Case allocation method and device based on natural language processing, equipment and medium

Abstract

The application relates to the technical field of artificial intelligence, and provides a case allocation method based on natural language processing and related devices, wherein the method comprises the following steps: if a to-be-allocated case meets a trigger condition, obtaining text data corresponding to the to-be-allocated case and a business type, and target user information; obtaining a to-do item corresponding to the text data, a business sub-type of the to-do item and a processing level based on a natural language processing model corresponding to the business type; if a case record of the to-do item exists in the target user information, searching for customer service information of a historical customer service corresponding to the to-be-allocated case from a pre-stored customer service information set; searching for a target customer service from customer services except the historical customer service based on the customer service information of the historical customer service, the business sub-type and the processing level; and sending the to-be-allocated case to the target customer service based on the to-do item. By using the application, the accuracy and processing efficiency of case allocation can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and mainly relates to a case allocation method and device based on natural language processing, equipment and medium. BACKGROUND

[0002] With the rapid development of the Internet and the improvement of people's service consciousness, network customer service is deployed in the corresponding application programs of various industries. Network customer service includes intelligent customer service and manual customer service. Intelligent customer service automatically replies based on pre-recorded corpus and common questions and answers, and involves some guidance functions. However, there are some problems (for example, complaints or complex problems) in actual life that need to be handled by manual customer service.

[0003] At present, the dispatching mode of manual customer service usually adopts a random allocation mode. Random allocation can easily lead to uneven workload of manual customer service and cannot bring the overall work efficiency. In addition, it may be randomly allocated to manual customer service who is not familiar with it, resulting in poor user experience. SUMMARY

[0004] The embodiments of the present application provide a case allocation method and device based on natural language processing, equipment and medium, which can allocate the to-be-allocated case to manual customer service, and can improve the accuracy and processing efficiency of case allocation.

[0005] In a first aspect, the embodiments of the present application provide a case allocation method based on natural language processing, wherein:

[0006] If the to-be-allocated case meets the trigger condition, the text data and the business type corresponding to the to-be-allocated case, and the target user information are obtained;

[0007] Based on the natural language processing model corresponding to the business type, the to-do list corresponding to the text data, and the business sub-type and processing level of the to-do list are obtained;

[0008] If the case record of the to-do list exists in the target user information, the customer service information of the historical customer service corresponding to the to-be-allocated case is searched from the pre-stored customer service information set;

[0009] Based on the customer service information of the historical customer service, the business sub-type and the processing level, the target customer service is searched from the customer service except the historical customer service;

[0010] The to-be-allocated case is sent to the target customer service based on the to-do list.

[0011] In a second aspect, the embodiments of the present application provide a case allocation device based on natural language processing, wherein:

[0012] The storage unit is used for storing customer service information;

[0013] a processing unit, configured to: if the to-be-assigned case meets a trigger condition, acquire text data and a business type corresponding to the to-be-assigned case, and target user information; acquire a to-do item corresponding to the text data, and a business sub-type and a processing level of the to-do item based on a natural language processing model corresponding to the business type; if there is a case record of the to-do item in the target user information, search for customer service information of a historical customer service corresponding to the to-be-assigned case from a pre-stored customer service information set; and search for a target customer service from customer services other than the historical customer service based on the customer service information of the historical customer service, the business sub-type, and the processing level.

[0014] a communication unit, configured to send the to-be-assigned case to the target customer service based on the to-do item.

[0015] In a third aspect, an embodiment of the present application provides a computer device, including a processor, a memory, and a communication interface, wherein the memory stores a computer program and is configured to be executed by the processor, and the computer program includes instructions for part or all of the steps described in the first aspect.

[0016] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program causes a computer to execute part or all of the steps described in the first aspect.

[0017] The embodiment of the present application has the following beneficial effects:

[0018] After the case assignment method, device, equipment and medium based on natural language processing are used, if the to-be-assigned case meets a trigger condition, the text data and the business type corresponding to the to-be-assigned case are acquired, and the target user information is acquired. Then, the to-do item corresponding to the text data, and the business sub-type and the processing level of the to-do item are acquired based on the natural language processing model corresponding to the business type. Then, it is determined whether there is a case record of the to-do item in the target user information. If yes, the to-do item is an unsolved item, and the customer service information of the historical customer service corresponding to the to-be-assigned case can be searched from the pre-stored customer service information set. Then, the target customer service is searched from the customer services other than the historical customer service based on the customer service information of the historical customer service, the business sub-type, and the processing level. In this way, the accuracy of case assignment of the unsolved item can be improved. After the to-be-assigned case is sent to the target customer service based on the to-do item, the target customer service can arrange work based on the to-do item, which is beneficial to improve the processing efficiency of the case. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to make the technical scheme of the present application or the prior art clearer, the accompanying drawings needed in the description of the embodiments or the prior art will be briefly introduced. Obviously, the accompanying drawings in the description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative effort based on these drawings.

[0020] Wherein:

[0021] Figure 1 A structural schematic diagram of a network architecture provided by the embodiment of the present application is shown in the figure.

[0022] Figure 2 A flowchart of a case allocation method based on natural language processing provided by the embodiment of the present application is shown in the figure.

[0023] Figure 3 A structural schematic diagram of a case allocation device based on natural language processing provided by the embodiment of the present application is shown in the figure.

[0024] Figure 4 A structural schematic diagram of a computer device provided by the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0025] In order to make the technical scheme of the present application or the prior art clearer, the accompanying drawings needed in the description of the embodiments or the prior art will be briefly introduced. Obviously, the accompanying drawings in the description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative effort based on these drawings.

[0026] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned accompanying drawings are used to distinguish different objects, and are not used to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units is not limited to the listed steps or units, but can optionally include steps or units not listed, or can optionally include other steps or units inherent to the process, method, product or device.

[0027] Reference to“an embodiment” herein means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase“in an embodiment” in various places in the specification are not necessarily referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is appreciated that embodiments described herein can be combined with other embodiments.

[0028] The network architecture to which the embodiments of the application are applied includes a server and an electronic device. The embodiments of the application do not limit the number of electronic devices and servers, and the server can simultaneously provide services for multiple electronic devices. The server can be a stand-alone server or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and basic cloud computing services such as big data and artificial intelligence platforms. The server can also be implemented by a server cluster composed of multiple servers.

[0029] The electronic device can be a personal computer (PC), a notebook computer, or a smart phone, and can also be an all-in-one machine, a palm computer, a pad, a smart television terminal, a vehicle-mounted terminal, or a portable device, etc. The operating system of the electronic device of the PC end, such as an all-in-one machine, can include but is not limited to a Linux system, a Unix system, a Windows series system (such as Windows xp, Windows 7, etc.), a Mac OS X system (the operating system of an Apple computer), etc. The operating system of the electronic device of the mobile end, such as a smart phone, can include but is not limited to an Android system, an IOS (the operating system of an Apple phone), a Window system, etc.

[0030] The electronic device can install and run an application program, and the server can be a server corresponding to the application program installed by the electronic device, providing application services for the application program. The application program can be a separately integrated application software, a small program embedded in other applications, or a system on a webpage, etc., which is not limited herein. In the embodiments of the application, the business types related to the application program can include government affairs, medical treatment, finance, and other business types related to customer services, which are not limited herein. The government affairs business type can include a financial business type, etc.

[0031] The electronic device can be a device used by a user (or a client) or a device used by a customer service. In the embodiments of the present application, the electronic device used by the user can be referred to as a user terminal, and the electronic device used by the customer service can be referred to as a customer service terminal. The server is configured to implement communication between the user terminal and the customer service terminal. The communication can be initiated by the user terminal to the server, and the server selects the customer service terminal to access. The communication can also be implemented by the server notifying the customer service terminal to the server specified user terminal, which is not limited herein.

[0032] As shown in Figure 1 , the user terminal 101 is a mobile phone, and the customer service terminal 103 is a notebook computer. When the user 111 clicks the customer service control on the display page of the user terminal 101, the user terminal 101 generates a customer service access request. The server 102 finds the artificial customer service 113 corresponding to the customer service access request, and then establishes a connection between the customer service terminal 103 corresponding to the artificial customer service 113 and the user terminal 101, thereby realizing the communication connection between the user 111 and the artificial customer service 113. The artificial customer service 113 can provide customer service for the user 111.

[0033] In the embodiments of the present application, the customer service case initiated by the user terminal and of the complaint type can be referred to as a complaint case, which is used to request the customer service to respond to the complaint or suggestion proposed by the user. The customer service case initiated by the user terminal and of the consultation type can be referred to as a consultation case, which is used to consult related problems.

[0034] The customer service case initiated by the customer service terminal can be understood as a case initiated by the server. That is, the case can be a case that meets a series of conditions detected by the server, and then is assigned to the artificial customer service or intelligent customer service corresponding to the customer service terminal. The intelligent customer service can be understood as a virtual character corresponding to a computer program in the server, which is used to perform customer service corresponding to a preset program.

[0035] The above-mentioned series of cases that meet the conditions can be time meeting, condition meeting, or both condition and time meeting, which is not limited herein. For example, when the application program detects an unfinished matter, the matter can be reported to the server, and the case corresponding to the matter is assigned to the artificial customer service by the server, and the reminder service is provided by the artificial customer service to complete or abandon the matter. For another example, when a preset time period is reached, a follow-up visit is made to each customer or a follow-up visit is made to the customer meeting a preset requirement.

[0036] In this application embodiment, the customer service case initiated by the application upon detecting an incomplete task is referred to as a breakpoint case. This task can be any stage of the application, including registration and usage. For example, if a user leaves their phone number in a promotional application for a product but does not submit a purchase intention, an incomplete task is detected, and customer service can proactively contact the user to determine if they are willing to purchase the product or register for the promotional application. Similarly, if a user enters an amount in a loan application but does not click the "Confirm" button, an incomplete task is detected, and customer service can proactively contact the user to determine if they are willing to borrow money, etc.

[0037] In this embodiment, the case types for customer service cases, in addition to the breakpoint types corresponding to breakpoint cases, customer complaint types corresponding to customer complaint cases, and consultation types corresponding to consultation cases described above, may also include promotion types corresponding to promotion cases, etc., which are not limited here. A promotion case can be a case in which customer service proactively calls or sends messages to users to promote an application or product. This promotion case can be executed at intervals, which can be a fixed value or a value generated based on actual conditions (e.g., the number of registered users, the frequency of user access to the application, the work efficiency of customer service, etc.), etc., which are not limited here.

[0038] In this embodiment, the server can store user information based on each user's user identifier. The user identifier can be an account registered or assigned by the user in the corresponding application on the server, or a mobile phone number or email address entered, etc., and is not limited here. The user information can include basic information such as the user's place of origin, region, age, occupation, gender, educational background, and work experience, and may also include the user's order records. For example, the user's historical customer service orders, recordings from historical orders, recorded feedback, complaints, etc., and the service quality score or satisfaction level provided by the user after the customer service session, etc., are also not limited here.

[0039] The server can pre-store customer service information for each human agent based on their customer service identifier. The customer service identifier can be an account registered or assigned to the human agent in the corresponding application on the server, or a mobile phone number or email address entered, etc., and is not limited here. This customer service information can include basic information such as the human agent's place of origin, region, age, occupation, gender, educational background, and work experience. This information may also include the human agent's functional information, such as their service quality rating, customer service level, work status, types of cases they are proficient in, number of cases currently being followed up, number of cases pending follow-up, and number of cases they can handle simultaneously.

[0040] The user information and the customer service information described above can be stored in a block created on a blockchain network. The blockchain is a new application mode of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, and encryption algorithm. The blockchain is essentially a decentralized database, and is a chain of data blocks associated using a cryptographic method. Each data block contains information of a batch of network transactions, and is used to verify the validity (anti-fake) of the information and generate the next block. The blockchain can include a blockchain underlying platform, a platform product service layer, and an application service layer. It can be seen that, by using the blockchain to store data in a distributed manner, the data security can be ensured, and data sharing between different platforms can be achieved.

[0041] In the embodiments of the present application, the working state of the artificial customer service includes an online state and an offline state. The online state means that the artificial customer service logs in an application program corresponding to a server, and the artificial customer service is processing a customer service order or waiting for processing a customer service order. The offline state is opposite to the online state, and means that the artificial customer service logs out of the application program, or the current state of the artificial customer service cannot process the customer service order. For example, after the artificial customer service logs in the application program of the server, if the artificial customer service is not at a work station, it can be determined that the working state of the artificial customer service is the offline state.

[0042] The method for determining the online state and the offline state is not limited in the present application. The detection information can be sent to the user terminal in the login state at a regular time, and if a response to the detection information is received, it is determined that the artificial customer service is in the online state. The detection information can be a preset static verification code, or a randomly generated dynamic verification code, which is not limited herein. Alternatively, the determination can be based on the processing state of the customer service terminal used by the artificial customer service. For example, based on the processing state of the customer service terminal, it can be determined that the artificial customer service is currently typing, or is in voice communication with a customer, and it is determined that the artificial customer service is in the online state.

[0043] The working state of the online state further includes an idle state and a busy state. The idle state means that the artificial customer service can currently receive a work assignment of a customer service case. The busy state means that the artificial customer service is working, for example, processing a customer service order, and performing operations such as sorting or uploading data after the customer service order is completed. The busy state can be determined according to the operation matters and / or operation frequency of the application account corresponding to the artificial customer service. The operation matters can be customer service business, or operations of the application account, for example, sending a notification, uploading data, etc.

[0044] In actual life, there are some problems (for example, complaints or complex problem, etc.) that need to be handled by artificial customer service. At present, the artificial customer service dispatching mode usually adopts random allocation mode. Random allocation is easy to lead to uneven workload of artificial customer service, and cannot play the overall work efficiency. And it may be randomly allocated to artificial customer service who is not familiar with, leading to poor user experience.

[0045] Based on this, the embodiment of the application provides a case allocation method based on natural language processing. The method can be executed by a case allocation device based on natural language processing. The device can be realized by software and / or hardware, and can be integrated in a server. The device can allocate a case to be allocated to artificial customer service, and can improve the accuracy and processing efficiency of case allocation.

[0046] Please refer to Figure 2 , Figure 2 is a flowchart of a case allocation method based on natural language processing provided by the application. The method is applied in a server as an example, which includes the following steps S201-S205:

[0047] S201: If the case to be allocated satisfies the trigger condition, the text data and business type corresponding to the case to be allocated, and the target user information are obtained.

[0048] The application does not limit the trigger condition. If the case to be allocated satisfies the trigger condition, it means that the case to be allocated should be allocated to artificial customer service. The trigger condition can be receiving a customer's artificial call request. For example, if the artificial service instruction information is 2, the customer inputs 2 or clicks the corresponding text of 2, it can be determined that the artificial call request is received.

[0049] The trigger condition can be detecting intelligent customer service failure. Intelligent customer service failure can include network problems such as intelligent customer service disconnection, and can also include situations where intelligent customer service cannot work normally, such as intelligent customer service closing external customer service function during maintenance.

[0050] The trigger condition can be intelligent customer service service failure, which can include that the question understood by the intelligent customer service is not the customer's question, or the answer of the intelligent customer service is not the answer the customer wants, and other invalid processing methods. It can be determined that the intelligent customer service service fails by receiving a reply of "not solved" to "whether the problem is solved" by the intelligent customer service, or by analyzing the received reply as "other" to the question option.

[0051] The trigger condition can be detecting occurrence of some preset types of cases. The preset types of cases can include breakpoint cases, complaint cases, consultation cases, and promotion cases, etc., which are not limited herein. Or some cases of users, for example, users with service quality scores lower than a threshold, or users who submit service requests for the same or similar problems in a short time, etc.

[0052] In the embodiments of the present application, the business type corresponding to the to-be-assigned case can include finance, medical treatment, finance, etc. Or the specific classification of the above information, for example, the finance type can include basic information library, budget compilation and review, budget execution, financial accounting, special fund, income, and comprehensive business, etc., which are not limited herein. The method for obtaining the business type of the to-be-assigned case is not limited in the present application, which can be determined based on the case type of the to-be-assigned case, or the application type of the application corresponding to the to-be-assigned case, or the product type corresponding to the browsing footprint in the application, etc.

[0053] In the embodiments of the present application, the text data corresponding to the to-be-assigned case includes information corresponding to the to-be-assigned case described in the form of text. It can be understood that if the to-be-assigned case is a telephone type case and the intelligent customer service has been processed before, the text data includes the text obtained by performing voice recognition on the voice file for communication between the intelligent customer service and the customer. If the to-be-assigned case is a text type case and the intelligent customer service has been processed before, the text data includes the chat record for communication between the intelligent customer and the customer. If the intelligent customer service has not been processed before, the text data can include the information carried by the customer when submitting the service request. For example, the question input by the customer in the customer service function component, the page browsed before submitting the service request, the browsing footprint of the customer before submitting the service request, etc.

[0054] In the embodiments of the present application, the target user information corresponding to the to-be-assigned case refers to the user information of the customer corresponding to the to-be-assigned case, which can be referred to the user information described above and will not be repeated here. The target user information can be all information of the customer corresponding to the to-be-assigned case, or can be user information related to the to-be-assigned case and basic information of the user. The basic information can include the region, age, occupation, gender, education background, work experience, etc. of the customer. The user information related to the to-be-assigned case can include order information of the to-be-assigned case, such as order number, time of submitting the to-be-assigned case, etc. The information related to the to-be-assigned case can also include order information of historical cases related to the to-be-assigned case. The historical cases related to the to-be-assigned case can be service cases with a time interval not more than a preset time length from the time of submitting the to-be-assigned case, or cases with the same case type as the to-be-assigned case, etc., which are not limited herein.

[0055] In a possible example, the obtaining the text data and the business type corresponding to the to-be-assigned case comprises: determining a case type of the to-be-assigned case based on the type of the trigger condition; obtaining the text data and the business type corresponding to the to-be-assigned case, and target user information based on the case type.

[0056] The case type can include the complaint type, the consultation type, the complaint type, and the promotion type as described above, or can include a telephone type or a text type, and the like, which are not limited herein. The method for determining the case type is not limited herein. It can be understood that, if the type of the trigger condition met by the to-be-assigned case is receiving a manual call request of a client, the case type can include the complaint type or the consultation type. Further, the case type of the to-be-assigned case can be determined based on information (such as text, etc.) carried in the manual call request. If the type of the trigger condition met by the to-be-assigned case is detecting a smart customer service failure, the case type can be determined based on information carried by the to-be-assigned case. If the type of the trigger condition met by the to-be-assigned case is determining that a smart customer service failure occurs, the case type is determined based on the service content of the smart customer service. If the type of the trigger condition met by the to-be-assigned case is detecting that some preset type of case occurs, the case type is determined based on the preset type, and the like.

[0057] The method for determining the text data and the business type and the target user information based on the case type is not limited herein. If the case type is a text type, the text data and the business type can be determined based on the content of the current page. If the case type is a telephone type, the telephone file can be subjected to semantic analysis, and then the text data and the business type are determined. If the case type is a complaint type, the complaint matter can be obtained by analyzing the complaint question first, and then the text data is searched based on the complaint matter, and the business type is determined based on the complaint matter, and the like. Then, the target user information corresponding to the business type can be selected from the pre-stored user information of the client.

[0058] It can be understood that, in this example, the case type of the to-be-assigned case is determined based on the type of the trigger condition. The text data and the business type corresponding to the to-be-assigned case, and the target user information are determined based on the case type. In this way, the accuracy of obtaining the text data and the business type, and the target user information can be improved, which is beneficial to improving the accuracy and efficiency of case assignment.

[0059] S202: obtaining a to-do item corresponding to the text data, and a business sub-type and a processing level of the to-do item based on a natural language processing model corresponding to the business type.

[0060] In the embodiments of the present application, the to-do item refers to the main information of the question raised by the customer, which can be composed of a short phrase or summary information composed of keywords. For example, the text data is "Now operate the budget review, the project refresh cannot be performed, and operation cannot be performed", and the to-do item can be "budget review cannot be operated". It can be understood that the customer may not be concise in language and the logic can be chaotic when asking questions. In the embodiments of the present application, the to-be-assigned case is sent to the artificial customer service based on the to-do item corresponding to the to-be-assigned case, so that the artificial customer service can quickly obtain the to-do item of the to-be-assigned case without identifying the text data corresponding to the to-be-assigned case again, and the processing efficiency and accuracy can be improved.

[0061] In the embodiments of the present application, the business sub-type can be understood as a sub-type of the business type. That is, the range of the business type should be greater than the business sub-type of the to-do item. Taking the business type as finance as an example, if the business type is the finance type, the business sub-type can include the basic information library, budget review, budget execution, financial accounting, special funds, income, and comprehensive business. If the business type is the basic information library, the business sub-type can include the sub-item project library and basic data. If the business type is the budget review, the business sub-type can include medium and long-term planning, budget review, and budget index. If the business type is the budget execution, the business sub-type can include the official card and the treasury payment. If the business type is the financial accounting, the business sub-type can include the total budget account, the government total account, the unit centralized finance, the department account, and the unit asset management. If the business type is the special fund, the business sub-type can include the capital construction project and the special fund. If the business type is the income, the business sub-type can include the non-tax collection, the transfer payment fund, and the government debt fund. If the business type is the comprehensive business, the business sub-type can include the government investment guide fund, the accounting personnel management, and the comprehensive performance management. It can be understood that the to-be-assigned case is assigned to the artificial customer service skilled in processing the to-be-assigned case, and the processing efficiency can be improved.

[0062] In the embodiments of the present application, the processing level is used to describe the classification of the processing urgency of the to-be-assigned case, which can be divided into special urgency, urgency, and general. Or it can be divided into level one, level two, and level three, and the higher the level number, the more urgent, or the lower the level number, the more urgent, etc., which is not limited here. It can be understood that the more urgent the processing level of the to-be-assigned case is, the faster the artificial customer service processing the to-be-assigned case is assigned. Therefore, the to-be-assigned case is assigned to the artificial customer service based on the processing level of the to-be-assigned case, and the processing efficiency can be improved.

[0063] In the embodiments of the present application, the natural language processing model can be used to obtain the to-do item and the business sub-type and processing level of the to-do item corresponding to the text data. It can be understood that the natural language processing model corresponds to the business type of the case to be allocated, which is beneficial to improve the accuracy of identifying the business sub-type and processing level of the to-do item and the to-do item.

[0064] The natural language processing model can be based on a natural language processing algorithm, and can use the jieba word segmentation tool or the word vector model of word2vec to obtain the target word in the text data, as well as the part of speech (such as the two categories of nouns and verbs, and the names of persons, places, and organizations, or auxiliary verbs, and nominal verbs) and the meaning of the target word, so as to determine the to-do item and the business sub-type and processing level of the to-do item corresponding to the text data.

[0065] The target word can be any word in the text data. It should be noted that the target word can include stop words. Stop words refer to certain words or characters that are automatically filtered out before or after processing natural language data (or text) in order to save storage space and improve search efficiency.

[0066] The present application does not limit the method of identifying stop words, and can identify based on a pre-stored stop word list. In the embodiments of the present application, the stop word list can include a plurality of stop words. Further, the stop words can be divided into strict stop words and loose stop words. The importance of strict stop words is relatively low compared to loose stop words, for example, mood words such as "ma" and "ne" indicating questions can be strict stop words, and words such as "de", "then", and "er" indicating a coherent relationship can be loose stop words.

[0067] It should be noted that the stop word list can be a stop word list corresponding to the business type. It can be understood that different types of business have different stop words. For example, in the financial business, it is necessary to interpret the official document, and the stop words can also include the format language of the region and the unit in the official document.

[0068] The present application does not limit the training process of the natural language processing model, and the text in the historical data can be labeled first, and then trained based on the labeled historical data. When the loss function of the training data is less than a preset threshold, or the number of training times exceeds a preset number, the training is determined to be completed. It should be noted that in the labeling process, the words related to the business type can increase their weight to improve the accuracy of identifying the to-do item and the business sub-type and processing level of the to-do item.

[0069] In one possible example, step S202 can include the following steps A1-A5:

[0070] A1: obtaining a target word in the text data based on a natural language processing model corresponding to the business type, and a part of speech and a meaning of the target word.

[0071] A2: selecting a key word from the target word based on the part of speech and the meaning of the target word.

[0072] In the embodiments of the present application, the target word can refer to the foregoing, which is not limited herein. The key word can be extracted based on a TextRank algorithm, which is not limited herein. In one possible example, step A2 can include: filtering a stop word in the target word based on a stop word table corresponding to the business type to obtain a reference word; calculating a key value of the reference word based on a part of speech and a meaning of the reference word and a preset weight of the reference word; and selecting a key word from the reference word based on the key value.

[0073] The reference word refers to a word in the target word other than the stop word. The preset weight can be determined based on the relevance of the reference word and the business type, or can be determined based on the position of the reference word in the text data, and the like, which is not limited herein.

[0074] The key value is used to describe the reference value of the reference word to the text data, which can be understood as the impact effect of omitting the reference word on the expression of the to-be-assigned case. The method for calculating the key value is not limited in the present application. It can be understood that the greater the preset weight is, the greater the key value can be for the words with similar part of speech and meaning. If the part of speech is the same and the meaning is different, the key value can be calculated based on the meaning and the preset weight, and the meaning of the adjacent word in the text data, and the like.

[0075] It can be understood that in this example, the stop words in the target word are first filtered based on the stop word table corresponding to the business type, so as to improve the accuracy of selecting the key word. Then the key value of the reference word is determined based on the part of speech and the meaning of the reference word and the preset weight, and the key word is selected, which can select the most relevant word as the key word, thereby improving the accuracy of selecting the key word.

[0076] A3: generating a to-do list corresponding to the text data based on the part of speech and the meaning of the key word.

[0077] It can be understood that the part of speech and the meaning of the key word can be used to determine the adjacent word in the to-do list, so as to form the to-do list based on the key word.

[0078] A4: obtaining a business sub-type of the to-do list based on the key word.

[0079] In one possible example, step A4 can include the following steps A41-A44:

[0080] A41: based on a first similarity value between the keyword and a reference word corresponding to the reference sub-type of the business type.

[0081] In the embodiments of the present application, the reference sub-type refers to a classification in the business type. The reference word refers to a word related to the reference sub-type, which can be determined based on a knowledge graph corresponding to the business type. The first similarity value is used to describe the similarity between the keyword and the reference word corresponding to the reference sub-type of the business type, which can be calculated based on the parts of speech and the meanings of the keyword and the reference word.

[0082] A42: based on a second similarity value between the text of the to-do item and the text of a reference item of the reference sub-type.

[0083] In the embodiments of the present application, the reference item refers to an item related to the reference sub-type. The second similarity value is used to describe the similarity between the text of the to-do item and the text of the reference item of the reference sub-type. In one possible example, step A42 can include the following steps: obtaining an edit distance and a similarity ratio between the text of the to-do item and the text of the reference item of the reference sub-type; and obtaining a second similarity value between the to-do item and the reference item based on the similarity ratio and the edit distance.

[0084] The edit distance refers to the number of single-character editing operations required to change from one string to another. Single-character operations include three operations: insertion, deletion, and replacement. If the number of operations is c, the calculation of the edit distance S1 can be as follows:

[0085] S1 = (max(length(doc1), length(doc2) - c) / max(length(doc1), length(doc2)))

[0086] Where length(doc1) is the text length of the to-do item, and length(doc2) is the text length of the reference item. Since the number of operations can be greater than the length of the text, the final S1 can be negative, and the maximum value between the value and 0 is taken, i.e., S1 = (max(0, S1).

[0087] The similarity ratio is used to describe the similarity between the text of the to-do item and the text of the reference item. It can be determined based on the ratio between the repeated words between the text of the to-do item and the text of the reference item, and the total number of words in the text. Or it can be determined based on the semantic similarity value between the text of the to-do item and the text of the reference item, etc., which is not limited herein.

[0088] It can be understood that in this example, the edit distance and the similarity ratio between the text of the to-do item and the text of the reference item are obtained first. Then, the second similarity value between the text of the to-do item and the text of the reference item is obtained based on the edit distance and the similarity ratio. The second similarity value can be a product or a weighted average of the similarity ratio and the edit distance, which is not limited herein. In this way, the similarity value is obtained by two methods of calculating the text similarity value, which can improve the accuracy of obtaining the business sub-type.

[0089] A43: Weighting the first similarity value and the second similarity value to obtain a target similarity value.

[0090] In the embodiments of the present application, the target similarity value can be a weighted average of the first similarity value and the second similarity value.

[0091] A44: Taking the reference sub-type corresponding to the maximum value in the target similarity value as the business sub-type of the to-do item.

[0092] It can be understood that in steps A41-A44, the target similarity value between the reference sub-type corresponding to the to-do item and the business type is determined from two aspects of the keyword and the to-do item. Then, the reference sub-type is selected as the business sub-type of the to-do item based on the maximum value in the target similarity value, which can improve the accuracy of obtaining the business sub-type.

[0093] A5: Analyzing the target sentence corresponding to the target word based on the part of speech and the word meaning of the target word to obtain the processing level of the to-do item.

[0094] In a possible example, the to-be-assigned case is a complaint type, and step A5 can include the following steps: performing emotion analysis on the target sentence corresponding to the target word based on the part of speech and the word meaning of the target word to obtain an emotion vector of the text data; determining the processing level of the to-do item based on the emotion vector.

[0095] The emotion vector includes probability values of various emotions. The method of emotion analysis is not limited in the present application, and can be obtained based on a pre-trained emotion recognition model. Alternatively, the emotion vector is determined based on the part of speech of the target word to determine the emotional tone of the target sentence, and then based on the word meaning of the target word.

[0096] It can be understood that the more pessimistic or angry the customer is in the complaint type of to-be-assigned case, the greater the opinion of the customer, and the to-do list processing level should be higher, so that the to-be-assigned case can be processed faster. In this example, the target sentence corresponding to the target word is analyzed based on the part of speech and the meaning of the target word to obtain the emotion vector of the target sentence, so as to obtain the emotion vector of the entire text data. Based on the emotion vector, the processing level of the to-do list is determined, which can improve the accuracy of determining the processing level.

[0097] In a possible example, the to-be-assigned case is of a non-complaint type, and step A5 can include the following steps: performing semantic analysis on the target sentence corresponding to the target word based on the part of speech and the meaning of the target word to obtain the urgency value of the to-do list; determining the processing requirement of the to-do list; and determining the processing level of the to-do list based on the urgency value and the processing requirement.

[0098] The urgency value is used to describe the degree of urgency of the to-do list. It can be determined based on time-related words in the target sentence, such as “quickly”, deadline, etc. It can be understood that the greater the urgency value, the greater the required processing rate. The processing requirement refers to the requirements of the to-do list itself on time, quality, etc. It can be understood that the more processing requirements, the higher the processing level.

[0099] In this example, the processing level of the to-do list is determined based on the urgency value and the processing requirement of the to-do list, which can improve the accuracy of determining the processing level and is beneficial to improve the accuracy and efficiency of case allocation.

[0100] S203: If the case record of the to-do list exists in the target user information, the customer service information of the historical customer service corresponding to the to-be-assigned case is searched from the pre-stored customer service information set.

[0101] In the embodiment of the present application, if the target user information has a case record of the to-do item, it indicates that the to-do item is an unsolved matter. It can be understood that if the to-do item is an unsolved matter, it can be assigned to a new artificial customer service to improve the processing efficiency and accuracy of the case. The historical customer service refers to an artificial customer service or intelligent customer service that previously served the customer for the to-do item. The customer service information can refer to the foregoing, and will not be described here. The customer service information of the historical customer service can include all information of the historical customer service, or can be the customer service information related to the to-be-assigned case and the basic information of the historical customer service. The customer service information related to the to-be-assigned case can include the order information of the to-be-assigned case, and can also include the order information of the historical case related to the to-be-assigned case. The historical case related to the to-be-assigned case can be a case with the same business type or business sub-type as the to-be-assigned case, and the like, which is not limited here. It can be understood that based on the customer service information of the historical customer service, the reason why the to-do item is not solved and the problems encountered in the processing process can be analyzed, which is beneficial to improve the accuracy of finding the target customer service.

[0102] S204: Based on the customer service information of the historical customer service, the business sub-type and the processing level, find a target customer service from the customer services except the historical customer service.

[0103] In the embodiment of the present application, the target customer service refers to an artificial customer service that processes the to-be-assigned case. The present application does not limit the method for determining the target customer service. If the historical customer service is an intelligent customer service, the target customer service can be selected from the artificial customer service. If the historical customer service is an artificial customer service, the target customer service can be found from the customer services except the historical customer service, and the like.

[0104] In a possible example, step S204 can include the following steps B1-B5, wherein:

[0105] B1: Determine the unsolved reason of the to-do item and the score value of the historical customer service based on the customer service information of the historical customer service.

[0106] In the embodiment of the present application, the unsolved reason can include that the historical customer service does not understand the intention of the customer corresponding to the to-be-assigned case, or can include that the historical customer service has insufficient level or permission to process the to-be-assigned case, and the like, which is not limited here. The score value is used to describe the service quality of the historical customer service.

[0107] In a possible example, step B1 can include the following steps B11-B13, wherein:

[0108] B11: Determine the historical text data of the to-do item and the score information of the historical customer service based on the customer service information of the historical customer service.

[0109] In the embodiment of the present application, the historical text data refers to relevant text data in the process of historical customer service providing customer service, which can include historical communication records or summary information corresponding to the historical communication records, etc. The score information includes evaluation information and score value of the customer received after the historical customer service providing customer service.

[0110] B12: analyzing the historical text data to obtain the unresolved reason of the to-do item.

[0111] The method of analyzing the unresolved reason is not limited in the present application. If there are different similar words in the historical text data, it indicates that the customer repeatedly asks the same question, and the unresolved reason can include that the historical customer does not understand the intention of the customer corresponding to the to-be-assigned case. If the historical text data includes the "other" option of the customer's reply, and the number of filling of the "other" option or other content is large, it indicates that the customer does not get a satisfactory option, and the unresolved reason can include that the historical customer does not understand the intention of the customer corresponding to the to-be-assigned case. If the historical text data includes the instruction submitted by the historical customer to the upper level, it indicates that the level or authority of the current historical customer processing the to-be-assigned case is insufficient.

[0112] B13: calculating the score value of the historical customer based on the score information.

[0113] The method of calculating the score information is not limited in the present application, which can analyze the comment information, or can weight the evaluation value to obtain the score value.

[0114] It can be understood that in steps B11-B13, the historical text data of the to-do item and the score information of the historical customer are first determined based on the customer service information of the historical customer. Then the historical text data is analyzed to obtain the unresolved reason of the to-do item, and the score value of the historical customer is calculated based on the score information. In this way, the accuracy of the unresolved reason and the score value can be improved.

[0115] B2: finding a to-be-assigned customer whose score value is higher than that of the historical customer from the customers other than the historical customer based on the customer information set.

[0116] In the embodiment of the present application, the customer information of the artificial customer service is stored in the form of a customer information set, which can be referred to as the foregoing, and will not be described here. The online customer service refers to the to-be-assigned customer whose working state is online. The to-be-assigned customer refers to the customer whose score value is higher than that of the historical customer among the customers other than the historical customer. The calculation method of the score value of the customer can be referred to as the foregoing, and will not be described here. The target customer service refers to the online customer service which is good at processing the business sub-type and can cope with the processing level among the artificial customer service.

[0117] B3: determining whether the to-be-assigned customer includes an online customer service.

[0118] If yes, step B4 is performed: selecting a target service from the online service based on the service information of the online service. Otherwise, step B5 is performed: selecting a target service from the to-be-allocated service based on the service information of the to-be-allocated service.

[0119] The application does not limit the method for selecting a target service from the online service and the to-be-allocated service, and the following is an example of selecting a target service from the online service. The first evaluation value of the to-be-allocated case of the service sub-type and the second evaluation value of the to-be-allocated case of the processing level can be determined based on the service information of the online service. Then, the target evaluation value is calculated based on the first evaluation value and the second evaluation value, and the target service is selected from the online service based on the target evaluation value. Alternatively, the first service is selected from the online service based on the first evaluation value. If the number of the first service is greater than 1, the first service with the processing level higher than the processing level of the to-be-allocated case is determined as the target service. It can be understood that the target service corresponding to the service sub-type and the processing level is determined based on the service information of the online service, which can improve the efficiency and accuracy of determining the allocation of the to-be-processed case to the target service.

[0120] In a possible example, if the number of the determined target service is greater than 1, the processing efficiency of the target service is determined, and the target service with the fastest processing efficiency is selected as the target service.

[0121] In the embodiment of the application, the processing efficiency is used to describe the efficiency of the target service in processing the to-be-allocated case. The application does not limit the method for determining the processing efficiency, which can be determined based on the case type skilled in the art, the number of currently followed cases, the number of to-be-followed cases, the number of cases that can be processed synchronously, and the like.

[0122] It can be understood that if the number of the target service is greater than 1, it means that the number of the services corresponding to the current service sub-type and the processing level is more than one. One target service can be selected from the target service based on the processing efficiency of the target service in processing the case to perform the to-be-allocated case. If the number of the target service is equal to 1, the target service can perform the to-be-allocated case. In this way, the accuracy of selecting the target service can be improved.

[0123] In a possible example, if the number of the determined target service is greater than 1, the matching value of the target service can also be determined based on the target user information and the service information of the target service, and the target service is selected from the target service based on the matching value and the processing efficiency.

[0124] The target user information is as described above and will not be repeated here. It can be understood that if the number of target customer service representatives is greater than one, the evaluation score of the target customer service representative in handling the current pending case can be determined based on the target user information and the customer service information of the target customer service representative, thus obtaining a matching value. Then, based on the matching value and processing efficiency, a target customer service representative is selected to execute the pending case. This improves the accuracy of selecting target customer service representatives.

[0125] In one possible example, after step S202, if there is no case record for the pending matter in the target user information, the target customer service corresponding to the business subtype and the processing level is searched from the pre-stored customer service information set.

[0126] The method for selecting target customer service representatives can be referenced above. It can be understood that if there are no pending case records for the pending items in the target user information, it indicates that the pending item is a new unresolved matter, and the target customer service representative corresponding to the business subtype and processing level can be found from the human customer service database. This helps improve the efficiency and accuracy of the target customer service representatives in handling assigned cases.

[0127] S205: Send the pending cases to the target customer service representative based on the pending items.

[0128] This application does not limit the method of sending pending cases to the target customer service representative; it can be based on SMS, email, or system messages, etc. It should be noted that by sending pending cases to the target customer service representative based on the to-do list, the representative can obtain the to-do list, as well as the business subtype and processing level of the to-do list, so that the representative can arrange work accordingly, thereby improving processing efficiency.

[0129] exist Figure 2 In the method described, if a case to be assigned meets the triggering conditions, the text data and business type corresponding to the case, as well as the target user information, are obtained. Then, based on the natural language processing model corresponding to the business type, the pending tasks corresponding to the text data, along with the business subtype and processing level of the pending tasks, are obtained. Next, it is determined whether a case record for the pending task exists in the target user information. If so, it indicates that the pending task is unresolved, and the customer service information of the historical customer service representative corresponding to the case to be assigned can be retrieved from the pre-stored customer service information set. Then, based on the historical customer service information, business subtype, and processing level, the target customer service representative is retrieved from customer service representatives other than the historical ones. This improves the accuracy of case assignment for unresolved tasks. After sending the case to be assigned to the target customer service representative based on the pending task, the target customer service representative can arrange work based on the pending task, which helps improve the efficiency of case processing.

[0130] The methods of the embodiments of this application have been described in detail above, and the apparatus of the embodiments of this application is provided below.

[0131] Consistent with the embodiments shown in Figure 2 , please refer to Figure 3 , Figure 3 is a structure diagram of a case allocation device based on natural language processing. As shown in Figure 3 , the case allocation device 300 comprises:

[0132] a storage unit 303 for storing a customer service information set;

[0133] The processing unit 301 is configured to, if a to-be-allocated case satisfies a trigger condition, acquire text data corresponding to the to-be-allocated case and a business type, and target user information; acquire a to-do item corresponding to the text data based on a natural language processing model corresponding to the business type, and a business sub-type and a processing level of the to-do item; if there is a case record of the to-do item in the target user information, search for customer service information of a historical customer service corresponding to the to-be-allocated case from the customer service information set; and search for a target customer service from customer services other than the historical customer service based on the customer service information of the historical customer service, the business sub-type, and the processing level.

[0134] The communication unit 302 is configured to send the to-be-allocated case to the target customer service based on the to-do item.

[0135] In a possible example, the processing unit 301 is specifically configured to acquire a target word in the text data based on a natural language processing model corresponding to the business type, and a part of speech and a word meaning of the target word; select a keyword from the target word based on the part of speech and the word meaning of the target word; generate a to-do item corresponding to the text data based on the part of speech and the word meaning of the keyword; acquire a business sub-type of the to-do item based on the keyword; and analyze a sentence corresponding to the target word based on the part of speech and the word meaning of the target word to obtain a processing level of the to-do item.

[0136] In a possible example, the processing unit 301 is specifically configured to acquire a first similarity value between the keyword and a reference word of a reference sub-type corresponding to the business type; acquire a second similarity value between a text of the to-do item and a text of a reference item of the reference sub-type; weight the first similarity value and the second similarity value to obtain a target similarity value; and select, as the business sub-type of the to-do item, a reference sub-type in the reference sub-type corresponding to a maximum value in the target similarity value.

[0137] In a possible example, the processing unit 301 is specifically configured to obtain an edit distance and a similarity ratio between the text of the to-do item and the text of the reference item of the reference sub-type; and obtain a second similarity value between the to-do item and the reference item based on the similarity ratio and the edit distance.

[0138] In a possible example, the to-be-assigned case is of a complaint type, and the processing unit 301 is specifically configured to perform sentiment analysis on a target sentence corresponding to the target word based on the part-of-speech and the word meaning of the target word, to obtain an emotion vector of the text data; and determine a processing level of the to-do item based on the emotion vector.

[0139] In a possible example, the processing unit 301 is specifically configured to determine an unresolved reason of the to-do item and a score value of the historical customer service based on the customer service information of the historical customer service; and based on the set of customer service information, find a to-be-assigned customer service whose score value is higher than the score value of the historical customer service from among customer services other than the historical customer service; if the to-be-assigned customer service includes an online customer service, select a target customer service corresponding to the business sub-type and the processing level from the online customer service based on the customer service information of the online customer service; or if the to-be-assigned customer service does not include an online customer service, select a target customer service corresponding to the business sub-type and the processing level from the to-be-assigned customer service based on the customer service information of the to-be-assigned customer service.

[0140] In a possible example, the processing unit 301 is specifically configured to determine historical text data of the to-do item and score information of the historical customer service based on the customer service information of the historical customer service; analyze the historical text data to obtain an unresolved reason of the to-do item; and calculate a score value of the historical customer service based on the score information.

[0141] The detailed processes performed by each unit in the case assignment apparatus 300 can be referred to the execution steps in the foregoing method embodiments, which will not be described here.

[0142] Consistent with the embodiments of Figure 2 , refer to Figure 4 , Figure 4 is a structural schematic diagram of a computer device. As shown in Figure 4 , the computer device 400 includes a processor 410, a memory 420, and a communication interface 430. The processor 410, the memory 420, and the communication interface 430 are connected to each other through a bus 450. Figure 3 The communication unit 302 shown in the figure can implement the related functions through the communication interface 430, Figure 3 The processing unit 301 shown in the figure can implement the related functions through the processor 410, Figure 3The functions implemented by the storage unit 303 shown can be implemented by the processor 410.

[0143] The memory 420 stores a computer program 440 and is configured to be executed by the processor 410, including instructions for performing the following steps:

[0144] If the to-be-assigned case meets the triggering condition, text data and a business type corresponding to the to-be-assigned case are obtained, and target user information is obtained;

[0145] Based on a natural language processing model corresponding to the business type, a to-do item corresponding to the text data, a business sub-type of the to-do item, and a processing level of the to-do item are obtained;

[0146] If there is a case record of the to-do item in the target user information, the customer service information of the historical customer service corresponding to the to-be-assigned case is searched from a pre-stored customer service information set;

[0147] Based on the customer service information of the historical customer service, the business sub-type, and the processing level, a target customer service is searched from customer services other than the historical customer service;

[0148] Based on the to-do item, the to-be-assigned case is sent to the target customer service.

[0149] In one possible example, in the aspect of obtaining, based on a natural language processing model corresponding to the business type, a to-do item corresponding to the text data, a business sub-type of the to-do item, and a processing level of the to-do item, the computer program 440 includes instructions specifically for performing the following steps:

[0150] Based on a natural language processing model corresponding to the business type, a target word in the text data, and a part of speech and a word meaning of the target word are obtained;

[0151] Based on the part of speech and the word meaning of the target word, a keyword is selected from the target word;

[0152] Based on the part of speech and the word meaning of the keyword, a to-do item corresponding to the text data is generated;

[0153] Based on the keyword, a business sub-type of the to-do item is obtained;

[0154] Based on the part of speech and the word meaning of the target word, a statement corresponding to the target word is analyzed to obtain a processing level of the to-do item.

[0155] In one possible example, in the aspect of obtaining, based on the keyword, a business sub-type of the to-do item, the computer program 440 includes instructions specifically for performing the following steps:

[0156] a first similarity value between the keyword and a reference keyword corresponding to the reference sub-type of the business type;

[0157] a second similarity value between the text of the to-do item and a reference text of a reference to-do item of the reference sub-type;

[0158] weighting the first similarity value and the second similarity value to obtain a target similarity value;

[0159] taking, as the business sub-type of the to-do item, a reference sub-type in the reference sub-types corresponding to a maximum value in the target similarity value.

[0160] In one possible example, in terms of the second similarity value between the text of the to-do item and the text of the reference to-do item of the reference sub-type, the computer program 440 comprises instructions specifically used to perform the following steps:

[0161] obtaining an edit distance and a similarity ratio between the text of the to-do item and the text of the reference to-do item of the reference sub-type;

[0162] obtaining the second similarity value between the to-do item and the reference to-do item based on the similarity ratio and the edit distance.

[0163] In one possible example, the case to be allocated is of a complaint type, and in terms of the analysis of the target sentence corresponding to the target keyword based on the part-of-speech and the word meaning of the target keyword to obtain the processing level of the to-do item, the computer program 440 comprises instructions specifically used to perform the following steps:

[0164] performing sentiment analysis on the target sentence corresponding to the target keyword based on the part-of-speech and the word meaning of the target keyword to obtain an emotion vector of the text data;

[0165] determining the processing level of the to-do item based on the emotion vector.

[0166] In one possible example, in terms of finding a target customer service from customer services other than the historical customer service based on the customer service information of the historical customer service, the business sub-type, and the processing level, the computer program 440 comprises instructions specifically used to perform the following steps:

[0167] determining an unresolved reason of the to-do item and a score value of the historical customer service based on the customer service information of the historical customer service;

[0168] finding, from the customer services other than the historical customer service, a to-be-allocated customer service with a score value higher than the score value of the historical customer service based on the set of customer service information;

[0169] If the to-be-assigned customer service includes an online customer service, then based on the customer service information of the online customer service, a target customer service corresponding to the business sub-type and the processing level is selected from the online customer service; or

[0170] If the to-be-assigned customer service does not include an online customer service, then based on the customer service information of the to-be-assigned customer service, a target customer service corresponding to the business sub-type and the processing level is selected from the to-be-assigned customer service.

[0171] In one possible example, in the aspect of determining the unresolved reason of the to-do item and the score value of the historical customer service based on the customer service information of the historical customer service, the computer program 440 includes instructions specifically used for performing the following steps:

[0172] determining historical text data of the to-do item and score information of the historical customer service based on the customer service information of the historical customer service;

[0173] analyzing the historical text data to obtain the unresolved reason of the to-do item;

[0174] calculating the score value of the historical customer service based on the score information.

[0175] It should be noted that the number of devices in the above computer device can be one or more, which is not limited herein.

[0176] The embodiment of the present application further provides a computer storage medium, wherein the computer storage medium stores a computer program, the computer program causes a computer to execute to realize part or all steps of any method described in the method embodiment, and the computer includes an electronic device and a server.

[0177] The embodiment of the present application further provides a computer program product, the computer program product includes a non-transitory computer readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute to realize part or all steps of any method described in the method embodiment. The computer program product can be a software installation package, and the computer includes an electronic device and a server.

[0178] It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all expressed as a combination of a series of actions, but those skilled in the art should know that the present application is not limited to the action order described, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modes involved are not necessarily necessary for the present application.

[0179] In the above embodiments, the description of each embodiment focuses on different aspects, and the parts not described in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0180] In several embodiments provided in the present application, it should be understood that the disclosed apparatus can be implemented by other manners. For example, the apparatus embodiments described above are only illustrative, for example, the division of units is only a logical function division, and in actual implementation, another division manner can be used, for example, at least one unit or component can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or components shown or discussed can be indirect coupling or communication connection through some interfaces, apparatuses or units, and can be electrical or other forms.

[0181] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, can be located in one place, or can be distributed to at least one network unit. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0182] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of software program mode.

[0183] If the integrated unit is realized in the form of software program mode and sold or used as an independent product, it can be stored in a computer readable storage medium. According to such understanding, the technical scheme of the present application or the essential part or all or part of the technical scheme that contributes to the prior art can be embodied in the form of software product, which is stored in a storage medium and includes a plurality of instructions for making a computer (which can be a personal computer, a server or a network device, etc.) execute all or part of the steps of the method of each embodiment of the present application. The storage medium mentioned above includes: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, magnetic disk or optical disk and various program code storage media.

[0184] The above has carried out the detailed introduction to the embodiment of the application, the principle and implementation mode of the application are described by applying specific examples in this paper, the above embodiment is only used to help understanding the method of the application and its core idea; at the same time, for the general technical personnel in the art, according to the idea of the application, the specific implementation mode and application range will have changes, and the above, the content of the specification should not be understood as the limitation of the application.

Claims

1. A case assignment method based on natural language processing, characterized by, The method comprises the steps of: if the to-be-assigned case meets the triggering condition, obtaining text data corresponding to the to-be-assigned case and a business type, and target user information, wherein the text data comprises chat records of communication between an intelligent customer and a customer; based on a natural language processing model corresponding to the business type, obtaining a to-do item corresponding to the text data, and a business sub-type and a processing level of the to-do item, wherein the to-do item is main information of a question raised by the customer, and the to-do item comprises a phrase or summary information composed of keywords, and the obtaining of the to-do item corresponding to the text data and the business sub-type and the processing level of the to-do item based on the natural language processing model corresponding to the business type comprises: based on the natural language processing model corresponding to the business type, obtaining a target word in the text data and a part of speech and a meaning of the target word; selecting a keyword from the target word based on the part of speech and the meaning of the target word; generating the to-do item corresponding to the text data based on the part of speech and the meaning of the keyword; obtaining a target similarity value by weighting a first similarity value between the keyword and a reference word of a reference sub-type corresponding to the business type, and a second similarity value between a text of the to-do item and a text of a reference item of the reference sub-type; taking a reference sub-type corresponding to a maximum value in the target similarity value as the business sub-type of the to-do item; and analyzing a sentence corresponding to the target word based on the part of speech and the meaning of the target word to obtain the processing level of the to-do item; if there is a case record of the to-do item in the target user information, searching for customer service information of a historical customer service corresponding to the to-be-assigned case from a pre-stored customer service information set; based on the customer service information of the historical customer service, the business sub-type and the processing level, searching for a target customer service from customer services other than the historical customer service; based on the to-do item, sending the to-be-assigned case to the target customer service.

2. The method of claim 1, wherein, The second similarity value between the text of the to-do item and the text of the reference item of the reference sub-type comprises: obtaining an edit distance and a similarity ratio between the text of the to-do item and the text of the reference item of the reference sub-type; based on the similarity ratio and the edit distance, obtaining the second similarity value between the to-do item and the reference item.

3. The method of claim 1, wherein, The to-be-assigned case is of a complaint type, and the processing level of the to-do item is obtained by analyzing a target sentence corresponding to the target word based on the part of speech and the meaning of the target word, which comprises: performing emotion analysis on the target sentence corresponding to the target word based on the part of speech and the meaning of the target word to obtain an emotion vector of the text data; determining the processing level of the to-do item based on the emotion vector.

4. The method according to any one of claims 1 to 3, characterized in that, The target customer service is searched from the customer services other than the historical customer service based on the customer service information of the historical customer service, the business sub-type and the processing level, which comprises: determine an unresolved reason of the to-do item and a score value of the historical customer service based on the customer service information of the historical customer service; find a to-be-assigned customer service with a score value higher than the score value of the historical customer service from customer services other than the historical customer service based on the set of customer service information; if the to-be-assigned customer service includes an online customer service, select a target customer service corresponding to the business sub-type and the processing level from the online customer service based on the customer service information of the online customer service; or if the to-be-assigned customer service does not include an online customer service, select a target customer service corresponding to the business sub-type and the processing level from the to-be-assigned customer service based on the customer service information of the to-be-assigned customer service.

5. The method of claim 4, wherein, The determining an unresolved reason of the to-do item and a score value of the historical customer service based on the customer service information of the historical customer service comprises: determining historical text data of the to-do item and score information of the historical customer service based on the customer service information of the historical customer service; analyzing the historical text data to obtain the unresolved reason of the to-do item; calculating the score value of the historical customer service based on the score information.

6. A natural language processing based case allocation apparatus for performing the method of any one of claims 1-5, characterized in that, comprise: a storage unit configured to store a set of customer service information; a processing unit configured to, if a to-be-assigned case meets a trigger condition, obtain text data and a business type corresponding to the to-be-assigned case, and target user information; obtain a to-do item corresponding to the text data and a business sub-type and a processing level of the to-do item based on a natural language processing model corresponding to the business type; if there is a case record of the to-do item in the target user information, find customer service information of a historical customer service corresponding to the to-be-assigned case from a pre-stored set of customer service information; find a target customer service from customer services other than the historical customer service based on the customer service information of the historical customer service, the business sub-type and the processing level; a communication unit configured to send the to-be-assigned case to the target customer service based on the to-do item.

7. A computer device, comprising: comprise a processor, a memory and a communication interface, wherein the memory stores a computer program configured to be executed by the processor, and the computer program comprises instructions for executing steps in the method of any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program causes a computer to execute to implement the method of any one of claims 1-5. The computer readable storage medium stores a computer program, and the computer program causes a computer to execute to implement the method of any one of claims 1-5.

Images