Response task allocation method and device, equipment and storage medium thereof

Abstract

The embodiment of the invention belongs to the technical field of artificial intelligence and intelligent service, is applied to a business consultation seat distribution scene, and relates to a response task distribution method, device and equipment and a storage medium thereof. Outputting an optimal consultation intention in combination with the intention prediction model; taking the optimal consultation intention as a screening basis, and screening out a target response seat for the target user side; and establishing a real-time communication relationship between the target user side and the target response seat, and completing the response task distribution. An intention prediction model is adopted, the optimal consultation intention of consultation data is obtained, and seat task allocation is carried out, so that a business service party can accurately allocate seats according to the consultation intention, the seats which are good at services corresponding to the consultation intention can be screened out for target users, consultation services can be reasonably allocated for seat personnel, and the consultation efficiency is improved. Therefore, the overall processing efficiency of the whole seat department on various consultation businesses is improved, and blind seat distribution is avoided.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence and smart service technology, and is applied to business consultation seat allocation scenarios, and in particular to a response task allocation method, device, equipment and storage medium thereof. Background Art

[0002] At present, service-oriented companies have many service products. For example, insurance service products in the financial business field often have a large number of users, but there is a serious imbalance in the ratio between the number of service seats and users. They are unable to provide users with high-quality services. Blindly expanding the recruitment of service seats is obviously not in line with the company's actual recruitment needs.

[0003] Therefore, some companies have introduced intelligent customer service to assist human service agents in answering questions. However, intelligent customer service has certain business limitations and cannot completely replace human service agents. This has led to the following problems in service, despite the introduction of intelligent customer service: First, service agents need to contact and communicate with many users every day, which is time-consuming and labor-intensive, and has low business processing efficiency. Second, service agents have limited knowledge reserves, and the questions asked by users are often very business-oriented or professional, which cannot guarantee the service quality of service agents, resulting in the allocation of service agents being unscientific. Summary of the invention

[0004] The purpose of the embodiments of the present application is to propose a method, device, equipment and storage medium for allocating answering tasks to solve the problems that the allocation of service seats is not scientific enough and the business processing efficiency is low when allocating answering tasks.

[0005] In order to solve the above technical problems, the embodiment of the present application provides a response task allocation method, which adopts the following technical solution:

[0006] A method for allocating response tasks comprises the following steps:

[0007] Receive consultation data from the target user end;

[0008] Inputting the consultation data into a pre-trained intention prediction model to obtain the best consultation intention output by the intention prediction model;

[0009] Using the best consultation intention as a screening basis, a target answering agent is screened out for the target user terminal from a preset consultation intention and answering agent relationship comparison table;

[0010] A real-time communication relationship is established between the target user terminal and the target answering agent to complete the answering task allocation.

[0011] Furthermore, the pre-trained intention prediction model includes a word representation layer, a context modeling layer, a sentence representation layer, a question-answer intention matching layer, and an optimal intention screening layer. The step of inputting the consultation data into the pre-trained intention prediction model to obtain the optimal consultation intention output by the intention prediction model specifically includes:

[0012] Mapping the text data contained in the consulting data to the target digital vector space through the word representation layer to obtain the word digital vector representation corresponding to the consulting data;

[0013] By means of the context modeling layer, constructing a word vector dependency relationship between text data included in the consulting data;

[0014] Based on the word digital vector representation and the word vector dependency, generating a sentence vector representation matrix corresponding to the consultation data at the sentence representation layer;

[0015] Calculate the correlation between the sentence vector representation matrix and the sentence vector representation matrix of all answer data provided in the question-answer intention matching layer to obtain a candidate answer data sequence arranged from high to low correlation;

[0016] Input the candidate answer data sequence into the best intention screening layer, and screen out the answer data with the highest correlation as the target answer data through a screening mechanism preset in the best intention screening layer, wherein the preset screening mechanism includes screening the answer data at the first position of the candidate answer data sequence;

[0017] The business intention corresponding to the target answer data is output from the intention prediction model as the optimal consulting intention.

[0018] Furthermore, the step of mapping the text data contained in the consulting data to the target digital vector space through the word representation layer to obtain the word digital vector representation corresponding to the consulting data specifically includes:

[0019] According to a preset sentence division strategy, the consultation data is divided into sentences to obtain a sentence division result, wherein the preset sentence division strategy includes dividing sentences according to specified punctuation marks;

[0020] Based on a preset sentence length value, performing secondary processing on the sentence segmentation result to obtain all target sentences after secondary processing;

[0021] Input all the target sentences after the secondary processing into the word vector model pre-trained by the Word2Vec tool to obtain the word vector sequences corresponding to all the target sentences respectively;

[0022] According to the sequential position information of all the target sentences in the consultation data, the word vector sequences corresponding to all the target sentences are concatenated to obtain a concatenated word vector sequence;

[0023] The concatenated word vector sequence is used as the word digital vector representation corresponding to the consultation data.

[0024] Furthermore, the step of performing secondary processing on the sentence segmentation result based on the preset sentence length value to obtain all target sentences after the secondary processing specifically includes:

[0025] Count the number of text characters contained in each sentence, and use the statistical result as the sentence length value of the corresponding sentence;

[0026] By comparison, the relationship between the sentence length value of each sentence and the pre-set sentence length value is identified;

[0027] If the sentence length value of the current sentence does not exceed the preset sentence length value, the current sentence is padded to the preset sentence length using a preset padding character, and the padded sentence is obtained as the target sentence after the current sentence is processed twice;

[0028] If the sentence length value of the current sentence exceeds the pre-set sentence length value, the pre-set sentence length value is used as the segmentation length, and the current sentence is segmented. The sub-sentences that do not reach the segmentation length after segmentation are filled to the pre-set sentence length using the preset filling character, and all sub-sentences with a length of the segmentation length are obtained as the target sentences after the current sentence is processed twice.

[0029] Furthermore, the context modeling layer is composed of a multi-head self-attention mechanism and a BiLSTM bidirectional long short-term memory network in a transformer neural network. The step of constructing the word vector dependency relationship between the text data contained in the consulting data through the context modeling layer specifically includes:

[0030] Calculate the self-attention similarity between all word vectors in the consulting data through the multi-head self-attention mechanism in the transformer neural network;

[0031] By using the BiLSTM bidirectional long short-term memory network, the sequence features corresponding to all the word vectors in the consulting data are extracted as the implicit vector representations corresponding to all the word vectors;

[0032] According to the self-attention similarities between all word vectors in the consulting data and the implicit vector representations corresponding to all the word vectors respectively, the word vector dependency relationship between the text data contained in the consulting data is generated.

[0033] Furthermore, the step of calculating the correlation between the sentence vector representation matrix and the sentence vector representation matrix of all answer data provided in the question-answer intention matching layer to obtain a candidate answer data sequence arranged from high to low correlation specifically includes:

[0034] Using a cosine similarity algorithm, calculate the cosine similarity value between the sentence vector representation matrix and the sentence vector representation matrix of all answer data provided in the question-answer intention matching layer;

[0035] The cosine similarity value is used as a correlation representation value, and all answer data provided in the question-answer intention matching layer are sorted according to the magnitude relationship of the cosine similarity value;

[0036] The ranking processing result is obtained as a candidate solution data sequence arranged from high to low relevance.

[0037] Furthermore, the preset consulting intention and answering seat relationship comparison table includes the answer evaluation values ​​of each answering seat for different consulting intentions. The step of using the best consulting intention as the screening basis and screening out the target answering seat for the target user terminal from the preset consulting intention and answering seat relationship comparison table specifically includes:

[0038] Filter out all answering seats that have an answering relationship with the best consulting intention from the consulting intention and answering seat relationship comparison table;

[0039] Identify the response evaluation values ​​of all the response seats to the best consultation intention through the response evaluation values ​​of each response seat to different consultation intentions;

[0040] According to the response evaluation values ​​of all the response seats to the best consultation intention, all the response seats are sorted to generate a response seat sequence arranged from high to low according to the response evaluation values;

[0041] The answering seat which is at the front of the sequence and in the call idle state is selected from the answering seat sequence as the target answering seat.

[0042] In order to solve the above technical problems, the embodiment of the present application also provides a response task allocation device, which adopts the following technical solution:

[0043] A response task allocation device, comprising:

[0044] A consultation data receiving module, used for receiving consultation data from a target user terminal;

[0045] The best consultation intention prediction module is used to input the consultation data into the pre-trained intention prediction model to obtain the best consultation intention output by the intention prediction model;

[0046] A target answering agent screening module is used to use the best consultation intention as a screening basis and screen out a target answering agent for the target user terminal from a preset consultation intention and answering agent relationship comparison table;

[0047] The real-time communication relationship establishment module is used to establish a real-time communication relationship between the target user terminal and the target answering agent to complete the current answering task allocation.

[0048] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0049] A computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the above-mentioned response task allocation method when executing the computer-readable instructions.

[0050] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0051] A computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the response task allocation method as described above.

[0052] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0053] The answering task allocation method described in the embodiment of the present application receives consulting data from the target user terminal; inputs the consulting data into a pre-trained intention prediction model to obtain the best consulting intention output by the intention prediction model; uses the best consulting intention as a screening basis to screen out a target answering agent for the target user terminal from a preset consulting intention and answering agent relationship comparison table; establishes a real-time communication relationship between the target user terminal and the target answering agent to complete the answering task allocation. The pre-trained intention prediction model is used to obtain the best consulting intention of the consulting data, and the agent task allocation is performed based on the best consulting intention, so that the business service provider can accurately allocate seats according to the consulting intention, which can not only screen out seats that are good at the business corresponding to the consulting intention for the target user, but also reasonably allocate consulting services to the agents, thereby improving the overall processing efficiency of the entire agent department for various consulting services and avoiding blind seat allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0055] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0056] Figure 2 is a flow chart of an embodiment of a method for allocating response tasks according to the present application;

[0057] Figure 3 yes Figure 2 A flowchart of a specific embodiment of step 202 is shown;

[0058] Figure 4 yes Figure 3 A flowchart of a specific embodiment of step 301 is shown;

[0059] Figure 5 yes Figure 4 A flowchart of a specific embodiment of step 402 is shown;

[0060] Figure 6 yes Figure 3 A flowchart of a specific embodiment of step 302 is shown;

[0061] Figure 7 yes Figure 3 A flowchart of a specific embodiment of step 304 is shown;

[0062] Figure 8 yes Figure 2 A flowchart of a specific embodiment of step 203 is shown;

[0063] Fig. 9 is a structural schematic diagram of an embodiment of a response task allocation device according to the present application;

[0064] Fig.10 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0065] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0066] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0067] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0068] like Figure 1 As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables.

[0069] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0070] The terminal device 101 can be any electronic device with a display screen and supporting web browsing. In addition to a laptop computer 1011, a tablet computer 1012 or a mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV), a laptop computer, a desktop computer, etc.

[0071] The server 103 may be a server that provides various services, such as a background server that provides support for a web page displayed on the terminal device 101 .

[0072] It should be noted that the response task allocation method provided in the embodiment of the present application is generally executed by a server, and accordingly, the response task allocation device is generally set in the server.

[0073] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0074] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for allocating response tasks according to the present application. The method for allocating response tasks comprises the following steps:

[0075] Step 201, receiving consultation data from a target user terminal.

[0076] In this embodiment, the target user terminal refers to the execution terminal used by the target user for business consultation, for example: a mobile phone terminal used by the user to conduct business consultation while in a call state, and a customer input terminal used by the user to conduct business consultation while in a network connection state; the consultation data includes voice call data for business consultation, and also includes consultation box input data for business consultation.

[0077] In this embodiment, the reception is different according to the form of the consultation data. For example, if the consultation data is voice call data for business consultation, the receiving method is a voice signal receiving method based on voice data; if the consultation data is consultation box input data for business consultation, the receiving method is a character stream receiving method based on text data.

[0078] By receiving the consultation data from the target user, the business consultation intention of the target user can be predicted and identified based on the consultation data. This ensures that the business service provider can accurately allocate seats to the target user based on the business consultation intention, improve the business adaptability and processing efficiency of the seats, and avoid blind seat allocation.

[0079] Step 202: input the consultation data into a pre-trained intention prediction model to obtain the best consultation intention output by the intention prediction model.

[0080] By inputting the consulting data into a pre-trained intention prediction model, the optimal consulting intention output by the intention prediction model is obtained, so that the business service provider can accurately allocate seats to the target users based on the optimal consulting intention. It can not only screen out seats for the target users that are good at the business corresponding to the consulting intention, but also reasonably allocate consulting business to the seats, thereby improving the overall processing efficiency of the entire seat department for various consulting businesses and avoiding blind seat allocation.

[0081] Step 203, taking the best consultation intention as a screening basis, a target answering agent is screened out for the target user terminal from a preset consultation intention and answering agent relationship comparison table.

[0082] By taking the best consulting intention as the screening basis, the target answering seat is screened out for the target user end from the preset consulting intention and answering seat relationship comparison table, so that the business service provider can accurately allocate seats to the target user according to the best consulting intention, and can not only screen out seats for the target user who are good at the business corresponding to the consulting intention, but also reasonably allocate consulting business to the seats, thereby improving the overall processing efficiency of the entire seat department for various consulting businesses and avoiding blind seat allocation.

[0083] Step 204: establish a real-time communication relationship between the target user terminal and the target answering agent to complete the answering task allocation.

[0084] In this embodiment, the real-time communication relationship between the target user terminal and the target answering agent is established. For example, during a call consultation, a real-time call relationship between the target user terminal and the target answering agent can be established by transferring. Another example is that during consultation in the intelligent customer service consultation box, a consultation communication box for the target user terminal to communicate with the agent can be constructed by matching the communication address. This achieves the goal of screening out agents who are good at the business corresponding to the consultation intention for the target user, and can also reasonably allocate consultation business to the agent personnel, thereby improving the overall processing efficiency of the entire agent department for various consultation businesses and avoiding blind seat allocation.

[0085] In this embodiment, the consultation data of the target user end is received; the consultation data is input into the pre-trained intention prediction model to obtain the best consultation intention output by the intention prediction model; the best consultation intention is used as the screening basis to screen out the target answering seat for the target user end from the preset consultation intention and answering seat relationship comparison table; a real-time communication relationship is established between the target user end and the target answering seat to complete the answering task allocation. The pre-trained intention prediction model is used to obtain the best consultation intention of the consultation data, and the seat task allocation is performed based on the best consultation intention, so that the business service provider can accurately allocate seats according to the consultation intention, which can not only screen out seats that are good at the business corresponding to the consultation intention for the target user, but also reasonably allocate consultation business to the seat personnel, thereby improving the overall processing efficiency of the entire seat department for various consultation businesses and avoiding blind seat allocation.

[0086] In this embodiment, the pre-trained intent prediction model includes a word representation layer, a context modeling layer, a sentence representation layer, a question-answer intent matching layer, and an optimal intent screening layer.

[0087] Continue to refer Figure 3 , in some optional implementations, Figure 3 yes Figure 2 The flowchart of a specific embodiment of step 202 shown includes the following steps:

[0088] Step 301, mapping the text data contained in the consulting data to the target digital vector space through the word representation layer to obtain the word digital vector representation corresponding to the consulting data;

[0089] Step 302, constructing a word vector dependency relationship between text data included in the consulting data through the context modeling layer;

[0090] Step 303, generating a sentence vector representation matrix corresponding to the consultation data at the sentence representation layer based on the word digital vector representation and the word vector dependency relationship;

[0091] Step 304, calculating the correlation between the sentence vector representation matrix and the sentence vector representation matrix of all answer data provided in the question-answer intention matching layer, and obtaining a candidate answer data sequence arranged in descending order of correlation;

[0092] Step 305, inputting the candidate solution data sequence into the best intention screening layer, and screening out the solution data with the highest correlation as the target solution data through the screening mechanism preset in the best intention screening layer, wherein the preset screening mechanism includes screening the solution data at the first position of the candidate solution data sequence;

[0093] Step 306: Output the business intention corresponding to the target answer data from the intention prediction model as the optimal consulting intention.

[0094] By inputting the consulting data into a pre-trained intention prediction model, the processing results of each processing layer in the intention prediction model are obtained, and finally the optimal consulting intention is output, so that the business service provider can accurately allocate seats to the target users according to the optimal consulting intention. It can not only screen out seats for the target users that are good at the business corresponding to the consulting intention, but also reasonably allocate consulting business to the seats, thereby improving the overall processing efficiency of the entire seat department for various consulting businesses and avoiding blind seat allocation.

[0095] Continue to refer Figure 4 , in some optional implementations, Figure 4 yes Figure 3 The flowchart of a specific embodiment of step 301 shown includes the following steps:

[0096] Step 401, according to a preset sentence division strategy, the consultation data is sentence-divided to obtain a sentence division result, wherein the preset sentence division strategy includes sentence division according to specified punctuation marks;

[0097] Step 402, based on a preset sentence length value, performing secondary processing on the sentence segmentation result to obtain all target sentences after secondary processing;

[0098] Step 403, inputting all the target sentences after the secondary processing into the word vector model pre-trained by the Word2Vec tool to obtain the word vector sequences corresponding to all the target sentences respectively;

[0099] Step 404, according to the sequential position information of all the target sentences in the consultation data, concatenate the word vector sequences corresponding to all the target sentences to obtain a concatenated word vector sequence;

[0100] Step 405: Use the concatenated word vector sequence as the word digital vector representation corresponding to the consultation data.

[0101] Since the neural network structure involved in the intention prediction model cannot directly process and recognize language text data, the text data contained in the consultation data is mapped to the target digital vector space through the word representation layer to obtain the word digital vector representation corresponding to the consultation data, so as to facilitate the computer to combine with the neural network to perform intent recognition and processing.

[0102] Continue to refer Figure 5, in some optional implementations, Figure 5 yes Figure 4 The flowchart of a specific embodiment of step 402 shown includes the following steps:

[0103] Step 501, counting the number of text characters contained in each sentence, and using the statistical result as the sentence length value of the corresponding sentence;

[0104] Step 502, identifying the relationship between the sentence length value of each sentence and the preset sentence length value by comparison;

[0105] Step 503, if the sentence length value of the current sentence does not exceed the preset sentence length value, the current sentence is padded to the preset sentence length using a preset padding character, and the padded sentence is obtained as the target sentence after the current sentence is processed twice;

[0106] Step 504: if the sentence length value of the current sentence exceeds the pre-set sentence length value, the pre-set sentence length value is used as the segmentation length, the current sentence is segmented, and the sub-sentences that do not reach the segmentation length after the segmentation are filled to the pre-set sentence length using the preset filling character, and all sub-sentences with a length of the segmentation length are obtained as the target sentence after the current sentence is processed twice.

[0107] According to the pre-set sentence length value, the sentence division result is secondary processed so that the length of all target sentences after the secondary processing is consistent, thereby ensuring that when all target sentences after the secondary processing are input into the word vector model pre-trained by the Word2Vec tool, the length of the word vector sequence of all target sentences obtained is also consistent. Precisely because the length of the word vector sequence of all target sentences is also consistent, one-dimensional splicing or two-dimensional splicing can be selectively performed, thereby ensuring the processing efficiency of the computer.

[0108] In this embodiment, the context modeling layer is composed of a multi-head self-attention mechanism in a transformer neural network and a BiLSTM bidirectional long short-term memory network.

[0109] Continue to refer Figure 6 , in some optional implementations, Figure 6 yes Figure 3 The flowchart of a specific embodiment of step 302 shown includes the following steps:

[0110] Step 601, calculating the self-attention similarity between all word vectors in the consultation data through the multi-head self-attention mechanism in the transformer neural network;

[0111] Step 602, extracting the sequence features corresponding to all the word vectors in the consultation data in the consultation data through the BiLSTM bidirectional long short-term memory network as the implicit vector representations corresponding to all the word vectors;

[0112] Step 603, generating a word vector dependency relationship between text data contained in the consulting data based on the self-attention similarities between all word vectors in the consulting data and the implicit vector representations corresponding to all the word vectors.

[0113] Through the multi-head self-attention mechanism in the transformer neural network and the BiLSTM bidirectional long short-term memory network, it is ensured that when the consultation data is sufficient, the word vector dependency relationship between the text data contained in the consultation data can still be accurately obtained, thereby achieving the accuracy of intent prediction.

[0114] Continue to refer Figure 7 , in some optional implementations, Figure 7 yes Figure 3 The flowchart of a specific embodiment of step 304 shown includes the following steps:

[0115] Step 701, using a cosine similarity algorithm, calculates the cosine similarity value between the sentence vector representation matrix and the sentence vector representation matrix of all answer data provided in the question-answer intention matching layer;

[0116] Step 702: using the cosine similarity value as a correlation representation value, and sorting all answer data provided in the question-answer intention matching layer according to the magnitude relationship of the cosine similarity value;

[0117] Step 703: Obtain the sorting processing result as a candidate solution data sequence arranged from high to low relevance.

[0118] By adopting the cosine similarity algorithm, the cosine similarity value of the sentence vector representation matrix and the sentence vector representation matrix of all answer data provided in the question and answer intention matching layer is calculated, thereby achieving question and answer intention matching between consultation data and answer data, making it easier to identify the user's best consultation intention based on the candidate answer data with the highest relevance.

[0119] In this embodiment, the preset consulting intention and answering agent relationship comparison table includes the answer evaluation value of each answering agent to different consulting intentions.

[0120] Continue to refer Figure 8 , in some optional implementations, Figure 8 yes Figure 2The flowchart of a specific embodiment of step 203 shown includes the following steps:

[0121] Step 801, selecting all answering seats that have an answering relationship with the best consulting intention from the consulting intention and answering seat relationship comparison table;

[0122] Step 802, identifying the response evaluation values ​​of all the response seats to the best consultation intention through the response evaluation values ​​of each response seat to different consultation intentions;

[0123] Step 803, sorting all the answering seats according to the response evaluation values ​​of all the answering seats to the best consultation intention, and generating a sequence of answering seats arranged from high to low according to the response evaluation values;

[0124] Step 804: Filter out the answering agent with the frontmost position in the sequence and in the call idle state from the answering agent sequence as the target answering agent.

[0125] Through the comparison table of the relationship between consulting intention and answering seat, combined with the best consulting intention corresponding to the consulting data, all seats with the answering evaluation values ​​corresponding to the best consulting intention are screened out for the target user, and the answering seat with the highest sequence position and in the idle state is screened out from the answering seat sequence as the target answering seat. It is possible to screen out seats that are good at the business corresponding to the consulting intention for the target user, and to reasonably allocate consulting business to the seat personnel when the answering seat is in the idle state, thereby improving the overall processing efficiency of the entire seat department for various consulting businesses and avoiding blind seat allocation.

[0126] This application receives consulting data from the target user terminal; inputs the consulting data into a pre-trained intention prediction model to obtain the best consulting intention output by the intention prediction model; uses the best consulting intention as a screening basis to screen out a target answering seat for the target user terminal from a preset consulting intention and answering seat relationship comparison table; establishes a real-time communication relationship between the target user terminal and the target answering seat to complete the answering task allocation. The pre-trained intention prediction model is used to obtain the best consulting intention of the consulting data, and the seat task allocation is performed based on the best consulting intention, so that the business service provider can accurately allocate seats according to the consulting intention, which can not only screen out seats that are good at the business corresponding to the consulting intention for the target user, but also reasonably allocate consulting services to the seat personnel, thereby improving the overall processing efficiency of the entire seat department for various consulting services and avoiding blind seat allocation.

[0127] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0128] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0129] In the embodiment of the present application, consultation data of the target user terminal is received; the consultation data is input into the pre-trained intention prediction model to obtain the best consultation intention output by the intention prediction model; the best consultation intention is used as the screening basis to screen out the target answering seat for the target user terminal from the preset consultation intention and answering seat relationship comparison table; a real-time communication relationship is established between the target user terminal and the target answering seat to complete the answering task allocation. The pre-trained intention prediction model is used to obtain the best consultation intention of the consultation data, and the seat task allocation is performed based on the best consultation intention, so that the business service provider can accurately allocate seats according to the consultation intention, which can not only screen out seats that are good at the business corresponding to the consultation intention for the target user, but also reasonably allocate consultation business to the seat personnel, thereby improving the overall processing efficiency of the entire seat department for various consultation businesses and avoiding blind seat allocation.

[0130] Further references Fig. 9 , as a response to the above Figure 2 In order to realize the method shown in the figure, the present application provides an embodiment of a response task allocation device, and the device embodiment is Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0131] like Fig. 9 As shown, the answering task allocation device 900 described in this embodiment includes: a consultation data receiving module 901, an optimal consultation intention prediction module 902, a target answering agent screening module 903 and a real-time communication relationship establishment module 904. Among them:

[0132] The consultation data receiving module 901 is used to receive the consultation data of the target user terminal;

[0133] The best consultation intention prediction module 902 is used to input the consultation data into the pre-trained intention prediction model to obtain the best consultation intention output by the intention prediction model;

[0134] The target answering agent screening module 903 is used to use the best consultation intention as a screening basis and screen out a target answering agent for the target user terminal from a preset consultation intention and answering agent relationship comparison table;

[0135] The real-time communication relationship establishing module 904 is used to establish a real-time communication relationship between the target user terminal and the target answering agent to complete the current answering task allocation.

[0136] This application receives consulting data from the target user terminal; inputs the consulting data into a pre-trained intention prediction model to obtain the best consulting intention output by the intention prediction model; uses the best consulting intention as a screening basis to screen out a target answering seat for the target user terminal from a preset consulting intention and answering seat relationship comparison table; establishes a real-time communication relationship between the target user terminal and the target answering seat to complete the answering task allocation. The pre-trained intention prediction model is used to obtain the best consulting intention of the consulting data, and the seat task allocation is performed based on the best consulting intention, so that the business service provider can accurately allocate seats according to the consulting intention, which can not only screen out seats that are good at the business corresponding to the consulting intention for the target user, but also reasonably allocate consulting services to the seat personnel, thereby improving the overall processing efficiency of the entire seat department for various consulting services and avoiding blind seat allocation.

[0137] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0138] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0139] To solve the above technical problems, the present application also provides a computer device. Fig.10 , Fig.10 This is a basic structural block diagram of the computer device in this embodiment.

[0140] The computer device 10 includes a memory 10a, a processor 10b, and a network interface 10c which are interconnected and communicated via a system bus. It should be noted that: Fig.10 Only a computer device 10 having a component memory 10a, a processor 10b, and a network interface 10c is shown, but it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to a microprocessor, an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital processor (DSP), an embedded device, etc.

[0141] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0142] The memory 10a includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 10a can be an internal storage unit of the computer device 10, such as a hard disk or memory of the computer device 10. In other embodiments, the memory 10a can also be an external storage device of the computer device 10, such as a plug-in hard disk equipped on the computer device 10, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the memory 10a can also include both the internal storage unit of the computer device 10 and its external storage device. In this embodiment, the memory 10a is generally used to store the operating system and various application software installed on the computer device 10, such as a computer-readable instruction of a method for answering task allocation, etc. In addition, the memory 10a can also be used to temporarily store various types of data that have been output or are to be output.

[0143] The processor 10b may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 10b is generally used to control the overall operation of the computer device 10. In this embodiment, the processor 10b is used to run the computer-readable instructions stored in the memory 10a or process data, such as running the computer-readable instructions of the answer task allocation method.

[0144] The network interface 10c may include a wireless network interface or a wired network interface. The network interface 10c is generally used to establish a communication connection between the computer device 10 and other electronic devices.

[0145] The computer device proposed in this embodiment belongs to the field of artificial intelligence and intelligent service technology, and is applied to the business consultation seat allocation scenario. This application receives the consultation data of the target user terminal; inputs the consultation data into the pre-trained intention prediction model to obtain the best consultation intention output by the intention prediction model; uses the best consultation intention as the screening basis, and screens out the target answering seat for the target user terminal from the preset consultation intention and answering seat relationship comparison table; establishes a real-time communication relationship between the target user terminal and the target answering seat, and completes the answering task allocation. The pre-trained intention prediction model is used to obtain the best consultation intention of the consultation data, and the seat task is allocated according to the best consultation intention, so that the business service provider can accurately allocate seats according to the consultation intention, which can not only screen out seats that are good at the business corresponding to the consultation intention for the target user, but also reasonably allocate consultation services to the seat personnel, thereby improving the overall processing efficiency of the entire seat department for various types of consultation services and avoiding blind seat allocation.

[0146] The present application also provides another implementation, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by a processor to enable the processor to perform the steps of the response task allocation method as described above.

[0147] The computer-readable storage medium proposed in this embodiment belongs to the field of artificial intelligence and intelligent service technology, and is applied to the business consultation seat allocation scenario. This application receives the consultation data of the target user terminal; inputs the consultation data into the pre-trained intention prediction model to obtain the best consultation intention output by the intention prediction model; uses the best consultation intention as the screening basis, and screens out the target answering seat for the target user terminal from the preset consultation intention and answering seat relationship comparison table; establishes a real-time communication relationship between the target user terminal and the target answering seat, and completes the answering task allocation. The pre-trained intention prediction model is used to obtain the best consultation intention of the consultation data, and the seat task is allocated according to the best consultation intention, so that the business service provider can accurately allocate seats according to the consultation intention, which can not only screen out seats that are good at the business corresponding to the consultation intention for the target user, but also reasonably allocate consultation services to the seat personnel, thereby improving the overall processing efficiency of the entire seat department for various types of consultation services and avoiding blind seat allocation.

[0148] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0149] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for allocating response tasks, characterized in that: The steps include: Receive consultation data from the target user end; Inputting the consultation data into a pre-trained intention prediction model to obtain the best consultation intention output by the intention prediction model; Using the best consultation intention as a screening basis, a target answering agent is screened out for the target user terminal from a preset consultation intention and answering agent relationship comparison table; A real-time communication relationship is established between the target user terminal and the target answering agent to complete the answering task allocation.

2. The response task allocation method according to claim 1, characterized in that: The pre-trained intention prediction model includes a word representation layer, a context modeling layer, a sentence representation layer, a question-answer intention matching layer, and an optimal intention screening layer. The step of inputting the consultation data into the pre-trained intention prediction model to obtain the optimal consultation intention output by the intention prediction model specifically includes: Mapping the text data contained in the consulting data to the target digital vector space through the word representation layer to obtain the word digital vector representation corresponding to the consulting data; By means of the context modeling layer, constructing a word vector dependency relationship between text data included in the consulting data; Based on the word digital vector representation and the word vector dependency, generating a sentence vector representation matrix corresponding to the consultation data at the sentence representation layer; Calculate the correlation between the sentence vector representation matrix and the sentence vector representation matrix of all answer data provided in the question-answer intention matching layer to obtain a candidate answer data sequence arranged from high to low correlation; Input the candidate answer data sequence into the best intention screening layer, and screen out the answer data with the highest correlation as the target answer data through a screening mechanism preset in the best intention screening layer, wherein the preset screening mechanism includes screening the answer data at the first position of the candidate answer data sequence; The business intention corresponding to the target answer data is output from the intention prediction model as the optimal consulting intention.

3. The response task allocation method according to claim 2, characterized in that: The step of mapping the text data contained in the consulting data to the target digital vector space through the word representation layer to obtain the word digital vector representation corresponding to the consulting data specifically includes: According to a preset sentence division strategy, the consultation data is divided into sentences to obtain a sentence division result, wherein the preset sentence division strategy includes dividing sentences according to specified punctuation marks; Based on a preset sentence length value, performing secondary processing on the sentence segmentation result to obtain all target sentences after secondary processing; Input all the target sentences after the secondary processing into the word vector model pre-trained by the Word2Vec tool to obtain the word vector sequences corresponding to all the target sentences respectively; According to the sequential position information of all the target sentences in the consultation data, the word vector sequences corresponding to all the target sentences are concatenated to obtain a concatenated word vector sequence; The concatenated word vector sequence is used as the word digital vector representation corresponding to the consultation data.

4. The response task allocation method according to claim 3, characterized in that: The step of performing secondary processing on the sentence segmentation result based on the preset sentence length value to obtain all target sentences after the secondary processing specifically includes: Count the number of text characters contained in each sentence, and use the statistical result as the sentence length value of the corresponding sentence; By comparison, the relationship between the sentence length value of each sentence and the pre-set sentence length value is identified; If the sentence length value of the current sentence does not exceed the preset sentence length value, the current sentence is padded to the preset sentence length using a preset padding character, and the padded sentence is obtained as the target sentence after the current sentence is processed twice; If the sentence length value of the current sentence exceeds the pre-set sentence length value, the pre-set sentence length value is used as the segmentation length, and the current sentence is segmented. The sub-sentences that do not reach the segmentation length after segmentation are filled to the pre-set sentence length using the preset filling character, and all sub-sentences with a length of the segmentation length are obtained as the target sentences after the current sentence is processed twice.

5. The response task allocation method according to claim 2, characterized in that: The context modeling layer is composed of a multi-head self-attention mechanism and a BiLSTM bidirectional long short-term memory network in a transformer neural network. The step of constructing the word vector dependency relationship between the text data contained in the consulting data through the context modeling layer specifically includes: Calculate the self-attention similarity between all word vectors in the consulting data through the multi-head self-attention mechanism in the transformer neural network; By using the Bi LSTM bidirectional long short-term memory network, the sequence features corresponding to all the word vectors in the consulting data are extracted as the implicit vector representations corresponding to all the word vectors; According to the self-attention similarities between all word vectors in the consulting data and the implicit vector representations corresponding to all the word vectors respectively, the word vector dependency relationship between the text data contained in the consulting data is generated.

6. The response task allocation method according to claim 2, characterized in that: The step of calculating the correlation between the sentence vector representation matrix and the sentence vector representation matrix of all answer data provided in the question-answer intention matching layer to obtain a candidate answer data sequence arranged from high to low correlation specifically includes: Using a cosine similarity algorithm, calculate the cosine similarity value between the sentence vector representation matrix and the sentence vector representation matrix of all answer data provided in the question-answer intention matching layer; The cosine similarity value is used as a correlation representation value, and all answer data provided in the question-answer intention matching layer are sorted according to the magnitude relationship of the cosine similarity value; The ranking processing result is obtained as a candidate solution data sequence arranged from high to low relevance.

7. The response task allocation method according to claim 1, characterized in that: The preset consulting intention and answering seat relationship comparison table includes the answer evaluation values ​​of each answering seat for different consulting intentions. The step of using the best consulting intention as a screening basis and selecting a target answering seat for the target user terminal from the preset consulting intention and answering seat relationship comparison table specifically includes: Filter out all answering seats that have an answering relationship with the best consulting intention from the consulting intention and answering seat relationship comparison table; Identify the response evaluation values ​​of all the response seats to the best consultation intention through the response evaluation values ​​of each response seat to different consultation intentions; According to the response evaluation values ​​of all the response seats to the best consultation intention, all the response seats are sorted to generate a response seat sequence arranged from high to low according to the response evaluation values; The answering seat which is at the front of the sequence and in the call idle state is selected from the answering seat sequence as the target answering seat.

8. A response task allocation device, characterized in that: include: A consultation data receiving module, used for receiving consultation data from a target user terminal; The best consultation intention prediction module is used to input the consultation data into the pre-trained intention prediction model to obtain the best consultation intention output by the intention prediction model; A target answering agent screening module is used to use the best consultation intention as a screening basis and screen out a target answering agent for the target user terminal from a preset consultation intention and answering agent relationship comparison table; The real-time communication relationship establishment module is used to establish a real-time communication relationship between the target user terminal and the target answering agent to complete the current answering task allocation.

9. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the response task allocation method according to any one of claims 1 to 7 when executing the computer-readable instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the response task allocation method according to any one of claims 1 to 7 are implemented.

Images