can speak simultaneously. If user 11 hasn't finished speaking (e.g., "How much down payment are you prepared? I'll see if it's enough…"), robot 12 can start speaking, meaning robot 12 can interrupt user 11.

[0063] Current chatbots primarily operate in half-duplex mode, mainly using single-turn conversations with multi-turn conversations as a secondary function. Multi-turn conversations are driven by a pre-set finite state machine. Chatbots mainly employ a multi-path recall and reordering process to generate responses. Recall channels include sub-functions such as classification, retrieval, and matching. For example, if a user asks, "How much down payment are you prepared?", the chatbot in half-duplex mode identifies the user's intention to buy a house. It then recalls corresponding answer templates through multiple channels such as rules, classification, retrieval, and matching. After scoring and ranking each answer template, the highest-scoring answer template is selected as the final response and returned to the user.

[0064] The applicant found that chatbots operating in half-duplex mode struggle to accurately simulate real-world conversations. In half-duplex mode, the chatbot only responds when the user asks a question; it remains silent unless asked. However, many scenarios require chatbots to interrupt users and proactively ask questions. For example, during virtual reality (VR) property viewings, when an agent inquires about a client's down payment budget, the client might not only reply with "approximately xx million," but also refuse to answer (e.g., remain silent), give a vague answer (e.g., say "I'm not sure"), or ask a follow-up question (e.g., "How much down payment do you need for this house?"), or even interrupt the agent before they finish speaking ("How much down payment are you prepared? I'll see...") by saying "Approximately xx million." Therefore, half-duplex chatbots cannot effectively simulate real-world customer behavior.

[0065] This invention proposes a proactive dialogue scheme for a chatbot. The chatbot can initiate dialogue based on dialogue parameters, meaning it can operate in full-duplex mode, thereby more accurately simulating the user (e.g., customer) communicating with the chat partner (e.g., agent).

[0066] Figure 1 This is an exemplary flowchart of the active dialogue method of the chatbot according to an embodiment of the present invention. Figure 1 The method shown can be executed by a chatbot. For example... Figure 1 As shown, the method includes:

[0067] Step 101: Detect the dialogue parameters of the dialogue object during the dialogue process between the chatbot and the dialogue object.

[0068] Here, the dialogue partner is typically a human speaking to the chatbot, but it can also be another chatbot speaking to the chatbot. The chatbot simulates the dialogue between the user corresponding to the user account and the dialogue partner. For example, the dialogue partner could be a broker, and the user account could be an identifier for the user served by that broker. The dialogue parameters of the dialogue partner are specifically implemented as duration parameters or dialogue content related to the dialogue partner. For example, dialogue parameters could be the continuous silence duration, continuous speaking duration, or speaking content of the dialogue partner, and so on.

[0069] Step 102: When the dialogue parameters meet the pre-set conditions for proactive dialogue, proactive dialogue content is generated based on the attribute information of the user account simulated by the chatbot.

[0070] In one implementation, the attribute information includes interest information; step 102 specifically includes: when the continuous silence duration of the dialogue object is greater than or equal to a preset silence threshold, generating proactive dialogue content based on a question template associated with the interest information.

[0071] Specifically: One or more interest information of the user account simulated by the chatbot can be pre-set, as well as one or more question templates corresponding to each interest information. For example, the user account's interest information includes two: (1) interest information 1, which represents the user's preference for three-bedroom apartment types; and (2) interest information 2, which represents the user's interest in tax rates. Among them, the question template for interest information 1 is: "What do you think of three-bedroom apartment types?"; and the question template for interest information 2 is: "What is the tax rate for this house?".

[0072] When the duration of continuous silence from the dialogue partner is greater than or equal to a pre-set silence threshold, the dialogue parameters are considered to meet the pre-set conditions for proactive dialogue. At this point, the chatbot can select a question template from the question templates of interest information 1 and interest information 2 based on a predetermined selection strategy (e.g., random selection or priority based on interest information), and generate proactive dialogue content based on that template. During the process of generating proactive dialogue content based on the question template, the question template can be directly designated as the proactive dialogue content, or it can be edited (e.g., adding or deleting content based on the context of the dialogue) to generate proactive dialogue content.

[0073] Therefore, the embodiments of the present invention can use the continuous silence duration of the dialogue object as the trigger condition for active dialogue, which can avoid the awkward situation of no dialogue for a long time.

[0074] In one implementation, the attribute information includes interest information; step 102 specifically includes: when the continuous speaking time of the dialogue object is greater than or equal to a preset speaking time threshold, generating proactive dialogue content based on a question template associated with the interest information.

[0075] Specifically: One or more interest information of the user account simulated by the chatbot can be pre-set, as well as one or more question templates corresponding to each interest information. For example, the user account's interest information includes two items: (1) interest information 1, which represents the user's interest in the age of the house; and (2) interest information 2, which represents the user's interest in the tax rate. Among them, the question template for interest information 1 is: "How many years has this house been used?"; the question template for interest information 2 is: "What is the tax rate for this house?".

[0076] When the duration of continuous narration by the dialogue partner is greater than or equal to a pre-set narration duration threshold, the dialogue parameters are considered to meet the pre-set conditions for proactive dialogue. At this point, the chatbot can select a question template from the question templates of interest information 1 and interest information 2 based on a predetermined selection strategy (e.g., random selection or priority based on interest information), and generate proactive dialogue content based on that question template. During the process of generating proactive dialogue content based on the question template, the question template can be directly determined as the proactive dialogue content, or the question template can be edited (e.g., adding or deleting content based on the context of the dialogue) to generate proactive dialogue content.

[0077] Therefore, the embodiments of the present invention can use the duration of continuous narration by the dialogue partner as the triggering condition for active dialogue, which can avoid the awkward scenario of continuous one-sided dialogue.

[0078] In one implementation, the attribute information includes interest information; step 102 specifically includes: when the content spoken by the dialogue object matches the interest information, generating proactive dialogue content based on the question template associated with the interest information.

[0079] One or more interest information of the user account simulated by the chatbot can be preset, as well as one or more question templates corresponding to each interest information.

[0080] Specifically, based on question templates associated with interest information, proactive dialogue content is generated, including:

[0081] (1) Generate proactive dialogue content based on individual question templates that match the content of the narrative:

[0082] For example, a user account's interest information includes two parts: (1) Interest information 1, representing the user's preference for three-bedroom apartments; and (2) Interest information 2, representing the user's interest in tax rates. The question template for interest information 1 is: "What do you think of a three-bedroom apartment?"; the question template for interest information 2 is: "What is the tax rate for this apartment?". When the content matches interest information 1 due to its association with "three-bedroom apartment", the corresponding question template for interest information 1, "What do you think of a three-bedroom apartment?", is selected, and active dialogue content is generated based on this question template. When the content matches interest information 2 due to its association with "tax rate", the question template for interest information 2, "What is the tax rate for this apartment?", is selected, and active dialogue content is generated based on this question template. Similarly, in the process of generating active dialogue content based on this question template, the question template can be directly determined as the active dialogue content, or the question template can be edited (e.g., adding or deleting content based on the context of the dialogue) to generate active dialogue content.

[0083] (2) From multiple question templates that match the content being told, filter out question templates that are closer to the content that has been told before based on context information, and generate active dialogue content based on the remaining question templates.

[0084] For example, a user account might have one interest profile, representing the user's preferred neighborhood A. The corresponding question templates include: Question Template 1: "Do you think it's a good time to buy in neighborhood A?"; Question Template 2: "What do you think of the three-bedroom apartment layout in neighborhood A?". When the content of a conversation matches the interest profile (e.g., the speaker mentions neighborhood A), it's considered a match. At this point, a question template can be selected from Question Template 1 and Question Template 2 based on a predetermined selection strategy, and proactive dialogue content can be generated based on that template. Specifically, during the selection process, the context of the conversation can be considered to filter out templates that are closer to previously mentioned content. For example, if the speaker has previously expressed their views on whether buying in neighborhood A is suitable, Question Template 1 will not be selected; instead, Question Template 2 will be chosen. Similarly, when generating proactive dialogue content based on Question Template 2, it can be directly selected as the proactive dialogue content, or it can be edited (e.g., adding or deleting content based on the context of the conversation to generate proactive dialogue content).

[0085] Therefore, the embodiments of the present invention can use the content of the narrative as a triggering condition for proactive dialogue, enabling in-depth dialogue and communication.

[0086] In one implementation, step 102 specifically includes: determining contextual information during the dialogue; determining the interest factors already mentioned by the dialogue partner based on the contextual information; removing the interest factors from the interest information to obtain the remaining interest information; and generating proactive dialogue content based on the question template associated with the remaining interest information.

[0087] Specifically: One or more interest information of the user account simulated by the chatbot can be pre-set, as well as one or more question templates corresponding to each interest information. For example, the user account's interest information includes three, namely: (1) interest information 1 representing the user's interest in the age of the house; (2) interest information 2 representing the user's interest in the tax rate; (3) interest information 3 representing the user's preference for three-bedroom apartment type. Among them: the question template for interest information 1 is: "How many years has this house been used?"; the question template for interest information 2 is: "What is the tax rate of this house?"; the question template for interest information 3 is: "What do you think of the three-bedroom apartment type?".

[0088] When the contextual information during the conversation includes topics related to three-bedroom apartments, the interest factor already mentioned by the conversation partner is determined to be: three-bedroom apartments. Then, interest information 3 related to three-bedroom apartments is removed from the user's account interest information, leaving interest information 1 and interest information 2. Then, based on the question templates of interest information 1 and interest information 2, proactive dialogue content is generated.

[0089] As can be seen, the embodiments of the present invention can remove interest information using contextual information, thus avoiding repeated discussions of the same topic.

[0090] Step 103: Play the active dialogue content based on an active dialogue mode that is independent of the current dialogue state of the dialogue partner.

[0091] Here, the meaning of the active dialogue mode, which is independent of the current dialogue state of the other party, is that regardless of whether the other party is currently speaking or silent, the active dialogue content is played immediately by voice, that is, there is no need to pay attention to the current dialogue state of the other party, especially no need to wait for the other party to finish speaking.

[0092] In one implementation, before playing the active dialogue content based on an active dialogue mode independent of the current dialogue state of the dialogue object, the method further includes: performing speech recognition processing on the speech of the dialogue object during the dialogue process to generate text content of the speech; performing semantic recognition on the text content; determining the response content of the chatbot based on the semantic recognition result; and updating the active dialogue content to the response content when the dialogue parameters do not meet the active dialogue conditions, or when the dialogue parameters meet the active dialogue conditions and the number of times the chatbot initiates active dialogue exceeds a preset active dialogue number threshold.

[0093] As can be seen, the embodiments of the present invention can generate response content. When the dialogue parameters do not meet the conditions for active dialogue, the response content is played, thus ensuring compatibility with half-duplex mode. Moreover, when the dialogue parameters meet the conditions for active dialogue and the number of times the chatbot initiates active dialogue exceeds a preset threshold for the number of active dialogues, the response content is still played instead of initiating an active dialogue to play the content, thereby avoiding frequent interruptions to the conversation.

[0094] In one embodiment, the method further includes: detecting a silence signal fragment of the dialogue object during the dialogue process; sending N consecutive silence signal fragments to the server, where N is a pre-set positive integer value; wherein on the server, the continuous silence duration is determined based on the total duration of the N silence signal fragments; wherein at least one of N, the silence threshold value, and the duration of the silence signal fragments is adjustable.

[0095] As can be seen, by sending the silence signal in segments to the server in a centralized manner, the server can avoid frequent response requests, thus reducing its processing load. Furthermore, since N, the silence threshold, and the duration of the silence signal segments are all adjustable, scalable dialogue control is achieved. For example, when the server is under heavy load, N can be increased to reduce the number of requests, at the cost of decreased sensitivity in the robot's proactive dialogue; conversely, N can be decreased to increase the sensitivity of the robot's proactive dialogue.

[0096] Figure 3 This is an exemplary schematic diagram of the dialogue processing logic of the chatbot according to an embodiment of the present invention.

[0097] exist Figure 3 In the dialogue, the dialogue subject 11 emits a voice message, and the dialogue processing logic of the dialogue robot 12 specifically includes:

[0098] (1) ASR: After preprocessing such as sentence segmentation, speech is converted into text to facilitate machine understanding; and a silence signal is sent to the data processing. Among them, ASR performs segmented caching of silence signals. For example, depending on the server performance, a silence signal is sent to the server once every 10 segments to avoid excessive request pressure on the server.

[0099] (2) Data processing, specific execution: (2.1): Perform sentence segmentation on the text to ensure that the NLU receives relatively complete segments. (2.2): Maintain global context variables during the session, such as the duration of silence, the duration of user statements, and the number of times the user has initiated a conversation, etc.

[0100] (3) NLU: Under the unified control of the dialogue control, extract semantic information from the text, such as emotion, sentence structure (interrogative or declarative), intention and slot.

[0101] (4) DM: Includes Dialogue Measurement (DP) and Dialogue State Tracking (DST), used to provide response strategies under the unified control of the dialogue center, combining conversation history, current semantic information, and relevant data. For example, it can provide answers to common questions based on context (Q / A question and answer) or generate response content based on a multi-way recall and reordering algorithm.

[0102] (5) Question detection: continuously detect the dialogue parameters of the dialogue object 11 during the dialogue process between the dialogue robot 12 and the dialogue object 11; when the dialogue parameters meet the conditions for active dialogue, generate active dialogue content that is suitable for asking questions based on the attribute information of the user account simulated by the dialogue robot 12.

[0103] (6) Response Decision: Used to select one output from the response content and the active dialogue content, wherein: when the dialogue parameters do not meet the active dialogue conditions, the response content is selected as the output; when the dialogue parameters meet the active dialogue conditions and the number of times the chatbot initiates active dialogue exceeds the active dialogue number threshold, the response content is selected as the output; when the dialogue parameters meet the active dialogue conditions and the number of times the chatbot initiates active dialogue does not exceed the active dialogue number threshold, the active dialogue content is selected as the output.

[0104] (7) NLG: Combines contextual information to convert the output of the response decision into natural language text.

[0105] (8) TTS: Convert natural language text into speech signals and play the speech signals.

[0106] As can be seen, the embodiments of the present invention improve upon the traditional pipelined architecture by adding data processing and question detection processing. During the dialogue, ASR does not ignore silence information but sends silence information to the data processing at certain time units (e.g., 200ms), reducing the request volume by appropriately caching silence signals. Each request updates the timer (e.g., silence duration, speaking duration, number of times the dialogue has been initiated). Question detection updates the silence information (continuous duration) according to the timer, while simultaneously recording the intent already stated by the dialogue object through context. When the conditions for initiated dialogue are not met, the robot responds normally; once the conditions for initiated dialogue are met (e.g., silence timeout (e.g., silence for 2 seconds), the dialogue object speaks for too long, or an interest is hit), the initiated question logic is triggered. In the initiated question logic, the robot actively asks a question, at which point the robot can interrupt the current statement of the dialogue object.

[0107] Figure 3 The dialogue processing logic enables the robot to have full-duplex conversation capabilities, allowing it to both passively answer questions and actively ask them. It does not require predefined real-time dialogue templates and can adjust the sensitivity by setting a silent buffer window (e.g., 10 segments) based on server performance. This not only simulates full-duplex dialogue mode well but also achieves low cost and controllable content.

[0108] Can Figure 3 The dialogue processing logic shown can be applied to various scenarios. For example, it can be applied to live streaming scenarios where a chatbot simulates a customer training a broker.

[0109] Figure 4 This is an exemplary schematic diagram illustrating the use of a conversational robot in training brokers according to an embodiment of the present invention. Figure 4 In the middle, chatbot 12 simulates a customer. Agent 11 issues a voice message to begin introducing the property. The dialogue processing logic of chatbot 12 specifically includes:

[0110] (1) ASR: After preprocessing such as sentence segmentation, speech is converted into text to facilitate machine understanding; and a silence signal is sent to the data processing. Among them, ASR performs fragmented caching of silence signals. For example, depending on the server performance, a single silence signal is sent to the server to avoid excessive request pressure on the server.

[0111] (2) Data processing, specific execution: (2.1): Perform sentence segmentation on the text to ensure that the NLU receives relatively complete segments. For example, the real-time text stream: "How much down payment are you prepared? I'll see if it's enough ___" (where __ indicates being interrupted) is segmented into 3 segments: "How much down payment are you prepared?", "I'll see if it's enough ___", and a mute signal; (2.2): Maintain global context variables during the conversation, such as the duration of mute, the duration of user statements, and the number of times the user has initiated a conversation, etc.

[0112] (3) NLU: Extract semantic information from text, such as mood, sentence structure (interrogative or declarative), intent, and slots.

[0113] (4) DM: Includes DP and DST, used to provide response strategies by combining conversation history, current semantic information, and relevant data. The chatbot 12 can support single-turn or multi-turn conversations, with multi-turn conversations driven by a preset finite state machine. In multi-turn conversations, the chatbot 12 uses a multi-path recall and reordering process to generate response content. Recall channels include sub-functions such as classification, retrieval, and matching. For example, if a user asks "How much down payment are you prepared?", after the NLU identifies the user's intention to buy a house, it recalls their respective answer templates through multiple channels such as rules, classification, retrieval, and matching. After scoring and ranking each answer template, the highest-scoring answer template (e.g., approximately XX0,000) is selected as the final response content.

[0114] (5) Question Detection: Detects the dialogue parameters of the dialogue partner 11 during the dialogue between the chatbot 12 and the agent 11. When the dialogue parameters meet the conditions for proactive dialogue, proactive dialogue content adapted to ask questions is generated based on the attribute information of the simulated customer by the chatbot 12. The customer attribute information can be obtained from the customer's user profile. The conditions for proactive dialogue include: 1. Continuous silence timeout (e.g., >5s); 2. Agent's narration timeout (e.g., >3 minutes); 3. Hitting the customer's concerns (e.g., house taxes). Proactive dialogue content can be generated using question templates, where the wording of the question templates is related to the customer's interests in the profile information.

[0115] (6) Response Decision: Used to select one output from the response content and the active dialogue content, wherein: when the dialogue parameters do not meet the active dialogue conditions, the response content is selected as the output; when the dialogue parameters meet the active dialogue conditions and the number of times the chatbot initiates active dialogue exceeds the active dialogue number threshold, the response content is selected as the output; when the dialogue parameters meet the active dialogue conditions and the number of times the chatbot initiates active dialogue does not exceed the active dialogue number threshold, the active dialogue content is selected as the output.

[0116] (7) NLG: Combines contextual information to convert the output of the response decision into natural language text.

[0117] (8) TTS: Convert natural language text into speech signals and play the speech signals.

[0118] Assuming that agent 11's presentation time exceeds a predetermined threshold, thus meeting the conditions for an initiating dialogue, agent 11 says, "How much down payment are you prepared? Let me see if it's enough ___" (where __ indicates being interrupted). The chatbot 12 then triggers its initiating question logic. Chatbot 12 queries the customer's user profile, discovers interest in tax rates, and outputs the question, "What is the tax rate for this house?", thus interrupting agent 11's presentation.

[0119] Figure 5 This is an exemplary structural diagram of the active dialogue device of the dialogue robot according to an embodiment of the present invention. Figure 5 As shown, the active dialogue device 500 of the chatbot includes:

[0120] Detection module 501 is used to detect the dialogue parameters of the dialogue object during the dialogue process between the chatbot and the dialogue object;

[0121] The generation module 502 is used to generate active dialogue content based on the attribute information of the user account simulated by the chatbot when the dialogue parameters meet the preset active dialogue conditions.

[0122] The playback module 503 is used to play active dialogue content via voice.

[0123] In an exemplary embodiment, the attribute information includes interest information; the generation module 502 is configured to perform at least one of the following: when the continuous silence duration of the dialogue object is greater than or equal to a preset silence threshold, generate proactive dialogue content based on a question template associated with the interest information; when the continuous speaking duration of the dialogue object is greater than or equal to a preset speaking duration threshold, generate proactive dialogue content based on a question template associated with the interest information; when the speaking content of the dialogue object matches the interest information, generate proactive dialogue content based on a question template associated with the interest information.

[0124] In an exemplary implementation, the generation module 502 is used to determine contextual information during the dialogue process; based on the contextual information, determine the interest factors that the dialogue object has mentioned; remove the interest factors from the interest information to obtain the remaining interest information; and generate proactive dialogue content based on the question template associated with the remaining interest information.

[0125] In an exemplary embodiment, the generation module 502 performs speech recognition processing on the speech of the dialogue object during the dialogue process before the playback module plays the active dialogue content in a voice manner to generate text content of the speech; performs semantic recognition on the text content; determines the response content of the chatbot based on the semantic recognition result; and updates the active dialogue content to the response content when the dialogue parameters do not meet the active dialogue conditions, or when the dialogue parameters meet the active dialogue conditions and the number of times the chatbot initiates active dialogue exceeds a preset active dialogue number threshold.

[0126] In an exemplary embodiment, the detection module 501 is used to detect the silence signal segments of the dialogue object during the dialogue process; to send N consecutive silence signal segments to the server, where N is a preset positive integer value; wherein on the server, the continuous silence duration is determined based on the total duration of the N silence signal segments; wherein at least one of N, the silence threshold value, and the duration of the silence signal segments is adjustable.

[0127] In summary, in this embodiment of the invention, the dialogue parameters of the dialogue object are detected during the dialogue process between the chatbot and the dialogue object; when the dialogue parameters meet the preset active dialogue conditions, active dialogue content is generated based on the attribute information of the user account simulated by the chatbot; and the active dialogue content is played based on an active dialogue mode independent of the current dialogue state of the dialogue object. Therefore, the chatbot of this embodiment can initiate active dialogue using the dialogue parameters of the dialogue object, realizing full-duplex human-machine dialogue and more accurately simulating the user communicating with the chat object.

[0128] Moreover, the embodiments of the present invention generate proactive dialogue content based on question templates associated with interest information, thereby making the dialogue content controllable and reducing the risk of use.

[0129] Furthermore, since N, the silence threshold, and the duration of the silence signal segmentation are all adjustable, this embodiment of the invention also achieves scalable dialogue control. For example, when the server-side pressure is high, N can be increased to reduce the request volume; conversely, N can be decreased to improve the sensitivity of the robot's proactive dialogue.

[0130] This invention also provides a computer-readable medium storing instructions that, when executed by a processor, can perform the steps of the above-described active dialogue method for a chatbot. In practical applications, the computer-readable medium may be included in the device / apparatus / system described in the above embodiments, or it may exist independently and not assembled into that device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, can implement the active dialogue method for a chatbot described in the above embodiments. According to the embodiments disclosed in this invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof, but this is not intended to limit the scope of protection of this invention. In the embodiments disclosed in this invention, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0131] like Figure 6 As shown, embodiments of the present invention also provide an electronic device in which the apparatus for implementing the method of embodiments of the present invention can be integrated. For example... Figure 6 As shown, it illustrates an exemplary structural diagram of an electronic device according to an embodiment of the present invention.

[0132] Specifically, the electronic device may include a processor 601 with one or more processing cores, a memory 602 with one or more computer-readable storage media, and a computer program stored in the memory and executable on the processor. When the program in the memory 602 is executed, the aforementioned media stream transmission method can be implemented.

[0133] In practical applications, this electronic device may also include components such as a power supply 603, an input unit 604, and an output unit 605. Those skilled in the art will understand that... Figure 6The structure of the electronic device shown does not constitute a limitation on the device and may include more or fewer components than shown, or combine certain components, or have different component arrangements. The processor 601 is the control center of the electronic device, connecting various parts of the device via various interfaces and lines. It performs overall monitoring of the electronic device by running or executing software programs and / or modules stored in memory 602, and by calling data stored in memory 602, thereby executing various server functions and processing data. Memory 602 can be used to store software programs and modules, i.e., the aforementioned computer-readable storage medium. The processor 601 executes various functional applications and data processing by running the software programs and modules stored in memory 602. Memory 602 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created according to the use of the server, etc. Furthermore, memory 602 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, memory 602 may also include a memory controller to provide processor 601 with access to memory 602.

[0134] The electronic device also includes a power supply 603 that supplies power to various components. This power supply is logically connected to the processor 601 via a power management system, enabling functions such as charging, discharging, and power consumption management. The power supply 603 may also include one or more DC or AC power supplies, a recharging system, a power fault detection circuit, a power converter or inverter, a power status indicator, or other arbitrary components. The electronic device may also include an input unit 604, which can receive input digital or character information and generate keyboard, mouse, joystick, optical, or trackball signal inputs related to user settings and function control. The electronic device may also include an output unit 605, which can display information input by the user or information provided to the user, as well as various graphical user interfaces (GUIs), which can be composed of graphics, text, icons, video, and any combination thereof.

[0135] This invention also provides a computer program product including computer instructions that, when executed by a processor, perform the methods described in any of the above embodiments.

[0136] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments disclosed in this invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in the order shown in different figures. For example, two blocks shown connected together may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0137] This document describes the principles and implementation methods of the present invention using specific embodiments. These embodiments are merely illustrative of the method and core concepts of the present invention and are not intended to limit the invention. Those skilled in the art can make changes to the specific implementation methods and application scope based on the ideas, spirit, and principles of the present invention. Any modifications, equivalent substitutions, or improvements made should be included within the scope of protection of this invention.

Claims

1. A method for proactive dialogue in a chatbot, characterized in that, include: Detect the dialogue parameters of the dialogue object during the dialogue process between the chatbot and the dialogue object; When the dialogue parameters meet the preset active dialogue conditions, active dialogue content is generated based on the attribute information of the user account simulated by the chatbot. The attribute information includes interest information. Play the active dialogue content based on an active dialogue mode that is independent of the current dialogue state of the dialogue object; When the dialogue parameters meet the preset active dialogue conditions, the generated active dialogue content includes at least one of the following: When the duration of continuous silence of the dialogue object is greater than or equal to a preset silence threshold, the proactive dialogue content is generated based on a question template associated with the interest information. When the continuous speaking time of the dialogue object is greater than or equal to a preset speaking time threshold, the active dialogue content is generated based on the question template associated with the interest information. When the content of the dialogue object matches the interest information, the proactive dialogue content is generated based on the question template associated with the interest information.

2. The active dialogue method for a chatbot according to claim 1, characterized in that, The generation of the proactive dialogue content based on the question template associated with the interest information includes: Determine the context information during the dialogue process; Based on the context information, determine the interest factors that the dialogue partner has already described; Remove the interest factors from the interest information to obtain the remaining interest information; The proactive dialogue content is generated based on the question template associated with the remaining interest information.

3. The active dialogue method for a chatbot according to claim 1, characterized in that, Before playing the active dialogue content based on an active dialogue mode independent of the current dialogue state of the dialogue object, the method further includes: Speech recognition processing is performed on the speech of the dialogue partner during the dialogue process to generate the text content of the speech; Perform semantic recognition on the text content; Based on the semantic recognition results, the response content of the chatbot is determined; When the dialogue parameters do not meet the conditions for proactive dialogue, or when the dialogue parameters meet the conditions for proactive dialogue but the number of times the chatbot initiates proactive dialogue exceeds a preset threshold, the proactive dialogue content will be updated to the response content.

4. The active dialogue method for a chatbot according to claim 1, characterized in that, Also includes: Detect the silence signal fragments of the dialogue object during the dialogue process; The N consecutive mute signals are divided into segments and sent to the server in a concentrated manner, where N is a pre-set positive integer value; On the server side, the continuous silence duration is determined based on the total duration of the N silence signal segments. At least one of N, the mute threshold, and the time length of the mute signal segment is adjustable.

5. An active dialogue device for a conversational robot, characterized in that, include: The detection module is used to detect the dialogue parameters of the dialogue object during the dialogue process between the chatbot and the dialogue object; The generation module is used to generate active dialogue content based on the attribute information of the user account simulated by the chatbot when the dialogue parameters meet the preset active dialogue conditions. The attribute information includes interest information. The playback module is used to play the active dialogue content based on an active dialogue mode that is independent of the current dialogue state of the dialogue object; The generation module is configured to perform at least one of the following: When the duration of continuous silence of the dialogue object is greater than or equal to a preset silence threshold, the proactive dialogue content is generated based on a question template associated with the interest information. When the continuous speaking time of the dialogue object is greater than or equal to a preset speaking time threshold, the active dialogue content is generated based on the question template associated with the interest information. When the content of the dialogue object matches the interest information, the proactive dialogue content is generated based on the question template associated with the interest information.

6. A computer-readable storage medium storing computer instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the steps of the active dialogue method of the chatbot according to any one of claims 1 to 4.

7. An electronic device, characterized in that, The electronic device includes: processor; Memory for storing the executable instructions of the processor; The processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the active dialogue method of the chatbot according to any one of claims 1 to 4.

8. A computer program product, characterized in that, It includes computer instructions that, when executed by a processor, implement the active dialogue method of the chatbot according to any one of claims 1 to 4.

Images