Response method and apparatus, electronic device, and storage medium

Abstract

Embodiments of the present specification provide a responding method and device, electronic equipment and a storage medium, wherein the responding method comprises: obtaining a responding access request of a target user; the responding access request comprises a first request identifier and a second request identifier; the first request identifier comprises at least one of a conversation identifier and a target user identifier of the target user; determining risk judgment data corresponding to the responding access request according to at least one of the conversation identifier and the target user identifier; determining a risk user type of the target user according to the risk judgment data; placing the second request identifier of the responding access request in a target queuing queue corresponding to the risk user type; each second request identifier included in the target queuing queue is arranged in sequence, and the arrangement order of the second request identifier represents the expected response order of the to-be-processed access request corresponding to the second request identifier. In this way, the satisfaction of the user in the process of waiting for accessing the artificial agent is improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a response method, apparatus, electronic device and storage medium. Background Technology

[0002] With the development of electronic technology, the application of artificial intelligence is becoming increasingly widespread. Using AI technology to provide users with consultation, complaint handling, and other interactive services can significantly save manpower and efficiently resolve user issues. However, AI-powered smart agents cannot completely replace human agents. When smart agents cannot solve a user's problem, the user often prefers to switch from a smart agent to a human agent.

[0003] If multiple users need to be transferred to a live agent, but the number and capacity of live agents are limited, the agent service system needs to queue users and assign them to agents in order. However, different users have different levels of urgency in their need for live agent services, meaning some users may be more tolerant of long queue times, while others may be more likely to become dissatisfied due to excessively long wait times. If the agent service system queues users solely based on the time they submit their requests to connect to a live agent, it may result in excessively long wait times for some users with urgent needs, reducing user satisfaction with connecting to a live agent. Summary of the Invention

[0004] This application provides a response method, apparatus, electronic device, and storage medium to improve user satisfaction while waiting to connect to a live agent.

[0005] In a first aspect, embodiments of this application provide a response method, including:

[0006] Obtain the target user's response access request; the response access request carries a first request identifier and a second request identifier; the first request identifier includes at least one of a dialogue identifier and the target user's target user identifier; the second request identifier is used to uniquely identify the response access request;

[0007] Based on at least one of the dialogue identifier and the target user identifier, determine the risk assessment data corresponding to the response access request;

[0008] Based on the risk assessment data, the risk user type of the target user is determined;

[0009] The second request identifier of the response access request is placed in the target queuing queue corresponding to the risk user type; each of the multiple second request identifiers included in the target queuing queue is arranged sequentially, and the arrangement order of the second request identifiers is used to characterize the expected response order of the access request to be processed corresponding to the second request identifier.

[0010] Secondly, embodiments of this application provide a response device, including:

[0011] An acquisition unit is configured to acquire a response access request from a target user; the response access request carries a first request identifier and a second request identifier; the first request identifier includes at least one of a dialogue identifier and a target user identifier of the target user; the second request identifier is used to uniquely identify the response access request.

[0012] The first determining unit is configured to determine the risk assessment data corresponding to the response access request based on at least one of the dialogue identifier and the target user identifier.

[0013] The second determining unit is used to determine the risk user type of the target user based on the risk assessment data;

[0014] The placement unit is used to place the second request identifier of the response access request into the target queuing queue corresponding to the risk user type; each second request identifier in the target queuing queue is arranged sequentially, and the arrangement order of the second request identifiers is used to characterize the expected response order of the access requests to be processed corresponding to the second request identifier.

[0015] Thirdly, embodiments of this application provide an electronic device, including: a processor; and a memory configured to store computer-executable instructions, which, when executed, cause the processor to perform the response method as described in the first aspect.

[0016] Fourthly, embodiments of this application provide a computer-readable storage medium for storing computer-executable instructions, which, when executed by a processor, implement the response method as described in the first aspect.

[0017] As can be seen, in this embodiment, firstly, the response access request of the target user is obtained; the response access request carries a first request identifier and a second request identifier; the first request identifier includes at least one of a dialogue identifier and a target user identifier of the target user; the second request identifier is used to uniquely identify the response access request; secondly, risk assessment data corresponding to the response access request is determined based on at least one of the dialogue identifier and the target user identifier; then, the risk user type of the target user is determined based on the risk assessment data; finally, the second request identifier of the response access request is placed in the target queuing queue corresponding to the risk user type; each of the multiple second request identifiers included in the target queuing queue is arranged sequentially, and the arrangement order of the second request identifiers is used to characterize the expected response order of the access request to be processed corresponding to the second request identifier. Therefore, by using at least one of the dialogue identifier and the target user identifier carried in the target user's response access request, the risk assessment data corresponding to the response access request can be determined. Furthermore, based on the risk assessment data, the risk user type of the target user can be determined. Each risk user type corresponds to a different queuing queue. By placing the second request identifier carried in the target user's response access request into the target queuing queue corresponding to the risk user type, target users of different risk user types can be pre-classified, indirectly reducing the number of pending access requests corresponding to each queuing queue, shortening the waiting time for response access requests, and improving user satisfaction while waiting to access a human agent. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating a response method provided in an embodiment of this application;

[0020] Figure 2 A framework diagram of a seat service system provided in this application embodiment;

[0021] Figure 3 A block diagram illustrating a response method provided in an embodiment of this application;

[0022] Figure 4 A flowchart illustrating another response method provided in this application embodiment:

[0023] Figure 5 A flowchart illustrating another response method provided in an embodiment of this application;

[0024] Figure 6 A schematic diagram of a response device provided in an embodiment of this application;

[0025] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0026] To enable those skilled in the art to better understand the technical solutions in the embodiments of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments in this specification, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this application.

[0027] Agent seats can be divided into human agents and intelligent agents. Human agents include, but are not limited to, text-based agents, video agents, and voice agents. Text-based agents primarily provide services to users through text chat; video agents primarily provide services through voice and video; and voice agents primarily provide services through mobile phones. To reduce manpower and material resources, the agent service system typically prioritizes providing intelligent agents, which use AI-powered robots to meet the needs of most customers. However, in financial or other business scenarios, there are special needs that cannot be met by intelligent agents alone. When intelligent agents cannot resolve a user's problem, the user needs to be transferred to a human agent.

[0028] When a large number of users require connection to a live agent, the limited number and capacity of live agents necessitate queuing of these requests. Typically, the system can queue these requests based on the order in which they are sent. However, different users have varying degrees of urgency. Some users with urgent needs are highly sensitive to queuing time, as prolonged waits can significantly reduce their satisfaction; others, more tolerant of waiting times, may be more problematic. Queuing these users together could lead to excessively long wait times for the latter, causing significant dissatisfaction.

[0029] To address the aforementioned issues, this application provides a response method.

[0030] The response method proposed in this application can be executed by an electronic device, specifically by a processor within the electronic device. The electronic device mentioned herein can be a terminal device, such as a smartphone, tablet computer, desktop computer, intelligent voice interaction device, wearable device, robot, or vehicle terminal, etc.; alternatively, the electronic device can also be a server, such as a standalone physical server, a server cluster consisting of multiple servers, or a cloud server capable of cloud computing.

[0031] Figure 1 This is a flowchart illustrating a response method provided in an embodiment of this application. Figure 2 This is a framework diagram of a seat service system provided in an embodiment of this application.

[0032] like Figure 1 The response method shown can be applied to an agent service system. The structure of the agent service system can be referenced. Figure 2 .

[0033] A call center service system can be a system that provides services to users through at least one of intelligent or human agents, such as a customer service system. In this specification, "agent" can refer to a service personnel in a TSR (Telephone Service Representative). A TSR typically consists of an agent computer, agent software, agent headset, and service personnel. The TSR uses agent software and hardware to implement relevant control functions to achieve customer service objectives and falls under the scope of customer service.

[0034] For example, a smart agent can be a robot that can provide services to users, while a human agent can be a customer service representative, a volunteer, or other types of service personnel.

[0035] like Figure 2 As shown, the agent service system includes a presentation layer 201, a service layer 202, a middleware layer 203, and a storage layer 204. The presentation layer is the user interface displayed by the agent service system to the user, and it includes an intelligent agent terminal 2011 and a human agent terminal 2012. The intelligent agent terminal 2011 refers to the first user interface providing intelligent agent services to the user, and the human agent terminal 2012 refers to the second user interface providing human agent services to the user.

[0036] The serial numbers “first,” “second,” “third,” etc., appearing in the embodiments of this application are merely for the purpose of distinguishing multiple features that are different in application but similar in content, and have no actual meaning. They will not be elaborated on further below.

[0037] Service layer 202 may include various system components for performing response methods, including but not limited to: response system DM (Dialogue Manager) 2021, NLP (Natural Language Processing) intent recognition 2022, robot knowledge base 2023, multi-turn dialogue engine 2025, human agent interface 2026, and session management service 2027.

[0038] Session Management Service 2027 can be used to manage conversations between smart agents / human agents and users.

[0039] The middleware layer 203 includes the framework components and connection components of the agent service system, which can be used for information transmission within the agent service system. The middleware layer 203 may include: a Kafka component 2031, a Redis component 2032, and a message bus 2033.

[0040] The storage layer can be used to store the data required by the agent service system. The storage layer can include: MySQL database, HBase database, and Elasticsearch database.

[0041] It is important to note that Figure 2 The structure of the agent service system shown is merely exemplary, and agent service systems may vary considerably due to differences in configuration or performance.

[0042] In practical applications of the response method, the agent service system can also be replaced by other systems that can provide agent services. This specification mainly uses the application of the response method to the agent service system as an example for explanation. When the response method is applied to other systems that can provide agent services, please refer to the corresponding description of the embodiment of the response method applied to the agent service system.

[0043] Next, a detailed introduction Figure 1 The response method shown.

[0044] Step S102: Obtain the target user's response access request; the response access request carries a first request identifier and a second request identifier; the first request identifier includes at least one of a dialogue identifier and a target user identifier of the target user; the second request identifier is used to uniquely identify the response access request.

[0045] The target users can be users of a customer service system who require access to live agent seats. Target users can be registered users of the customer service system or temporary users. Temporary users are those who have not yet registered their personal information with the customer service system and have been assigned a temporary user ID by the system.

[0046] In step S102, the number of target users can be one or more. It's important to note that having multiple target users means obtaining the response access request from each of the multiple target users simultaneously. If the multiple response access requests are obtained at different times, the process can be performed separately for each target user. Figure 1 The response method shown in the embodiment.

[0047] The response access request can be a response access request that is transferred from the smart agent to the human agent. If the human agent responds to the response access request, the human agent who responded to the response access request will replace the smart agent to provide services to the target user.

[0048] The response access request can also be a response access request sent directly to a human agent. If the human agent responds to the response access request, the human agent who responded to the response access request will provide services to the target user.

[0049] Services that intelligent or human agents can provide to target users include, but are not limited to: answering user inquiries, providing shopping links, providing information on promotional activities, responding to points inquiry results, recommending services suitable for user needs, providing interfaces for participating in activities, and reassuring users who intend to complain, etc. This manual does not impose any special limitations on the services provided by intelligent or human agents.

[0050] The response to the access request may include a first request identifier and a second request identifier.

[0051] The first request identifier may include only the dialogue identifier, only the target user identifier of the target user, or both the dialogue identifier and the target user identifier.

[0052] The dialogue identifier can be a string or other preset identifier. The dialogue identifier can be used to uniquely identify a dialogue between the target user and the smart agent. This dialogue can occur before step S102 is executed, that is, before obtaining the target user's response to the access request.

[0053] The target user identifier can be a user identifier used to uniquely identify the target user. The target user identifier can be an official user identifier assigned to a registered user by the agent service system. The target user identifier can be used to query the target user's historical service data within the agent service system. Historical service data includes, but is not limited to: historical dialogue data saved after the target user has interacted with at least one of the intelligent agent and the human agent, as well as historical response records of access requests and responses to those requests, etc.

[0054] The second request identifier can be used to uniquely identify the response access request. The second request identifier can be in string form or other preset form. It should be noted that for each target user, if the target user sends the response access request at different times, then the response access request sent at each time point corresponds to a second request identifier.

[0055] The target user can either send a response access request to be transferred to a human agent after having a conversation with the intelligent agent, or send a response access request to a human agent without having a conversation with the intelligent agent.

[0056] If a target user first engages in a conversation with the smart agent and then sends a request to be transferred to a human agent, the first request identifier may include the conversation identifier of the conversation with the smart agent.

[0057] When a target user sends a response access request directly to a human agent without first engaging in dialogue with the intelligent agent, the first request identifier may not include the dialogue identifier.

[0058] If the target user is a registered user of the agent service system, the first request identifier may include the target user identifier of the target user.

[0059] If the target user is a temporary user of the agent service system, the user identifier of the temporary user is a temporary user identifier and cannot be used to query historical service data in the agent service system. Therefore, the first request identifier of the human agent may not include the target user identifier of the target user.

[0060] Obtaining the target user's response access request can be done by obtaining the specified text entered by the target user in the text input box, such as "human agent", or by obtaining the target user's click operation on a pre-configured response access request control. It can also be done by pre-configuring other ways to trigger the response access request, receiving user input corresponding to the other way to trigger the response access request, and triggering the response access request based on the user input, etc.

[0061] Optionally, before obtaining the target user's response access request, the method further includes: obtaining dialogue data between the target user and the intelligent agent; determining whether the dialogue data includes at least one target word that matches a sensitive word based on a pre-configured set of sensitive words; if the dialogue data includes at least one target word, then performing data embedding processing based on the target word to obtain embedding data.

[0062] In practical implementation, a set of sensitive words can be pre-configured in the agent service system. This set can include multiple sensitive terms, such as "complaint." Furthermore, in different business scenarios, corresponding sets of sensitive words can be configured according to the actual business requirements.

[0063] When a target user engages in conversation with the intelligent agent, the conversation data between the target user and the intelligent agent can be obtained. The conversation data includes every sentence of chat content between the target user and the intelligent agent.

[0064] Based on the sensitive word set, the system can perform sensitive word matching on each chat message between the target user and the bot to determine whether each chat message contains at least one target word that matches the sensitive word.

[0065] The detection of sensitive words can include the following steps:

[0066] (1) Each chat message can be segmented into words to obtain the segmentation results, which include multiple words corresponding to the chat message. For example, a chat message is: I want to complain about XXX because his service attitude is extremely poor. After segmenting the chat message, the segmentation results include the following words: "I", "want", "complain", "XXX", "he", "of", "service", "attitude", "extremely", and "poor".

[0067] (2.1) In one embodiment, for each of the multiple words, the similarity between each sensitive word in the sensitive word set and the word can be calculated. If the similarity is greater than or equal to a preset threshold, then the word is determined to match the sensitive word.

[0068] For example, the sensitive word set includes: A, B, etc. The word segmentation result of a chat message includes multiple words: x1, x2, x3, x4, x5, x6. The similarity between the sensitive word "A" in the sensitive word set and each of the aforementioned word segmentation results (x1, x2, x3, x4, x5, x6) is calculated. If the similarity is greater than a preset threshold, the word is determined to match the sensitive word "A"; if the similarity is less than or equal to the preset threshold, the word is determined not to match the sensitive word "A". Similarly, the similarity between the sensitive word "B" in the sensitive word set and each of the aforementioned word segmentation results (x1, x2, x3, x4, x5, x6) is calculated. If the similarity is greater than a preset threshold, the word is determined to match the sensitive word "B"; if the similarity is less than or equal to the preset threshold, the word is determined not to match the sensitive word "B", and so on.

[0069] (2.2) In another implementation, a set of synonyms for each sensitive word can be pre-configured. For each of the multiple words, it can be queried whether the word belongs to the set of synonyms of each sensitive word. If so, it is determined that the word matches the sensitive word.

[0070] For example, the sensitive word set includes "A", "B", etc. The set of synonyms for the sensitive word "A", set 1, includes: A1, A2, A3, A4. The set of synonyms for the sensitive word "B", set 2, includes: B1, B2, B3, B4, B5. The word segmentation result of a chat message includes multiple words: x1, x2, x3, x4, x5, x6. For x1, query whether the word belongs to set 1. If yes, x1 matches the sensitive word "A"; otherwise, x1 does not match the sensitive word "A". Query whether the word belongs to set 2. If yes, x1 matches the sensitive word "B"; otherwise, x1 does not match the sensitive word "B". x2, x3, x4, x5, and x6 are similar to x1; refer to the explanation section for x1.

[0071] Data tracking is performed based on target words to obtain tracking data. This can be determined by the number of target words or by the sentence containing the target words. Tracking data can be stored in a database or a Redis component.

[0072] Step S104: Determine the risk assessment data corresponding to the response access request based on at least one of the dialogue identifier and the target user identifier.

[0073] Each dialogue identifier corresponds to one dialogue between a user and the smart agent. Each user can have multiple dialogues with the smart agent. The target user's target user identifier can correspond to one or more dialogue identifiers.

[0074] The dialogue identifier included in the response access request can be the dialogue identifier of the most recent dialogue between the target user and the smart agent before the target user sends the response access request to be transferred from the smart agent to the human agent.

[0075] Based on this dialogue identifier, the content of the most recent dialogue between the target user and the smart agent can be obtained. Then, risk assessment data for the access request can be determined based on the dialogue content.

[0076] Based on the target user identifier, the target user's historical dialogue data, historical event tracking data, and other types of historical data can be queried, thereby determining the risk assessment data corresponding to the access request based on each type of historical data.

[0077] Risk assessment data can be parameters used to determine whether a target user is a risky user.

[0078] Step S106: Determine the risk user type of the target user based on the risk assessment data.

[0079] Risk user types can include risk users and non-risk users, multiple risk levels, or other pre-configured risk user types.

[0080] Optionally, based on the dialogue identifier, the risk assessment data corresponding to the response access request is determined, including: based on the dialogue identifier, querying the tracking data corresponding to the dialogue identifier in historical tracking data to obtain the first tracking data; reading the number of target words from the first tracking data to obtain the first quantity, and determining the first quantity as the risk assessment data; and based on the risk assessment data, determining the risk user type of the target user, including: if the first quantity is greater than a first preset quantity threshold, then determining the risk user type of the target user as the first risk type.

[0081] When a target user switches from a smart agent to a human agent, the risk assessment data corresponding to the response access request can be determined based on the dialogue identifier.

[0082] Historical event tracking data can include the correspondence between pre-generated dialogue identifiers and event tracking data. The generation method of event tracking data corresponding to each dialogue identifier in historical event tracking data can refer to the previous explanation section: "Acquire dialogue data between the target user and the intelligent agent; based on a pre-configured sensitive word set including multiple sensitive words, determine whether the dialogue data includes at least one target word that matches the sensitive word; if the dialogue data includes at least one target word, then perform data tracking processing based on the target word to obtain the event tracking data."

[0083] In the multi-turn dialogue process of the agent service system, the agent service system can determine whether the dialogue corresponding to the dialogue identifier carried by the target user's response access request hits sensitive words. Specifically, it queries the cumulative number of hits A of the dialogue based on the dialogue identifier sessionid from the Redis component or the db database, that is, it queries the tracking data corresponding to the dialogue identifier to obtain the first tracking data; it reads the number of target words from the first tracking data to obtain the first quantity.

[0084] The first preset threshold can be the sensitive word hit threshold X pre-configured by the agent service system. If the cumulative number of hits A is greater than X, it means that the conversation has triggered sensitive words, and the risk user type of the target user can be determined as the first risk type.

[0085] Optionally, based on the target user identifier, the risk assessment data corresponding to the response access request is determined, including: querying the tracking data corresponding to the target user identifier in historical tracking data based on a preset time length and the target user identifier to obtain second tracking data; reading the number of target words from the second tracking data to obtain a second quantity, and determining the second quantity as risk assessment data; determining the risk user type of the target user based on the risk assessment data, including: if the second quantity is greater than a second preset quantity threshold, then determining the risk user type of the target user as a second risk type.

[0086] First, query the sensitive word hit data table to count the number of times (A) sensitive words were hit in all conversations of the target user within X days, where X can represent the preset time length. The second preset threshold can be the historical conversation sensitive word configuration threshold (N). Obtain the historical conversation sensitive word configuration threshold (N) from the agent service system configuration table. If A > N, it indicates that sensitive words were triggered in a historical conversation, and the target user's risk user type can be determined as the second risk type.

[0087] The second preset quantity threshold can be a value greater than or equal to the first preset quantity threshold.

[0088] Specifically, the target time range can be determined based on the preset time length and the current time point; based on the target time range and the target user identifier, the second tracking data can be obtained by querying the tracking data corresponding to the target user identifier from the historical tracking data.

[0089] Within the target time frame, the number of conversations corresponding to the target user identifier may be one, non-existent, or multiple. The conversation identifiers corresponding to the target user identifier can be retrieved from historical service data. Furthermore, if the number of conversation identifiers is one or more, the tracking data corresponding to each conversation identifier can be retrieved from historical tracking data. If the number of conversation identifiers is zero, there is no tracking data corresponding to the target user identifier in the historical tracking data; in this case, the second tracking data can be represented by a preset null value, such as "0".

[0090] In another implementation, based on a preset time range and target user identifier, the tracking data corresponding to the target user identifier can be queried in the historical tracking data to obtain the second tracking data; the number of target words can be read from the second tracking data to obtain the second quantity, and the second quantity can be determined as the risk judgment data; based on the risk judgment data, the risk user type of the target user can be determined, including: if the second quantity is greater than the second preset quantity threshold, the risk user type of the target user can be determined as the second risk type.

[0091] Optionally, based on the target user identifier, the risk assessment data corresponding to the response access request is determined, including: based on the target user identifier, querying in a preset storage space whether there is at least one target complaint work order, obtaining the work order query result, and determining the work order query result as risk assessment data; the target complaint work order meets the preset work order screening conditions; the preset storage space stores the correspondence between user identifiers and complaint work orders; based on the risk assessment data, the risk user type of the target user is determined, including: if the work order query result indicates that there is at least one target complaint work order, then the risk user type of the target user is determined as the third risk type.

[0092] Preset work order filtering conditions include, but are not limited to, one or more of the following: customer source filtering conditions, status filtering conditions, time filtering conditions, and work order tag filtering conditions.

[0093] The default storage space can be a CRM (Customer Relationship Management) system.

[0094] The third risk type can include multiple risk subtypes, which can be determined based on preset work order screening criteria. The following examples illustrate several risk subtypes of the third risk type.

[0095] With preset work order filtering conditions including customer source filtering conditions and status filtering conditions, the work order data of the day in the CRM system can be queried by target user identifier. If the status of the work order data is one of "processing", "out of contact", "suspended", "pending assignment" or "completed and pending follow-up", and the customer source is one of "online platform" or "media", it means that the target user has unprocessed complaint work orders on that day. The risk user type of the target user can be identified as the first subtype of the third risk type.

[0096] Given that the preset work order filtering conditions include time filtering, customer source filtering, and status filtering, the CRM system retrieves work order data for the most recent X days (configured by business personnel in the system business configuration table) based on the target user identifier. If there is work order data with the customer source being either "online platform" or "media" and the status being "closed", it indicates that the target user has historical complaint work orders that have been processed. The risk user type of the target user can be determined as the second subtype in the third risk type.

[0097] Given that the preset work order filtering conditions include work order tag filtering conditions and status filtering conditions, the CRM system retrieves the work order data for the day based on the target user identifier. If there are work order data with a status of "processing", "out of contact", "suspended", "pending assignment" and "completed and pending follow-up", and the work order tag is one of "consultation", "level 1 complaint" and "level 2 complaint", it means that the target user has unprocessed complaint work orders for the day. The risk user type of the target user can be determined as the third subtype of the third risk type.

[0098] Given preset work order filtering conditions including time filtering, work order tag filtering, and status filtering, work order data from the CRM system for the most recent X days is retrieved based on the target user identifier. If there are work order data with a status of "Processing," "Out of Contact," "Suspended," "Pending Assignment," or "Completed but Pending Follow-up," and a work order tag of "Inquiry," "Level 1 Complaint," or "Level 2 Complaint," it indicates that the target user has historical complaint work orders that have not yet been processed. Therefore, the target user's risk user type can be identified as the fourth subtype within the third risk type. The aforementioned first, second, third, and fourth subtypes can be four different types of risk subtypes belonging to the third risk type.

[0099] Optionally, based on a pre-configured set of sensitive words including multiple sensitive words, after determining whether the dialogue data includes at least one target word matching a sensitive word, the process further includes: if the number of target words included in the dialogue data is greater than a third preset threshold, a request for human agent access is triggered; if a human agent access request is successfully made, a target behavior tracking record is generated; based on the target user identifier, risk assessment data corresponding to the access request is determined, including: based on the target user identifier, the number of target behavior tracking records corresponding to the target user identifier is queried from historical tracking data to obtain a third number, and the third number is determined as risk assessment data; based on the risk assessment data, the risk user type of the target user is determined, including: if the third number is greater than a fourth threshold, the risk user type of the target user is determined as a fourth risk type.

[0100] After a target user has a conversation with the intelligent agent, the conversation may be automatically transferred to a human agent if the conversation contains multiple sensitive words. Each time the user successfully triggers a human agent transfer, the data of the transfer will be embedded in the Redis component and the database.

[0101] When determining whether to transfer a user to a human agent during a multi-round dialogue process, the number of times the target user has been transferred to a human agent (A) can be obtained from Redis or the database based on the target user identifier. The fourth preset threshold can be the number of times the user has been transferred to a human agent (X) configured in the agent service system configuration table. If A > X, it means that the target user has previously transferred to a human agent multiple times, and the target user's risk user type is determined as the fourth risk type.

[0102] Optionally, based on the target user identifier, the risk assessment data corresponding to the response access request is determined, including: querying the number of dialogue data corresponding to the target user identifier from historical dialogue data based on the target user identifier and a preset time range to obtain a fourth quantity; and querying the number of interactive voice response (IVR) records corresponding to the target user identifier from historical call records based on the target user identifier and a preset time range to obtain a fifth quantity; the fourth quantity and the fifth quantity are determined as risk assessment data; based on the risk assessment data, the risk user type of the target user is determined, including: if the fourth quantity is greater than the fifth quantity threshold and the fifth quantity is greater than the sixth quantity threshold, then the risk user type of the target user is determined as the fifth risk type.

[0103] The risk user type for the target user is the fifth risk type, which can be used to indicate that the target user has made repeated calls. Repeated calls can be divided into two situations: 1. Repeated calls from smart agents; 2. Repeated calls from IVR voice systems.

[0104] The definition of repeated calls to the smart agent is as follows: When a target user enters a multi-round conversation at time A, the number of times the target user accesses the smart agent within the time range before time A is counted (C1), and the number of calls configured by the business personnel (C2) is obtained from the business configuration table. If C1 > C2, it indicates that the smart agent has made repeated calls.

[0105] The definition of repeated IVR voice calls is as follows: When a target user enters a multi-round conversation at time A, the number of times the target user actively dials the agent's phone number A1 within the same day before time A is counted. At the same time, the threshold for dialing the agent's phone number B1 is obtained from the system's service configuration table. If A1 > B1, it indicates that the IVR voice call is repeated.

[0106] If both the intelligent agent's repeated call and the IVR voice repeated call are true simultaneously, it means that the target user meets the pre-configured repeated call conditions on that day, and thus the target user's risk user type can be determined as the fifth risk type.

[0107] In another implementation, the system can also query whether the target user is a member of the call center service system based on the target user identifier. If so, the membership level can be further queried. If the target user is a member, the risk user type of the target user can be determined as the sixth risk type. Furthermore, the subtype corresponding to the target user in the sixth risk type can be determined based on the membership level, so as to provide different levels of queuing services for members of different levels in the future.

[0108] It should be noted that the above-mentioned implementation methods for determining the risk user type of the target user based on risk assessment data can be executed simultaneously, or in a preset order. They can be executed only one or more of the implementation methods, or all of the implementation methods can be executed.

[0109] When multiple implementation methods for determining the risk user type of the target user are executed simultaneously, if the number of risk user types of the target user is greater than one, the risk user type with the highest priority can be determined as the target risk user type of the target user according to the pre-configured priority order, and then the target queue corresponding to the target risk user type can be determined in the subsequent step S108.

[0110] For example, at time point T1, multiple implementation methods for determining the risk user type of the target user can be executed simultaneously, and the processing flow and results of each implementation method are as follows:

[0111] (a1) Based on the dialogue identifier, query the corresponding tracking data in the historical tracking data to obtain the first tracking data; read the number of target words from the first tracking data to obtain the first quantity. If the first quantity is less than the first preset quantity threshold, it can be determined that the risk user type of the target user is not the first risk type.

[0112] (a2) Based on the preset time length and target user identifier, query the tracking data corresponding to the target user identifier in the historical tracking data to obtain the second tracking data; read the number of target words from the second tracking data to obtain the second quantity; if the second quantity is greater than the second preset quantity threshold, then the risk user type of the target user is determined as the second risk type.

[0113] (a3) Based on the target user identifier and the preset time range, query the number of dialogue data corresponding to the target user identifier from the historical dialogue data to obtain the fourth quantity; and based on the target user identifier and the preset time range, query the number of interactive voice response (IVR) records corresponding to the target user identifier from the historical call records to obtain the fifth quantity; if the fourth quantity is greater than the fifth quantity threshold and the fifth quantity is greater than the sixth quantity threshold, then the risk user type of the target user is determined as the fifth risk type.

[0114] (a4) Based on the target user identifier, query whether the target user is a member of the call center service system and obtain the query results. If the query results determine that the target user is a member, then the target user's risk user type can be determined as the sixth risk type.

[0115] In summary, the risk user types of the target user can be identified as including the second risk type, the fifth risk type, and the sixth risk type. Furthermore, the priority order of the first risk user type r1, the second risk user type r2, the fifth risk user type r5, and the sixth risk user type r6, from highest to lowest, is r1, r2, r5, r6. Therefore, based on this priority order, the final risk user type of the target user can be determined to be the second risk type. Consequently, the target queuing queue corresponding to the second risk type can be determined in subsequent step S108.

[0116] When multiple implementation methods for determining the risk user type of a target user are executed sequentially according to a preset order, after obtaining a risk user type, the implementation method for determining the risk user type of a target user based on risk judgment data that has not yet been executed can be skipped, and the currently determined risk user type can be determined as the target risk user type of the target user. Then, the target queue corresponding to the target risk user type can be determined in the subsequent step S108.

[0117] For example, the preset order can be used to indicate sequentially determining whether the risk user type of the target user is a first risk type, a second risk type, a fifth risk type, and a sixth risk type. Multiple implementation methods for determining the risk user type of the target user are executed sequentially according to the preset order, and the processing flow and results of each implementation method are as follows:

[0118] (b1) First, based on the dialogue identifier, query the corresponding tracking data in the historical tracking data to obtain the first tracking data; read the number of target words from the first tracking data to obtain the first quantity. If the first quantity is less than the first preset quantity threshold, it can be determined that the risk user type of the target user is not the first risk type.

[0119] (b2) Secondly, based on the preset time length and target user identifier, query the tracking data corresponding to the target user identifier in the historical tracking data to obtain the second tracking data; read the number of target words from the second tracking data to obtain the second quantity; if the second quantity is less than the second preset quantity threshold, it can be determined that the risk user type of the target user is not the second risk type.

[0120] (b3) Next, based on the target user identifier and the preset time range, the number of dialogue data corresponding to the target user identifier is queried from the historical dialogue data to obtain the fourth number; and based on the target user identifier and the preset time range, the number of interactive voice response (IVR) records corresponding to the target user identifier is queried from the historical call records to obtain the fifth number; if the fourth number is greater than the threshold of the fifth number and the fifth number is greater than the threshold of the sixth number, then the risk user type of the target user is determined as the fifth risk type.

[0121] After determining the target user's risk user type as the fifth risk type, since one risk user type has already been obtained, the previously unexecuted step of "querying whether the target user is a member of the agent service system based on the target user identifier and obtaining the query result; determining whether the target user is a member based on the query result; if so, determining the target user's risk user type as the sixth risk type" can be skipped. The currently determined risk user type is the fifth risk type. By determining the fifth risk type as the target user's target risk user type, the target queuing queue corresponding to the fifth risk type can be determined in the subsequent step S108.

[0122] Step S108: Place the second request identifier of the access request in the target queuing queue corresponding to the risk user type; each second request identifier in the target queuing queue is arranged in sequence, and the arrangement order of the second request identifiers is used to characterize the expected response order of the access request to be processed corresponding to the second request identifier.

[0123] The correspondence between risky user types and queuing queues can be one-to-one, one-to-many, many-to-one, or many-to-many. For example, each risky user type can correspond to one queuing queue or K queuing queues. Alternatively, each queuing queue can correspond to L risky user types.

[0124] When each risk user type corresponds to multiple queues, based on the risk user type of the target user, first query to obtain the corresponding multiple queues, then query the queue length of each queue, and determine the queue with the shortest queue length as the target queue.

[0125] In another implementation, when each risk user type corresponds to multiple queues, based on the risk user type of the target user, the corresponding multiple queues are first queried. Then, the queue with the highest queue priority can be determined as the target queue according to the preset queue priority order of the multiple queues.

[0126] Based on the pre-configured correspondence between risk user types and queuing queues, the target queuing queue can be obtained by querying the risk user type of the target user determined in step S106. In the case of simultaneously executing multiple implementation methods for determining the risk user type of the target user, the target queuing queue can also be obtained by querying the final risk user type of the target user determined in step S106.

[0127] When the risk user type includes both risky and non-risky users, the mapping between risk tags and the first queuing queue, and the mapping between risk-free tags and the second queuing queue, can be pre-configured. If, based on risk assessment data, the target user is determined to be a risky user, a risk tag can be added to that target user, so that the first queuing queue corresponding to the risk tag is identified as the target queuing queue corresponding to the risky user type during step S108. If, based on risk assessment data, the target user is determined to be a risk-free user, a risk-free tag can be added to that target user, so that the second queuing queue corresponding to the risk-free tag is identified as the target queuing queue corresponding to the risky user type in subsequent steps.

[0128] In cases where the risk user type includes both risky and non-risky users, the mapping between users carrying risk labels and the first queuing queue, as well as the mapping between users not carrying risk labels and the second queuing queue, can also be pre-configured.

[0129] The first queue and the second queue can be two different queues. The first queue can include multiple second request identifiers, each arranged sequentially. The order of the second request identifiers represents the response order of the pending access requests corresponding to each second request identifier. The second queue can also include multiple second request identifiers, each arranged sequentially. The order of the second request identifiers represents the response order of the pending access requests corresponding to each second request identifier.

[0130] The implementation of risk user types including multiple risk levels, as well as the implementation of risk user types including other pre-configured risk user types, can be found in the corresponding description section for risk user types including risk users and non-risk users.

[0131] The second request identifier of the response access request is placed in the target queuing queue corresponding to the risk user type. This can be done by inserting the second request identifier of the response access request in the middle of the target queuing queue, or by inserting the second request identifier of the response access request after the last second request identifier in the target queuing queue.

[0132] Typically, the target queuing queue may include at least one second request identifier. Each of these at least one second request identifier may be a second request identifier carried in the response access request of a user other than the target user before executing step S108. The process for placing the second request identifier carried in the response access request of other users into the target queuing queue can be referred to... Figure 1 Steps S102-S108 of the response method provided in the embodiment.

[0133] To facilitate the distinction between the second request identifier included in the target queue before execution step S108 and the second request identifier carried in the target user's response access request, the second request identifier carried in the target user's response access request can also be referred to as the target request identifier.

[0134] The second request identifier of the response access request is placed in the target queuing queue corresponding to the risky user type, or the target request identifier is placed in the target queuing queue corresponding to the risky user type.

[0135] Before executing step S108, among the multiple second request identifiers included in the target queuing queue, each second request identifier is arranged sequentially. The order of the second request identifiers is used to characterize the expected response order of the access requests to be processed corresponding to the second request identifiers. The access requests to be processed corresponding to the second request identifiers can be response access requests from users other than the aforementioned target user.

[0136] After executing step S108, the target queuing queue includes multiple second request identifiers, including both second request identifiers carried in the response access requests of other users and the target request identifier. Each second request identifier is arranged sequentially, and the order of the second request identifiers represents the expected response order of the access requests to be processed corresponding to the second request identifier. The access request to be processed corresponding to the second request identifier can be a response access request from a user other than the aforementioned target user, or it can be a response access request from the target user.

[0137] Optionally, placing the response access request in the target queuing queue corresponding to the risk user type includes: placing the second request identifier of the response access request in the target queuing queue such that the second request identifier of the response access request is located after the last second request identifier included in the target queuing queue; the target queuing queue includes a plurality of second request identifiers arranged sequentially according to the order of placement time; each of the plurality of second request identifiers corresponds to a pending access request waiting to access the target human agent; the target human agent is the human agent corresponding to the risk user type.

[0138] In practice, each queuing queue can correspond to one risky user type or multiple risky user types. Similarly, a target queuing queue can correspond to one risky user type or multiple risky user types. Furthermore, one risky user type can also correspond to multiple queuing queues.

[0139] When each risk user type corresponds to multiple queuing queues, the target queuing queue to which the second request identifier of the response access request should be placed can be determined based on the priority order of the multiple queuing queues and the generation time of the response access request, and the second request identifier of the response access request can be placed into the target queuing queue.

[0140] Each of the multiple second request identifiers corresponds to a pending access request waiting to be processed by a target human agent. This pending access request can be a response access request from a user other than the target user. The second request identifier of the pending access request is placed in the target queuing queue before step S108 is executed. For a detailed implementation of the process from obtaining the pending access request to placing the second request identifier of the pending access request into the target queuing queue, please refer to [reference needed]. Figure 1 Steps S102-S108 of the response method provided in the embodiment.

[0141] For each risk user type, a corresponding agent number can be configured, thereby providing targeted human services for different risk user types and improving the satisfaction of target users.

[0142] Optionally, placing the response access request in the target queuing queue corresponding to the risk user type includes: determining the insertion position of the second request identifier of the response access request in the target queuing queue based on risk assessment data; and placing the second request identifier of the response access request in the target queuing queue based on the insertion position.

[0143] Determining the insertion position of an access request in the target queuing queue based on risk assessment data can be achieved by sorting the risk assessment data corresponding to each access request when multiple response requests are obtained simultaneously, and then determining the insertion position of each access request in the target queuing queue based on the sorting result. Based on the insertion position, the access request is then placed in the target queuing queue.

[0144] For example, simultaneously acquire access request 1 from target user 1, access request 2 from target user 2, and access request 3 from target user 3. Then, simultaneously execute the following steps:

[0145] (c1) Based on the dialogue identifier carried by the response access request 1, query the corresponding tracking data of the dialogue identifier in the historical tracking data to obtain the first tracking data; read the number of target words from the first tracking data to obtain the first quantity corresponding to the response access request 1, and determine the first quantity corresponding to the response access request 1 as the risk judgment data of the response access request 1.

[0146] (c2) Based on the dialogue identifier carried by the response access request 2, query the corresponding tracking data of the dialogue identifier in the historical tracking data to obtain the first tracking data; read the number of target words from the first tracking data to obtain the first quantity corresponding to the response access request 2, and determine the first quantity corresponding to the response access request 2 as the risk judgment data of the response access request 2.

[0147] (c3) Based on the dialogue identifier carried by the response access request 3, query the corresponding tracking data of the dialogue identifier in the historical tracking data to obtain the first tracking data; read the number of target words from the first tracking data to obtain the first quantity corresponding to the response access request 3, and determine the first quantity corresponding to the response access request 1 as the risk judgment data of the response access request 3.

[0148] Then, the first number corresponding to access request 1, access request 2, and access request 3 are sorted to obtain the following sorting results: the first number corresponding to access request 1 is the largest, the first number corresponding to access request 3 is in the middle, and the first number corresponding to access request 2 is the smallest.

[0149] Since the target queue already includes 5 second request identifiers, the insertion position of response access request 1 in the target queue can be determined as the sixth position, the insertion position of response access request 3 in the target queue can be determined as the seventh position, and the insertion position of response access request 2 in the target queue can be determined as the eighth position.

[0150] In another implementation, the insertion position of the second request identifier for responding to the access request in the target queuing queue can be determined based on the comparison result between the risk assessment data and the preset threshold; and the second request identifier for responding to the access request can be placed in the target queuing queue based on the insertion position.

[0151] For example, the target queuing queue already includes 6 second request identifiers. The risk assessment data is the aforementioned third quantity, namely the number of target behavior tracking records corresponding to the target user identifier. If the third quantity belongs to the preset threshold range [n1, n2], the insertion position of the second request identifier for the response access request in the target queuing queue can be determined as the middle position in the target queuing queue, i.e., the fourth position in the target queuing queue. Based on the insertion position, the second request identifier for the response access request is placed in the target queuing queue, so that the second request identifier for the response access request is located in the fourth position in the target queuing queue, and there are three second request identifiers before and three second request identifiers after the second request identifier for the response access request. Of the original 6 second request identifiers included in the target queuing queue, the last three second request identifiers are each shifted one position to the right after performing the step of "placing the second request identifier for the response access request in the target queuing queue based on the insertion position".

[0152] In such Figure 1 In the illustrated response method embodiment, firstly, the response access request of the target user is obtained; the response access request carries a first request identifier and a second request identifier; the first request identifier includes at least one of a dialogue identifier and a target user identifier of the target user; the second request identifier is used to uniquely identify the response access request; secondly, risk assessment data corresponding to the response access request is determined based on at least one of the dialogue identifier and the target user identifier; then, the risk user type of the target user is determined based on the risk assessment data; finally, the second request identifier of the response access request is placed in the target queuing queue corresponding to the risk user type; each of the multiple second request identifiers included in the target queuing queue is arranged sequentially, and the arrangement order of the second request identifiers is used to characterize the expected response order of the pending access requests corresponding to the second request identifier. Therefore, by using at least one of the dialogue identifier and the target user identifier carried in the target user's response access request, the risk assessment data corresponding to the response access request can be determined. Furthermore, based on the risk assessment data, the risk user type of the target user can be determined. Each risk user type corresponds to a different queuing queue. By placing the second request identifier carried in the target user's response access request into the target queuing queue corresponding to the risk user type, target users of different risk user types can be pre-classified, indirectly reducing the number of pending access requests corresponding to each queuing queue, shortening the waiting time for response access requests, and improving user satisfaction while waiting to access a human agent.

[0153] Based on the same technical concept as the aforementioned method embodiments, this application also provides an embodiment of a block diagram of a response method. Figure 3 This is a block diagram of a response method provided in an embodiment of this application.

[0154] like Figure 3 As shown, the agent service system uses response DM301, NLP intent recognition 302, and process engine 303 to calculate whether a risk label needs to be added to the target user's response access request. The agent service system can pre-execute manual queue priority segmentation, with different segmented queues corresponding to different risk user types. It can also perform agent business line segmentation based on different business types. Based on whether the response access request carries a risk label and the risk user type corresponding to the risk label, the agent service system can place the target user's response access request into the corresponding queue to wait for an available agent to answer the call.

[0155] Based on the same technical concept as the foregoing method embodiments, this application also provides an embodiment of another response method.

[0156] Figure 4 A flowchart illustrating another response method provided in this application embodiment.

[0157] like Figure 4 As shown, firstly, the target user inputs "human customer service" 402. Then, the agent service system transmits the user input to the response system 404. Then, the response system 404 calls NLP intent recognition 406 to perform intent recognition processing on the user input. Then, it performs intent recognition classification 408, that is, converts the user input into intelligent agent knowledge classification intent.

[0158] The response system classifies the intent identified by NLP and then transfers it to a human operator. It then calls a multi-turn dialogue engine system to determine whether to start the process 410 based on the intent judgment conditions at the process entry point.

[0159] If so, return the prompt message 412 to the target user and quickly proceed to the judgment of the 9 risk user tag nodes. If any one of the business points is judged successfully, it means that the target user is a risk user. If all 9 tags fail to be judged, it means that the target user is a normal user.

[0160] The nine tag nodes are judged as follows: sensitive word verification 414, number of times to human agent verification 416, sensitive word verification of historical conversation 418, whether the call was repeated 420, whether the person is a member 422, whether there is a complaint in transit 1 (424), whether there is a historical complaint 1 (426), whether there is a complaint in transit 2 (428), and whether there is a historical complaint 2 (430) after x days.

[0161] The results are determined by the target user's tag node, and the user is ultimately transferred to a human agent. If the user is a high-risk user, the user is transferred to the special human agent channel 436; if the user is a regular user, the user is transferred to the normal human agent channel 432.

[0162] The strategy for transferring users to human agents can be to determine whether the channel identifier 438 is a normal identifier or a special identifier. If the channel identifier 438 is a normal identifier, users transferring normally to human agents are placed in the regular human agent queue segment 440. If the channel identifier 438 is a special identifier, users in the special channel are placed in the higher priority queue segment 442. This allows for agent allocation 444 to target users waiting in different segments.

[0163] The special manual channels are segmented for queuing services for high-risk users. When configuring manual agents, dedicated agents can be configured for handling these special manual channels. When an agent for a special manual channel becomes available, the agent service system will automatically assign the user to an available agent in the queue corresponding to that special manual channel.

[0164] Once a human agent connects with a target user, they can use their expertise to solve the user's problems, thus enabling professional services to be handled by professional agents.

[0165] After the agent has met the needs of the current target user and determined that the service can be terminated, the connection with the current target user is terminated, and other target users are assigned to the line.

[0166] Since the technical concept is the same, the description in this embodiment is relatively simple. For the relevant parts, please refer to the corresponding description of the response method embodiment provided above.

[0167] Based on the same technical concept as the foregoing method embodiments, this application also provides an embodiment of another response method. Figure 5 This is a flowchart illustrating a response method provided in an embodiment of this application.

[0168] like Figure 5 As shown, upon entering a multi-round dialogue prompt 502, the following steps are taken: sensitive word verification result 504, number of times to human agent is requested 506, whether sensitive words were triggered in the past dialogue 508, whether there were repeated calls 510, whether the user is a member 512, whether there are any first-type in-transit complaints 514, whether there are any first-type historical complaints 516, whether there are any second-type in-transit complaints 518, and whether there are any second-type historical complaints 520. Based on one or more of the above-determined 504-520, the risk user type 522 can be determined.

[0169] In the above embodiments, a response method is provided, and correspondingly, a response device is also provided, which will be described below with reference to the accompanying drawings.

[0170] Figure 6 This is a schematic diagram of a response device provided in an embodiment of this application.

[0171] This embodiment provides a response device, including:

[0172] The acquisition unit 601 is used to acquire the response access request of the target user; the response access request carries a first request identifier and a second request identifier; the first request identifier includes at least one of a dialogue identifier and a target user identifier of the target user; the second request identifier is used to uniquely identify the response access request.

[0173] The first determining unit 602 is used to determine the risk assessment data corresponding to the response access request based on at least one of the dialogue identifier and the target user identifier.

[0174] The second determining unit 603 is used to determine the risk user type of the target user based on the risk assessment data;

[0175] The placement unit 604 is used to place the second request identifier of the response access request into the target queuing queue corresponding to the risk user type; each second request identifier in the target queuing queue is arranged sequentially, and the arrangement order of the second request identifiers is used to characterize the expected response order of the access request to be processed corresponding to the second request identifier.

[0176] Optionally, the response device further includes:

[0177] The dialogue acquisition unit is used to acquire dialogue data between the target user and the intelligent agent;

[0178] The third determining unit is used to determine whether the dialogue data includes at least one target word that matches a sensitive word, based on a pre-configured set of sensitive words that includes multiple sensitive words.

[0179] The data tracking unit is used to perform data tracking based on the target word if the dialogue data includes at least one target word, and obtain the tracking data.

[0180] Optionally, the first determining unit 602 is specifically used for:

[0181] Based on the dialogue identifier, query the corresponding event data in the historical event data to obtain the first event data;

[0182] The number of target words is read from the first data point to obtain the first quantity, which is then used as the risk assessment data.

[0183] The second determining unit 603 is specifically used for:

[0184] If the first quantity is greater than the first preset quantity threshold, then the risk user type of the target user is determined as the first risk type.

[0185] Optionally, the first determining unit 602 is specifically used for:

[0186] Based on the preset time length and target user identifier, query the tracking data corresponding to the target user identifier in the historical tracking data to obtain the second tracking data;

[0187] The number of target words is read from the second data point to obtain the second quantity, which is then used as the risk assessment data.

[0188] The second determining unit 603 is specifically used for:

[0189] If the second quantity is greater than the second preset quantity threshold, then the risk user type of the target user is determined as the second risk type.

[0190] Optionally, the first determining unit 602 is specifically used for:

[0191] Based on the target user identifier, query the preset storage space to see if there is at least one target complaint work order, obtain the work order query results, and determine the work order query results as risk assessment data; the target complaint work order meets the preset work order screening conditions; the preset storage space stores the correspondence between user identifiers and complaint work orders;

[0192] The second determining unit 603 is specifically used for:

[0193] If the work order query result indicates that there is at least one target complaint work order, then the risk user type of the target user is determined as the third risk type.

[0194] Optionally, the response device further includes:

[0195] The triggering unit is used to trigger a request for human agent access if the number of target words included in the dialogue data is greater than a third preset threshold.

[0196] The generation unit is used to generate a target behavior tracking record if the connection to a human agent is successfully established through a human agent access request.

[0197] The first determining unit 602 is specifically used for:

[0198] Based on the target user identifier, query the number of target behavior tracking records corresponding to the target user identifier from the historical tracking data to obtain the third quantity, and determine the third quantity as the risk assessment data.

[0199] The second determining unit 603 is specifically used for:

[0200] If the third quantity is greater than the fourth quantity threshold, then the risk user type of the target user is determined as the fourth risk type.

[0201] Optionally, the first determining unit 602 is specifically used for:

[0202] Based on the target user identifier and a preset time range, the number of dialogue data corresponding to the target user identifier is retrieved from historical dialogue data to obtain the fourth quantity; and based on the target user identifier and a preset time range, the number of interactive voice response (IVR) records corresponding to the target user identifier is retrieved from historical call records to obtain the fifth quantity; the fourth and fifth quantities are determined as risk assessment data;

[0203] The second determining unit 603 is specifically used for:

[0204] If the fourth quantity is greater than the fifth quantity threshold and the fifth quantity is greater than the sixth quantity threshold, then the risk user type of the target user is determined as the fifth risk type.

[0205] Optionally, unit 604 is placed for:

[0206] The second request identifier that responds to the access request is placed in the target queuing queue, such that the second request identifier that responds to the access request is located after the last second request identifier included in the target queuing queue; the target queuing queue includes multiple second request identifiers arranged in order of placement time; each of the multiple second request identifiers corresponds to a pending access request waiting to access the target human agent; the target human agent is the human agent corresponding to the risk user type.

[0207] Optionally, unit 604 is placed for:

[0208] Determine the insertion position of the second request identifier in the target queue based on the risk assessment data;

[0209] Based on the insertion position, the second request identifier of the response access request is placed in the target queuing queue.

[0210] In this embodiment, the response device includes an acquisition unit, a first determination unit, a second determination unit, and a placement unit. The acquisition unit acquires a response access request from a target user. The response access request carries a first request identifier and a second request identifier. The first request identifier includes at least one of a dialogue identifier and a target user identifier of the target user. The second request identifier uniquely identifies the response access request. The first determination unit determines risk assessment data corresponding to the response access request based on at least one of the dialogue identifier and the target user identifier. The second determination unit determines the risk user type of the target user based on the risk assessment data. The placement unit places the second request identifier of the response access request in a target queuing queue corresponding to the risk user type. Each second request identifier in the target queuing queue is arranged sequentially, and the arrangement order of the second request identifiers represents the expected response order of the pending access requests corresponding to the second request identifier. Therefore, by using at least one of the dialogue identifier and the target user identifier carried in the target user's response access request, the risk assessment data corresponding to the response access request can be determined. Furthermore, based on the risk assessment data, the risk user type of the target user can be determined. Each risk user type corresponds to a different queuing queue. By placing the second request identifier carried in the target user's response access request into the target queuing queue corresponding to the risk user type, target users of different risk user types can be pre-classified, indirectly reducing the number of pending access requests corresponding to each queuing queue, shortening the waiting time for response access requests, and improving user satisfaction while waiting to access a human agent.

[0211] Corresponding to the response method described above, based on the same technical concept, this application also provides an electronic device for executing the response method provided above. Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0212] like Figure 7As shown, electronic devices can vary considerably due to differences in configuration or performance. They may include one or more processors 701 and memories 702, with the memory 702 storing one or more application programs or data. The memory 702 can be temporary or persistent storage. The application programs stored in the memory 702 may include one or more modules (not shown), each module including a series of computer-executable instructions from the electronic device. Furthermore, the processor 701 may be configured to communicate with the memory 702, executing the series of computer-executable instructions stored in the memory 702 on the electronic device. The electronic device may also include one or more power supplies 703, one or more wired or wireless network interfaces 704, one or more input / output interfaces 705, one or more keyboards 706, etc.

[0213] In one specific embodiment, the electronic device includes a memory and one or more programs, wherein the one or more programs are stored in the memory, and the one or more programs may include one or more modules, and each module may include a series of computer-executable instructions for use in the electronic device, and is configured to be executed by one or more processors. The one or more programs include computer-executable instructions for performing the following:

[0214] Obtain the target user's response access request; the response access request carries a first request identifier and a second request identifier; the first request identifier includes at least one of a dialogue identifier and the target user's target user identifier; the second request identifier is used to uniquely identify the response access request.

[0215] Based on at least one of the dialogue identifier and the target user identifier, determine the risk assessment data corresponding to the response access request;

[0216] Based on risk assessment data, determine the risk user type of the target users;

[0217] The second request identifier that responds to the access request is placed in the target queuing queue corresponding to the risk user type; each of the multiple second request identifiers included in the target queuing queue is arranged sequentially, and the arrangement order of the second request identifiers is used to characterize the expected response order of the access requests to be processed corresponding to the second request identifier.

[0218] Corresponding to the response method described above, and based on the same technical concept, embodiments of this application also provide a computer-readable storage medium.

[0219] The computer-readable storage medium provided in this embodiment is used to store computer-executable instructions, which, when executed by a processor, implement the following process:

[0220] Obtain the target user's response access request; the response access request carries a first request identifier and a second request identifier; the first request identifier includes at least one of a dialogue identifier and the target user's target user identifier; the second request identifier is used to uniquely identify the response access request.

[0221] Based on at least one of the dialogue identifier and the target user identifier, determine the risk assessment data corresponding to the response access request;

[0222] Based on risk assessment data, determine the risk user type of the target users;

[0223] The second request identifier that responds to the access request is placed in the target queuing queue corresponding to the risk user type; each of the multiple second request identifiers included in the target queuing queue is arranged sequentially, and the arrangement order of the second request identifiers is used to characterize the expected response order of the access requests to be processed corresponding to the second request identifier.

[0224] It should be noted that the embodiments of the computer-readable storage medium in this specification and the embodiments of the response method in this specification are based on the same inventive concept. Therefore, the specific implementation of this embodiment can be referred to the implementation of the corresponding method described above, and the repeated parts will not be described again.

[0225] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0226] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, embodiments of this application can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, this specification can take the form of a computer program product embodied on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0227] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0228] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0229] These computer program instructions may also be loaded onto a computer or other programmable device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable device for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0230] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0231] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0232] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0233] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0234] The embodiments of this application can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. One or more embodiments of this specification can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules can reside in local and remote computer storage media, including storage devices.

[0235] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0236] The above description is merely an embodiment of this document and is not intended to limit the scope of this document. Various modifications and variations can be made to this document by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this document should be included within the scope of the claims of this document.

Claims

1. A response method, characterized in that, include: Acquire dialogue data between the target user and the intelligent agent; Based on a pre-configured set of sensitive words including multiple sensitive words, it is determined whether the dialogue data includes at least one target word that matches the sensitive word; if the dialogue data includes at least one target word, data tracking is performed based on the target word to obtain tracking data, and historical tracking data is obtained based on the tracking data. Obtain the target user's response access request; the response access request carries a first request identifier and a second request identifier; the first request identifier includes at least one of a dialogue identifier and the target user's target user identifier; the second request identifier is used to uniquely identify the response access request; Based on at least one of the dialogue identifier and the target user identifier, determine the risk assessment data corresponding to the response access request from the historical data. Based on the risk assessment data, the risk user type of the target user is determined; Place the second request identifier of the response access request into the target queuing queue corresponding to the risky user type; The target queuing queue includes multiple second request identifiers, each of which is arranged sequentially. The order of the second request identifiers is used to characterize the expected response order of the access requests to be processed corresponding to the second request identifier.

2. The method according to claim 1, characterized in that, Based on the dialogue identifier, risk assessment data corresponding to the response access request is determined from the historical data, including: Based on the dialogue identifier, query the corresponding event data in the historical event data to obtain the first event data; The number of target words is read from the first data point data to obtain a first number, and the first number is determined as the risk assessment data; The step of determining the risk user type of the target user based on the risk assessment data includes: If the first quantity is greater than the first preset quantity threshold, then the risk user type of the target user is determined as the first risk type.

3. The method according to claim 1, characterized in that, Based on the target user identifier, risk assessment data corresponding to the response access request is determined from the historical data, including: Based on the preset time length and the target user identifier, query the tracking data corresponding to the target user identifier in the historical tracking data to obtain the second tracking data; The number of target words is read from the second data point to obtain a second number, and the second number is determined as the risk assessment data. The step of determining the risk user type of the target user based on the risk assessment data includes: If the second quantity is greater than the second preset quantity threshold, then the risk user type of the target user is determined as the second risk type.

4. The method according to claim 1, wherein risk assessment data corresponding to the response access request is determined from the historical data based on the target user identifier, comprising: Based on the target user identifier, query the preset storage space to see if there is at least one target complaint work order, obtain the work order query result, and determine the work order query result as the risk judgment data; The target complaint work order meets the preset work order screening conditions; The preset storage space stores the correspondence between user identifiers and complaint work orders; The step of determining the risk user type of the target user based on the risk assessment data includes: If the work order query result indicates that there is at least one target complaint work order, then the risk user type of the target user is determined as the third risk type.

5. The method according to claim 1, characterized in that, After determining whether the dialogue data includes at least one target word matching the sensitive word based on a pre-configured sensitive word set containing multiple sensitive words, the process further includes: If the number of target words included in the dialogue data is greater than a third preset threshold, a request for human agent access is triggered. If the connection to a human agent is successfully established through the aforementioned human agent access request, a target behavior tracking record is generated. Based on the target user identifier, determine the risk assessment data corresponding to the response access request, including: Based on the target user identifier, the number of target behavior tracking records corresponding to the target user identifier is queried from the historical tracking data to obtain a third quantity, and the third quantity is determined as the risk judgment data. The step of determining the risk user type of the target user based on the risk assessment data includes: If the third quantity is greater than the fourth quantity threshold, then the risk user type of the target user is determined as the fourth risk type.

6. The method according to claim 1, characterized in that, Based on the target user identifier, risk assessment data corresponding to the response access request is determined from the historical data, including: Based on the target user identifier and a preset time range, the number of dialogue data corresponding to the target user identifier is retrieved from historical dialogue data to obtain a fourth quantity; and based on the target user identifier and a preset time range, the number of interactive voice response (IVR) records corresponding to the target user identifier is retrieved from historical call records to obtain a fifth quantity; the fourth quantity and the fifth quantity are determined as the risk assessment data; The step of determining the risk user type of the target user based on the risk assessment data includes: If the fourth quantity is greater than the fifth quantity threshold and the fifth quantity is greater than the sixth quantity threshold, then the risk user type of the target user is determined as the fifth risk type.

7. The method according to claim 1, characterized in that, The step of placing the second request identifier of the response access request into the target queuing queue corresponding to the risky user type includes: The second request identifier of the response access request is placed in the target queuing queue, such that the second request identifier of the response access request is located after the last second request identifier included in the target queuing queue; the target queuing queue includes a plurality of second request identifiers arranged in order of placement time; each of the plurality of second request identifiers corresponds to a pending access request waiting to access the target human agent; the target human agent is the human agent corresponding to the risk user type.

8. The method according to claim 1, characterized in that, The step of placing the second request identifier of the response access request into the target queuing queue corresponding to the risky user type includes: Based on the risk assessment data, determine the insertion position of the second request identifier of the response access request in the target queuing queue; Based on the insertion position, the second request identifier of the response access request is inserted into the target queuing queue.

9. A response device, characterized in that, include: The acquisition unit is used to acquire the target user's response access request; The response access request carries a first request identifier and a second request identifier; The first request identifier includes at least one of a dialogue identifier and a target user identifier of the target user; the second request identifier is used to uniquely identify the response access request; The first determining unit is configured to determine the risk assessment data corresponding to the response access request based on at least one of the dialogue identifier and the target user identifier. The second determining unit is used to determine the risk user type of the target user based on the risk assessment data; A placement unit is used to place the second request identifier of the response access request into the target queuing queue corresponding to the risky user type; The target queuing queue includes multiple second request identifiers, each of which is arranged sequentially. The order of the second request identifiers is used to characterize the expected response order of the access requests to be processed corresponding to the second request identifier.

10. An electronic device, characterized in that, include: processor; And a memory configured to store computer-executable instructions, which, when executed, cause the processor to perform the response method as described in any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store computer-executable instructions that, when executed by a processor, implement the response method as described in any one of claims 1-8.

Images