Intelligent session method and device

Abstract

The invention discloses an intelligent session method and device, and relates to the technical field of computers. A specific embodiment of the method comprises the steps of monitoring a session between a user and an artificial customer service, and when the session meets a preset triggering condition, determining a session intervention scene and session intervention logic according to a session intention of the user; generating an intermediate response according to the session intervention scene and the session intervention logic; the intermediate response is displayed on a session interface; and in response to an operation triggered by the user based on the intermediate response, generating a response result. According to the embodiment, the manual customer service and the machine customer service can be fused, and the response can be automatically performed according to the determined session intervention scene and session intervention logic, so that the session cost is reduced, and the session efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to an intelligent conversation method and device. Background Art

[0002] At present, intelligent conversation services are mainly divided into manual customer service conversations and machine customer service conversations. The two conversation modes are independent of each other. During manual customer service conversations, if the response time of manual customer service exceeds the processing upper limit, it is easy for processing to be delayed. During machine customer service conversations, the existing machine customer service intelligence level and handling capabilities are insufficient, and cannot provide service capabilities to solve scenario-based problems. At this time, the system still switches to manual customer service for response based on the request sent by the user to transfer to manual customer service, and the conversation efficiency is low.

[0003] In the process of implementing the present invention, the inventors found that there are at least the following problems in the related art:

[0004] Human customer service and machine customer service cannot be integrated, and the conversation cost is high and the efficiency is low. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides an intelligent conversation method and device, which can integrate human customer service and machine customer service, and can automatically respond according to the determined conversation intervention scenario and conversation intervention logic, thereby reducing conversation costs and improving conversation efficiency.

[0006] To achieve the above object, according to one aspect of an embodiment of the present invention, an intelligent conversation method is provided, including:

[0007] Monitor the conversation between the user and the manual customer service, and when the conversation meets the preset trigger condition, determine the conversation intervention scenario and conversation intervention logic according to the user's conversation intention;

[0008] Generate an intermediate response according to the conversation intervention scenario and the conversation intervention logic; and display the intermediate response on the conversation interface;

[0009] In response to an operation triggered by the user based on the intermediate response, a response result is generated.

[0010] Furthermore, the preset triggering condition includes:

[0011] The response time of the manual customer service exceeds a preset threshold or the user's level is a preset level.

[0012] Furthermore, before determining the conversation intervention scenario and the conversation intervention logic according to the conversation intention of the user, the method includes:

[0013] Perform semantic recognition on the historical conversation information between the user and the manual customer service to obtain semantic recognition results;

[0014] Based on the semantic recognition result and a preset scenario event library, the user's conversation intention is determined.

[0015] Further, based on the semantic recognition result and a preset scene event library, determining the user's conversation intention includes:

[0016] Determining event data corresponding to the semantic recognition result from the preset scene event library;

[0017] The semantic recognition results and event data are input into a pre-trained intent recognition model to obtain the user's conversation intent.

[0018] Furthermore, the pre-trained intent recognition model is obtained by the following steps:

[0019] Obtain historical event data and historical text data sent by users;

[0020] determining an event weight based on the historical event data;

[0021] Based on the event weights, historical event data and historical text data, an intent recognition model is trained.

[0022] Furthermore, determining a conversation intervention scenario and a conversation intervention logic according to the conversation intention of the user includes:

[0023] Determining a conversation intervention scenario based on the user's conversation intention and a pre-set matching rule;

[0024] The conversation intervention logic is determined according to the scenario parameters and conditional rules corresponding to the intervention scenario.

[0025] Furthermore, before monitoring the conversation between the user and the manual customer service, it includes:

[0026] Make sure the session has the add flag.

[0027] According to a second aspect of an embodiment of the present invention, there is provided an intelligent conversation device, comprising:

[0028] A conversation monitoring module is used to monitor the conversation between the user and the manual customer service, and when the conversation meets the preset trigger condition, determine the conversation intervention scenario and conversation intervention logic according to the user's conversation intention;

[0029] An intermediate response module, used to generate an intermediate response according to the conversation intervention scenario and the conversation intervention logic; the intermediate response is displayed on the conversation interface;

[0030] The result response module is used to generate a response result in response to the user's operation triggered based on the intermediate response.

[0031] According to a third aspect of an embodiment of the present invention, there is provided an electronic device, including:

[0032] one or more processors;

[0033] a memory for storing one or more programs,

[0034] When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method described in any one of the above embodiments.

[0035] According to a fourth aspect of an embodiment of the present invention, a computer-readable medium is provided, on which a computer program is stored, and when the program is executed by a processor, the method described in any one of the above embodiments is implemented.

[0036] An embodiment of the above invention has the following advantages or beneficial effects: by monitoring the conversation between the user and the manual customer service, when the conversation meets the preset trigger condition, the conversation intervention scenario and the conversation intervention logic are determined according to the user's conversation intention; according to the conversation intervention scenario and the conversation intervention logic, an intermediate response is generated; the intermediate response is displayed on the conversation interface; in response to the user's operation triggered by the intermediate response, a response result is generated; thereby, manual customer service and machine customer service can be integrated, and responses can be automatically made according to the determined conversation intervention scenario and conversation intervention logic, thereby reducing conversation costs and improving conversation efficiency.

[0037] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with specific implementation examples. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0039] Figure 1 is a schematic diagram of the main process of the intelligent conversation method according to an embodiment of the present invention;

[0040] Figure 2 is a schematic diagram of the main process of an intelligent conversation method according to a reference embodiment of the present invention;

[0041] Figure 3 is a schematic diagram of the main process of an intelligent conversation method according to another reference embodiment of the present invention;

[0042] Figure 4 is a schematic diagram of the main process of an intelligent conversation method according to another reference embodiment of the present invention;

[0043] Figure 5is a schematic diagram of the main process of an intelligent conversation method according to another reference embodiment of the present invention;

[0044] Figure 6 It is a schematic diagram of the main process of the intelligent conversation method of a preferred embodiment of the present invention;

[0045] Figure 7 It is a flowchart of a conversation intervention scenario hit according to a preferred embodiment of the present invention;

[0046] Figure 8 is a schematic diagram of the order cancellation session intervention logic according to a preferred embodiment of the present invention;

[0047] Fig. 9 is a schematic diagram of main modules of an intelligent conversation device according to an embodiment of the present invention;

[0048] Fig.10 is an exemplary system architecture diagram to which embodiments of the present invention may be applied;

[0049] Fig.11 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0051] It should be noted that the acquisition, storage and application of personal information etc. involved in the embodiments of the present invention are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0052] Taking online shopping malls, online teaching Q&A, online medical care, auto repair maintenance and other scenarios as examples, online systems often provide users with consulting, teaching Q&A, medical consultation and other services. During the service process, users can initiate consultations with human customer service or machine customer service (here, reception staff in different fields such as teaching teachers, auto repair engineers, doctors, and medical guides are collectively referred to as customer service) for the issues involved, and human customer service or machine customer service provides users with the service capabilities of corresponding question replies and solutions. In the consulting service session, users are supported to conduct Q&A interactions with human customer service or machine customer service and solve problems through the supporting system. The two service types and corresponding sessions are independent of each other and do not interfere with each other. In the human customer service session, often due to factors such as manpower input cost control, the human customer service response often exceeds the upper limit of manual processing and is not processed in time, resulting in a poor user experience. In the machine customer service session, often due to the current robot intelligence level and insufficient problem-solving capabilities, it is impossible to provide human-like services and scenario-based problem solving services. At the same time, when the machine customer service session cannot solve the problem, it is still transferred to human customer service, which will also increase service costs and reduce the efficiency of problem solving.

[0053] In view of this, according to one aspect of an embodiment of the present invention, an intelligent conversation method is provided.

[0054] Figure 1 FIG. 1 is a schematic diagram of the main process of the intelligent conversation method according to an embodiment of the present invention. Figure 1 As shown, the intelligent conversation method according to an embodiment of the present invention includes the following steps S101 to S103.

[0055] Step S101 , monitoring a conversation between a user and a human customer service representative, and when the conversation meets a preset trigger condition, determining a conversation intervention scenario and a conversation intervention logic according to the conversation intention of the user.

[0056] The user refers to an individual, organization or entity that interacts with a service or system. For example, a user may be an individual who visits a website or application to seek help, ask questions or perform specific operations. The human customer service refers to a professional who is responsible for answering user questions or providing service support. The human customer service may communicate with the user in real time over the phone, chat window or other communication channels to answer questions, provide suggestions or process requests.

[0057] To monitor the conversation between the user and the manual customer service, the established real-time monitoring system can be integrated into the communication platform or application, and the real-time monitoring system can be used to capture the real-time conversation between the user and the manual customer service. The conversation content between the user and the customer service can also be saved in a database or recorded through a log file, and the system can retrieve and view the session information from the above database and log file at any time.

[0058] During the monitoring process, natural language processing technology can also be used to conduct real-time analysis of the conversation between the user and the manual customer service, such as text classification, sentiment analysis, intent recognition and other technologies, to pre-determine the user's intention, emotion and conversation scenario, so as to intervene quickly and improve the system's response processing speed and efficiency. In addition, the user's conversation intention can also be determined when the conversation meets the preset trigger conditions to reduce the system's monitoring cost. For example, the historical conversation information between the user and the manual customer service can be analyzed using natural language processing technology to extract the user's intention; for example, the historical conversation information between the user and the manual customer service can be matched with a pre-established keyword library to infer the user's intention.

[0059] Trigger conditions can be set based on the content of the conversation, for example, intervention is triggered when the user mentions the same question or keyword multiple times; trigger conditions can also be set based on the recognized emotional changes of the user, for example, intervention is triggered when the user expresses dissatisfaction or abnormal emotions; trigger conditions can also be set based on the duration of the conversation, for example, intervention is triggered when the length of time that the manual customer service does not respond to a conversation exceeds a threshold; or trigger conditions can be set based on specific events, for example, intervention is triggered when the user submits a specific form, selects an option, or clicks a button; or trigger conditions can be set based on the user's level, for example, intervention is triggered when the user is at a member level, and so on.

[0060] The intervention scenario refers to a specific behavior or operation that is automatically triggered according to the triggering conditions, which is used to change the progress of the session or provide corresponding help, such as canceling an order scenario, modifying an order scenario, etc. The session intervention logic refers to the specific operations and processing procedures performed by the system after the intervention scenario is triggered. The corresponding intervention operations can be performed according to the preset intervention scenarios and priorities, such as canceling an order, modifying an order, sending an automatic reply, transferring a session, providing help information, etc.

[0061] Step S102: Generate an intermediate response according to the conversation intervention scenario and the conversation intervention logic; the intermediate response is displayed on the conversation interface.

[0062] The intermediate response refers to the session process response generated after executing the session intervention logic. The session interface refers to the interface for the user to have a conversation with the manual customer service, which can be a chat window, a message page of an application, etc.

[0063] According to the preset intervention scenario and intervention logic, the system can automatically generate intermediate responses in the form of text, images, links or buttons, that is, the intermediate responses can include clear questions or prompts and operation options or buttons, etc. For example, for the intervention scenario of canceling an order, the following intermediate response can be generated: "Are you sure you want to cancel the order?" and provide relevant operation buttons. For another example, for the intervention scenario of modifying an order, prompt information such as "Are you sure you want to modify the order?" or "Please provide the order number to confirm the modification request" is generated, and operation buttons such as "Confirm modification" and "Do not modify for the time being" are provided to the user.

[0064] Intermediate responses can be displayed immediately or continuously on the conversation interface, and users can view, reply or operate in real time. Specifically, the intermediate responses are displayed on the conversation interface in the form of prompt boxes or bubbles to distinguish between user input and system-generated intermediate responses. When displaying intermediate responses, animations or special effects can also be added to improve user attention and interactive experience.

[0065] Step S103: generating a response result in response to the user's operation triggered based on the intermediate response.

[0066] The system monitors the user's operation on the intermediate response in real time, for example, the user clicks the "Confirm Cancel Order" button or selects other options; according to the user's operation, the system performs the corresponding business operation and generates the corresponding response result. Taking the cancellation order scenario as an example, if the user clicks the "Confirm Cancel Order" button, the system performs the order cancellation operation and generates the corresponding response result, such as "The order has been successfully canceled" or "Sorry, the order cancellation failed, please try again later"; if the user selects the "Do not cancel for now" button, the system can generate a corresponding prompt message, such as "OK, if you need to cancel the order, please let me know at any time", and continue to monitor the user's operation.

[0067] The generated response result is fed back to the user, such as displaying the response result on the conversation interface in the form of text messages, images, links or buttons, so that the user can clearly understand that the operation has been performed or provide further information. If an abnormal situation occurs during the execution of the operation, such as system failure or network problem, the system can handle it in time and provide the user with corresponding prompts or solutions, such as re-execute the operation, provide contact information or suggest the user to try again later.

[0068] The embodiment of the present invention monitors the conversation between the user and the manual customer service, and when the conversation meets the preset trigger condition, determines the conversation intervention scenario and the conversation intervention logic according to the user's conversation intention; generates an intermediate response according to the conversation intervention scenario and the conversation intervention logic; the intermediate response is displayed on the conversation interface; and generates a response result in response to the user's operation triggered by the intermediate response. The manual customer service and the machine customer service can be integrated, and the response can be automatically made according to the determined conversation intervention scenario and the conversation intervention logic, thereby reducing the conversation cost and improving the conversation efficiency.

[0069] Optionally, the preset trigger condition includes: the response time of the manual customer service exceeds a preset threshold or the user's level is a preset level.

[0070] The system can monitor the duration of the conversation between the human customer service and the user, and trigger the intervention operation when it exceeds the preset threshold. For example, the system sets the human customer service response time to no more than 3 minutes. When the human customer service response time exceeds 3 minutes, the system can automatically trigger the intervention scenario. The system can identify the user's level, and trigger the intervention operation when the identified user's level is the preset level, for example, the preset level is the membership level, etc.

[0071] By setting preset trigger conditions, the embodiment of the present invention allows the system to process conversations more intelligently and perform personalized intervention and processing according to different conditions, thereby improving service efficiency and user experience.

[0072] Figure 2 FIG. 1 is a schematic diagram of the main process of an intelligent conversation method according to a reference embodiment of the present invention. Figure 2 As shown, before determining the conversation intervention scenario and the conversation intervention logic according to the user's conversation intention, step S201 and step S202 are included.

[0073] Step S201: perform semantic recognition on historical conversation information between the user and the manual customer service to obtain a semantic recognition result.

[0074] Step S202: determining the user's conversation intention based on the semantic recognition result and a preset scenario event library.

[0075] In the above step S201, the historical conversation information is first preprocessed, and the preprocessing includes removing special characters, stop words, stemming, etc. to reduce noise and improve the accuracy of semantic recognition; then, natural language processing technology is used to extract features such as keywords, entities, phrases, etc. in the historical conversation information; a deep learning model or rule-based semantic analysis technology is used to perform semantic understanding on the historical conversation information that has been preprocessed and feature extracted, identify user intentions, emotional tendencies, etc., and generate semantic recognition results, which include information such as user intentions, topic keywords, emotional polarity, etc.

[0076] The pre-set scenario event library refers to a set of possible user intentions, corresponding processing logic, answer templates and other information that are pre-defined and collected for specific application scenarios or tasks. For each scenario or task, the scenario database lists possible user intentions, such as querying information, submitting requests, seeking help, and filing complaints; for each user intention, the corresponding processing logic and process are defined, such as how the system should respond, how to process user requests, and how to guide users to the next step; for each user intention and processing logic, the corresponding answer template is set. The answer template includes text replies, graphic messages, link recommendations, operation instructions, etc., so that the system can quickly give an appropriate response after identifying the user's intention.

[0077] In step S202, the semantic recognition result obtained in step S201 is matched with the scene event library to find the best matching intent. Specifically, a rule-based matching algorithm, a fuzzy matching algorithm or a machine learning model can be used for intent matching; based on the matching result, the user's session intent is determined, and the user's request is further processed and responded to. If the matching result is uncertain or the matching degree is low, the system can make further inquiries or provide other alternatives to improve the accuracy of intent recognition.

[0078] Figure 3 FIG. 1 is a schematic diagram of the main process of an intelligent conversation method according to another reference embodiment of the present invention. Figure 3 As shown, based on the semantic recognition result and the preset scene event library, the user's conversation intention is determined, including step S301 and step S302.

[0079] Step S301, determining event data corresponding to the semantic recognition result from the preset scene event library.

[0080] Step S302: input the semantic recognition result and event data into a pre-trained intention recognition model to obtain the user's conversation intention.

[0081] Specifically, step S301 determines that the event data corresponding to the semantic recognition result can be achieved through text matching and entity recognition. First, the semantic recognition result is text matched, and the event data matching the semantic recognition result is found in the pre-set scene event library by using methods such as keyword matching and fuzzy matching; secondly, the entities in the semantic recognition result are identified and extracted, and then matched with the entities in the scene event library to find the corresponding event data; finally, the best matching event data is selected based on factors such as matching degree and rule priority.

[0082] The above step S302 preprocesses the event data and semantic recognition results determined in step S301, and the preprocessing includes text cleaning, feature extraction, vectorization and other operations; uses a pre-trained intent recognition model, such as a rule-based model, a deep learning model or an integrated model, and uses the preprocessed semantic recognition results and event data as the input of the model. The user's session intention is obtained through model calculation, and the intent recognition model outputs the user's intention or the confidence or probability of the user's intention; according to the intention result output by the model, the user's specific intention and needs are analyzed, which can be a predefined intent label or a set of actions and parameters, such as query, submission, consultation, etc.

[0083] The embodiment of the present invention determines the task of user conversation intention based on semantic recognition results and a pre-set scenario event library, which can improve the accuracy and stability of intention recognition, thereby realizing intelligent conversation interaction.

[0084] Figure 4 FIG. 1 is a schematic diagram of the main process of an intelligent conversation method according to another reference embodiment of the present invention. Figure 4 As shown, the pre-trained intent recognition model is obtained through the following steps S401 to S403.

[0085] Step S401, obtaining historical event data and historical text data sent by the user.

[0086] Step S402: determining event weights based on the historical event data.

[0087] Step S403: training an intent recognition model based on the event weights, historical event data and historical text data.

[0088] The above step S401 can obtain historical event data and historical text data sent by users through log records, database storage or other data collection methods. Historical event data includes various events processed by the system and user requests, such as conversations initiated by users, system replies, user operations, etc. Historical text data is text information sent by users in conversations, including questions, requests, descriptions, etc.

[0089] The event weights in step S402 can be determined by statistical analysis, machine learning methods, or expert evaluation. For example, the frequency of occurrence, impact range, user feedback, and other indicators of various events in historical event data can be counted to calculate the weight values ​​of each event, and the weights can be adjusted when training the model based on the weight values.

[0090] The above step S403 first uses historical event data and text data as training sets, annotates corresponding intent labels, and then uses machine learning algorithms (such as support vector machines, deep neural networks, naive Bayes, etc.) to train the model. Event weights can be added to the model training process by weighted processing samples or adjusting the loss function to improve the model's attention to important events and recognition accuracy.

[0091] Figure 5 FIG. 1 is a schematic diagram of the main process of an intelligent conversation method according to another reference embodiment of the present invention. Figure 5 As shown, determining a conversation intervention scenario and a conversation intervention logic according to the conversation intention of the user includes step S501 and step S502.

[0092] Step S501: determining a conversation intervention scenario based on the user's conversation intention and a preset matching rule.

[0093] Step S502: determining the conversation intervention logic according to the scenario parameters and conditional rules corresponding to the intervention scenario.

[0094] Specifically, matching rules are formulated for each pre-set conversation intervention scenario, and the matching rules include aspects such as keyword matching, semantic matching, and intent confidence. For example, for the intent of "cancel order", the matching rule can be set as user input containing keywords such as "cancel" and "order". The above step S501 matches the user's conversation intent with the matching rules to find the most matching conversation intervention scenario. If there are multiple matching conversation intervention scenarios, they can be selected according to the preset priority; finally, the hit conversation intervention scenario is determined, the user's intent is recorded, the corresponding scene processing logic is triggered, and the corresponding service or function is called to respond to the user's request. For example, for the scenario of canceling an order, the system can trigger the conversation intervention logic for canceling the order, provide the user with the option of canceling the order, and process related operations.

[0095] Scenario parameters are set for each session intervention scenario. Scenario parameters include required input information, operation object, operation target, etc. For example, scenario parameters for canceling an order include order number, cancellation reason, etc. The session intervention logic is triggered according to different established conditional rules. The conditional rules include time conditions, status conditions, user information conditions, etc. For example, if the user initiates a request to cancel an order within a specified time, the session intervention logic for canceling an order is triggered. Step S502 determines the session intervention logic according to the scenario parameters and conditional rules. For example, confirm that the user's intention is to cancel the order, obtain the order number and cancellation reason provided by the user, the system checks the order status, and can only continue the cancellation operation when the order status is "pending" or "confirmed". Verify whether the user's cancellation request meets the conditional rules, including order status check and cancellation number limit. If the cancellation condition verification passes, the system executes the operation of canceling the order.

[0096] Optionally, before monitoring the conversation between the user and the manual customer service, the method includes: determining whether there is a hang-up mark in the conversation.

[0097] Determining that a session has a tag can be used to instruct the system to perform special processing on the session or have certain permissions. Tags can be set based on a variety of conditions, such as user identity, session content, trigger conditions, etc. Tags are used to grant special permissions or specific operating capabilities to a session. For example, for member users or important partners, the system can add special permission tags to indicate monitoring of the session, etc., to more flexibly and accurately manage and process the session between users and manual customer service, and improve the user experience.

[0098] Figure 6 FIG. 1 is a schematic diagram of the main process of the intelligent conversation method of a preferred embodiment of the present invention. Figure 6 As shown in the figure, the user sends a consultation message, and the message is transmitted to the conversation service module through the terminal processing module. The conversation service module assigns the message to the skill group, and the human customer service responds. The robot customer service listens to the response of the human customer service through the message monitoring module, and performs natural language processing, such as intent recognition, language recognition, and conversation intervention scenario judgment, and then executes the conversation intervention logic (scenario capability) to generate a response message, and displays the response message to the user through the conversation interface. The user receives the response information and generates operation information based on the response information. The robot customer service continues the above conversation process based on the operation information and generates the final response result.

[0099] The user initiates a consultation or operation request through the front-end conversation interface. The terminal processing module classifies customer messages into general consultation or intervention execution. General consultation messages are sent and received for processing, and intervention execution messages include touch and display, obstacle avoidance operation, and receipt display. The conversation service module assigns messages to the corresponding skill group and locates the customer service. The human customer service picks up the conversation and replies. The robot customer service monitors the conversation through the message monitoring module. The robot customer service can also receive general chat messages, perform text summarization, semantic recognition, intent recognition, and predict conversation intervention scenarios. When the trigger conditions are met, the conversation intervention logic is executed and the results are returned to the user. The intervention platform monitors the responses of the human customer service, performs rule matching, and handles intervention events.

[0100] Specifically, when a user initiates a consultation with a human customer service representative, the system receives the first question sent by the user, allocates human customer service personnel according to the currently configured customer service skill group and personnel, and creates a human customer service session. In the process of creating a human customer service session, the system determines whether to add a robot customer service representative based on the current service robot customer service situation, such as whether robot-assisted service is supported or whether the robot-assisted service capability includes the current problem coverage. If the addition is currently allowed, the addition mark is added to start the robot customer service representative.

[0101] After creating a human customer service session and adding a robot customer service, the robot customer service will monitor the current human customer service session (the earlier the monitoring, the more complete the model training and the more accurate the subsequent recognition). During the current session monitoring process, the robot customer service receives the messages sent by the user to the human customer service and the response messages from the human customer service to the user. After receiving the message, the robot customer service collects the message and processes the text summary. The robot customer service can also perform intervention identification, monitor the changes in the response time of the human customer service, and when the response time exceeds the threshold, it is considered that the human customer service response has timed out, triggering an intervention event; and the user session intention recognition is pre-placed, and semantic recognition is performed when receiving the message sent by the user, and combined with the intervention intention scenario event library classification, the user's final intention is identified. Specifically, the event library is designed according to the session intervention scenario, and the corresponding event report is added based on the event library. The reported event data is collected and written into the event library. After the event data is collected, the event weight and deep learning model are combined to classify and predict the event to identify the final intention.

[0102] As shown in Tables 1 to 5 below, based on self-service scenarios (cancelling orders, expediting shipments, etc.) as examples, combined with user behavior trajectories and consultation situations, data inference is performed (the following are only examples, but the actual application is not limited to the following examples)

[0103]

[0104]

[0105]

[0106]

[0107]

[0108] Based on the above user event library and text data collection, the best scene intent matching result is obtained through deep learning model training. As shown in Table 6 below, the scene semantic list includes but is not limited to scene classification and event name. Among them, the deep learning model can be BERT (Bidirectional Encoder Representations from Transformers) or MLP (Multilayer Perceptron), etc.

[0109]

[0110] The scenario is predicted based on the user's conversation intention, combined with semantic recognition, and matched according to the currently set rules, and finally the corresponding conversation intervention scenario is identified and the scenario is hit. The rule matching includes but is not limited to product status, order status, amount, merchant, etc.

[0111] Figure 7 is a flow chart of a conversation intervention scenario hit according to a preferred embodiment of the present invention; Figure 7 As shown in the figure, during the consultation message processing, key content is extracted from the consultation message to obtain a text summary, semantic recognition is performed to analyze the semantics of the text summary to understand the user's intention, and the message is matched with the predefined rules based on the recognized semantics to hit the corresponding conversation intervention scenario. During the event message processing, relevant event data is collected, and the machine learning model is trained using the event data. The user's intention is identified through the trained model, and the identified user intention is matched with the predefined rules to hit the corresponding conversation intervention scenario.

[0112] Figure 8 is a schematic diagram of the order session intervention logic cancellation according to a preferred embodiment of the present invention; Figure 8 As shown, the order cancellation session intervention scenario is only an example. According to the session intervention scenario, the session intervention logic (scenario capability) is set, and the session intervention logic is preset and the processing flow is closed. A single session intervention scenario may involve a single session intervention logic or multiple session intervention logics depending on the business. The session intervention logic can be a complete set of functions to complete the scenario function and business process, including front-end and back-end interactions based on the business and business logic processing.

[0113] After the session intervention scenario is hit, the session intervention logic hit (capability hit) is screened according to the corresponding scenario parameters and processing rule conditions. If there is only a single session intervention logic, it is directly hit according to the conditional rules, and a response is issued and the corresponding session intervention logic is triggered; if there are multiple session intervention logics, they are screened according to the conditional rules, and the session intervention logic that meets the conditional rules is matched and hit, and a response is issued and the corresponding session intervention logic is triggered according to the set logic.

[0114] After the session intervention logic is hit, the intervention execution is triggered. According to the current capability characteristics, different forms of session intervention scenarios and session intervention logic responses are issued. After the session intervention scenario response reaches the front-end session interface, the session intervention logic processing is completed through two or more interactions between the front and back ends, and the results are returned, the session intervention scenario problem is accurately solved, and the intervention is ended. After the scenario response reaches the front-end session interface, the front-end session interface interacts with the robot customer service for the second time through the preset session intervention scenario and session intervention logic; during the interaction, the necessary parameters for logic processing are submitted. After the robot customer service receives the operation, it performs operation processing according to the input parameters and processing identifiers; the robot customer service processes the business process by processing and calling the interface provided by the capability-dependent service; after the capability-dependent service completes the business logic processing, it returns the result; the robot customer service performs secondary processing based on the returned result and reaches the front-end session interface; the front-end session interface displays the processing result and the intervention is completed.

[0115] During a manual customer service conversation, the embodiment of the present invention actively mounts a robot customer service, performs conversation intervention scenarios and conversation intervention logic presets, monitors conversation records between users and manual customer service, dynamically identifies intervention service scenarios through semantic recognition, conversation context scenario parameters, and manual customer service response time, and actively provides scalable and interactive intervention capabilities to quickly and accurately solve user problems, thereby improving conversation efficiency.

[0116] According to a second aspect of the embodiment of the present invention, Figure 7 As shown, an intelligent conversation device 900 is provided, including a conversation monitoring module 901, an intermediate response module 902 and a result response module 903.

[0117] The conversation monitoring module 901 is used to monitor the conversation between the user and the manual customer service, and when the conversation meets the preset trigger condition, determine the conversation intervention scenario and conversation intervention logic according to the conversation intention of the user;

[0118] An intermediate response module 902 is used to generate an intermediate response according to the conversation intervention scenario and the conversation intervention logic; the intermediate response is displayed on the conversation interface;

[0119] The result response module 903 is used to generate a response result in response to the user's operation triggered based on the intermediate response.

[0120] Optionally, the preset triggering conditions in the session monitoring module 901 include:

[0121] The response time of the manual customer service exceeds a preset threshold or the user's level is a preset level.

[0122] Optionally, the intelligent conversation device 900 further includes an intention recognition module, and the intention recognition module is used to:

[0123] Perform semantic recognition on the historical conversation information between the user and the manual customer service to obtain semantic recognition results;

[0124] Based on the semantic recognition result and a preset scenario event library, the user's conversation intention is determined.

[0125] Optionally, the intention recognition module is further used to:

[0126] Determining event data corresponding to the semantic recognition result from the preset scene event library;

[0127] The semantic recognition results and event data are input into a pre-trained intent recognition model to obtain the user's conversation intent.

[0128] Optionally, the intelligent conversation device 900 further includes a model training module, and the model training module is used to train an intent recognition model:

[0129] Obtain historical event data and historical text data sent by users;

[0130] determining an event weight based on the historical event data;

[0131] Based on the event weights, historical event data and historical text data, an intent recognition model is trained.

[0132] Optionally, the intermediate response module 902 is used to:

[0133] Determining a conversation intervention scenario based on the user's conversation intention and a pre-set matching rule;

[0134] The conversation intervention logic is determined according to the scenario parameters and conditional rules corresponding to the intervention scenario.

[0135] Optionally, the intelligent conversation device 900 further includes a tag determination module, and the tag determination module is used to:

[0136] Make sure the session has the add flag.

[0137] It should be noted that the specific implementation content of the intelligent conversation device of the present invention has been described in detail in the above-mentioned intelligent conversation method, so the repeated content will not be described here.

[0138] According to a third aspect of an embodiment of the present invention, there is provided an electronic device, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method provided by the first aspect of the embodiment of the present invention.

[0139] According to a fourth aspect of an embodiment of the present invention, a computer-readable medium is provided, on which a computer program is stored, and when the program is executed by a processor, the method provided by the first aspect of the embodiment of the present invention is implemented.

[0140] Figure 8 An exemplary system architecture 1000 is shown to which an intelligent conversation method or an intelligent conversation device according to an embodiment of the present invention can be applied.

[0141] like Figure 8 As shown, the system architecture 1000 may include terminal devices 1001, 1002, 1003, a network 1004, and a server 1005. The network 1004 is used to provide a medium for communication links between the terminal devices 1001, 1002, 1003 and the server 1005. The network 1004 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0142] Users can use terminal devices 1001, 1002, 1003 to interact with server 1005 through network 1004 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 1001, 1002, 1003, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0143] The terminal devices 1001 , 1002 , and 1003 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers, etc.

[0144] The server 1005 may be a server that provides various services, such as a backend management server (only an example) that provides support for shopping websites browsed by users using the terminal devices 1001, 1002, and 1003. The backend management server may analyze and process the received data such as the data analysis request, and feed back the processing result (response result - only an example) to the terminal device.

[0145] It should be noted that the intelligent conversation method provided in the embodiment of the present invention is generally executed by the server 1005, and accordingly, the intelligent conversation device is generally set in the server 1005. The intelligent conversation method provided in the embodiment of the present invention can also be executed by the terminal devices 1001, 1002, and 1003, and accordingly, the intelligent conversation device can be set in the terminal devices 1001, 1002, and 1003.

[0146] It should be understood that Figure 8 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0147] Reference below Fig. 9 , which shows a schematic diagram of the structure of a computer system 1100 of a terminal device suitable for implementing an embodiment of the present invention. Fig. 9 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0148] like Fig. 9 As shown, the computer system 1100 includes a central processing unit (CPU) 1101, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1102 or a program loaded from a storage portion 1108 into a random access memory (RAM) 1103. In the RAM 1103, various programs and data required for the operation of the system 1100 are also stored. The CPU 1101, the ROM 1102, and the RAM 1103 are connected to each other via a bus 1104. An input / output (I / O) interface 1105 is also connected to the bus 1104.

[0149] The following components are connected to the I / O interface 1105: an input section 1106 including a keyboard, a mouse, etc.; an output section 1107 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1108 including a hard disk, etc.; and a communication section 1109 including a network interface card such as a LAN card, a modem, etc. The communication section 1109 performs communication processing via a network such as the Internet. A drive 1110 is also connected to the I / O interface 1105 as needed. A removable medium 1111, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1110 as needed, so that a computer program read therefrom is installed into the storage section 1108 as needed.

[0150] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 1109, and / or installed from the removable medium 1111. When the computer program is executed by the central processing unit (CPU) 1101, the above-mentioned functions defined in the system of the present invention are executed.

[0151] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0152] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0153] The modules involved in the embodiments of the present invention may be implemented by software or hardware. The modules described may also be set in a processor, for example, it may be described as: a processor includes a result response module, an intermediate response module and a session monitoring module, wherein the names of these modules do not constitute a limitation on the modules themselves in some cases, for example, the session monitoring module may also be described as "a module for monitoring the conversation between a user and a manual customer service".

[0154] As another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiment; or it may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by a device, the device implements the following method: monitor the conversation between the user and the manual customer service, and when the conversation meets the preset trigger condition, determine the conversation intervention scenario and conversation intervention logic according to the user's conversation intention; generate an intermediate response according to the conversation intervention scenario and conversation intervention logic; the intermediate response is displayed on the conversation interface; and generate a response result in response to the user's operation triggered by the intermediate response.

[0155] According to the technical solution of the embodiment of the present invention, the intelligent conversation method of the embodiment of the present invention monitors the conversation between the user and the manual customer service, and when the conversation meets the preset trigger condition, determines the conversation intervention scenario and the conversation intervention logic according to the user's conversation intention; generates an intermediate response according to the conversation intervention scenario and the conversation intervention logic; the intermediate response is displayed on the conversation interface; and generates a response result in response to the user's operation triggered by the intermediate response; thereby, the manual customer service and the machine customer service can be integrated, and the response can be automatically made according to the determined conversation intervention scenario and the conversation intervention logic, thereby reducing the conversation cost and improving the conversation efficiency.

[0156] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0157] It should be noted that the acquisition, storage and application of user personal information involved in the technical solution of the present disclosure are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

Claims

1. An intelligent conversation method, characterized in that: include: Monitor the conversation between the user and the manual customer service, and when the conversation meets the preset trigger condition, determine the conversation intervention scenario and conversation intervention logic according to the user's conversation intention; generating an intermediate response according to the conversation intervention scenario and the conversation intervention logic; The intermediate response is displayed on the conversation interface; In response to an operation triggered by the user based on the intermediate response, a response result is generated.

2. The method according to claim 1, characterized in that The preset trigger conditions include: The response time of the manual customer service exceeds a preset threshold or the user's level is a preset level.

3. The method according to claim 1, characterized in that Before determining the conversation intervention scenario and conversation intervention logic according to the conversation intention of the user, the method includes: Perform semantic recognition on the historical conversation information between the user and the manual customer service to obtain semantic recognition results; Based on the semantic recognition result and a preset scenario event library, the user's conversation intention is determined.

4. The method according to claim 3, characterized in that Determining the user's session intention based on the semantic recognition result and a preset scenario event library includes: Determining event data corresponding to the semantic recognition result from the preset scene event library; The semantic recognition results and event data are input into a pre-trained intent recognition model to obtain the user's conversation intent.

5. The method according to claim 4, characterized in that The pre-trained intent recognition model is obtained by the following steps: Obtain historical event data and historical text data sent by users; determining an event weight based on the historical event data; Based on the event weights, historical event data and historical text data, an intent recognition model is trained.

6. The method according to claim 1, characterized in that Determining a conversation intervention scenario and conversation intervention logic according to the conversation intention of the user includes: Determining a conversation intervention scenario based on the user's conversation intention and a pre-set matching rule; The conversation intervention logic is determined according to the scenario parameters and conditional rules corresponding to the intervention scenario.

7. The method according to claim 1, characterized in that Before monitoring the conversation between the user and the manual customer service, including: Make sure the session has the add flag.

8. An intelligent conversation device, characterized in that: include: A conversation monitoring module is used to monitor the conversation between the user and the manual customer service, and when the conversation meets the preset trigger condition, determine the conversation intervention scenario and conversation intervention logic according to the user's conversation intention; An intermediate response module, used for generating an intermediate response according to the conversation intervention scenario and the conversation intervention logic; The intermediate response is displayed on the conversation interface; The result response module is used to generate a response result in response to the user's operation triggered based on the intermediate response.

9. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 7.

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

Images