Work order generation method, device, equipment and storage medium based on email information

Abstract

The present application relates to a method, apparatus, device, and storage medium for generating a work order based on email information. The method comprises: parsing an email to be processed, obtaining the body information and sender information of the email to be processed, obtaining voice description information input by a customer service user for the body information, and converting the voice description information into target text information, wherein the target text information is semantically associated with the body information, extracting the semantic text of the email to be processed based on the body information and the target text information, extracting semantic keywords corresponding to the semantic text and the problem type represented by the semantic text, determining a work order template based on the sender information and the problem type, and filling the semantic keywords and semantic text into the work order template to generate a target work order. The present application can accurately generate a work order by combining the body information and target text information of the email.

Description

Technical Field

[0001] The present application relates to the technical field of work order generation, and in particular to a work order generation method, apparatus, device and storage medium based on email information. Background Art

[0002] Currently, creating work orders is typically done manually by customer service personnel. This consumes a significant amount of customer service manpower when the volume of work orders is high, and work order creation is inefficient. While there are existing solutions for automatically generating work orders, most of these solutions are based on preset templates and fixed rules. This lacks flexibility, and because the rules don't cover all scenarios, the accuracy of automatically generated work orders is low.

[0003] Therefore, how to improve the accuracy of work order generation has become a technical problem that needs to be solved urgently by those skilled in the art. It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] In view of the above, the present application provides a method, apparatus, device and storage medium for generating a work order based on email information, the purpose of which is to solve the above technical problems.

[0005] In a first aspect, the present application provides a method for generating a work order based on email information, the method comprising:

[0006] Parsing the email to be processed to obtain the body information and sender information of the email to be processed;

[0007] Acquiring voice description information input by a customer service user for the main text information, and converting the voice description information into target text information, wherein the target text information is semantically associated with the main text information;

[0008] Extracting the semantic text of the email to be processed based on the body text information and the target text information;

[0009] Extracting semantic keywords corresponding to the semantic text and the question type represented by the semantic text;

[0010] A work order template is determined based on the sender information and the problem type, and the semantic keywords and the semantic text are filled into the work order template to generate a target work order.

[0011] In a second aspect, the present application provides a work order generation device based on email information, the device comprising:

[0012] Parsing module: used to parse the email to be processed and obtain the body information and sender information of the email to be processed;

[0013] A conversion module is configured to obtain voice description information input by a customer service user for the main text information and convert the voice description information into target text information, wherein the target text information is semantically associated with the main text information;

[0014] A first extraction module: configured to extract the semantic text of the email to be processed based on the body information and the target text information;

[0015] A second extraction module: used to extract semantic keywords corresponding to the semantic text and the question type represented by the semantic text;

[0016] A generation module is used to determine a work order template based on the sender information and the problem type, and fill the semantic keywords and the semantic text into the work order template to generate a target work order.

[0017] In a third aspect, the present application provides an electronic device, comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus;

[0018] Memory for storing computer programs;

[0019] The processor is used to implement the work order generation method based on email information as described in any embodiment of the first aspect when executing the program stored in the memory.

[0020] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for generating a work order based on email information as described in any embodiment of the first aspect is implemented.

[0021] The above technical solution provided by the embodiment of the present application has the following advantages compared with the prior art:

[0022] This application parses the email to be processed to obtain the body information and sender information of the email to be processed, obtains the voice description information input by the customer service user for the body information, and converts the voice description information into target text information. Since the target text information is associated with the semantics of the body information, that is, the target text information is based on the customer service staff's understanding of the semantics of the body information of the email, when the body information of the email contains abbreviated products, numbers instead of products, scattered logic, etc., it avoids the situation where the semantics of the body information of the email are recognized and the deviation of the recognition results is large. Then, the semantic text of the email to be processed is extracted based on the body information and the target text information, and the semantic keywords corresponding to the semantic text and the problem type represented by the semantic text are extracted. Since the semantic text of the email to be processed combines the characteristics of the body information and the target text information, the work order template is determined according to the sender information and the problem type. Filling the semantic keywords and semantic text into the work order template can accurately generate the target work order, without the need for customer service personnel to manually create work orders, thereby improving the efficiency of work order generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0024] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0025] Figure 1 This is a flowchart of an embodiment of a method for generating a work order based on email information of this application;

[0026] Figure 2 This is a module diagram of an embodiment of a work order generation device based on email information of this application;

[0027] Figure 3 A schematic diagram of an electronic device embodiment of the present application;

[0028] The realization of the objectives, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0029] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0030] The disclosure below provides many different embodiments or examples for implementing different structures of the present application. In order to simplify the disclosure of the present application, the components and settings of specific examples are described below. Of course, these are merely examples and are not intended to limit the present application. In addition, the present application may repeat reference numbers and / or letters in different examples. Such repetition is for the purpose of simplicity and clarity and does not in itself indicate the relationship between the various embodiments and / or settings discussed.

[0031] This application provides a method for generating a work order based on email information. Figure 1 The figure shows a method flow diagram of an embodiment of the method for generating a work order based on email information of the present application. The method can be executed by an electronic device, which is implemented by software and / or hardware. The method for generating a work order based on email information includes:

[0032] Step S10: Parse the email to be processed to obtain the body information and sender information of the email to be processed;

[0033] Step S20: obtaining voice description information input by the customer service user for the main text information, and converting the voice description information into target text information, wherein the target text information is semantically associated with the main text information;

[0034] Step S30: extracting the semantic text of the email to be processed based on the body text information and the target text information;

[0035] Step S40: extracting semantic keywords corresponding to the semantic text and the question type represented by the semantic text;

[0036] Step S50: Determine a work order template based on the sender information and the problem type, and fill the semantic keywords and the semantic text into the work order template to generate a target work order.

[0037] In this embodiment, the emails to be processed may refer to emails received by the customer service system. Taking the customer service system as an example, the customer service system may receive emails from users such as the upstream suppliers and downstream channel merchants of the enterprise. After the customer service personnel read the emails of the customer service system, they usually need to create a corresponding work order and send the work order to the relevant handler.

[0038] Since in actual applications, a large proportion of emails in the customer service system need to generate work orders, manually creating work orders by customer service personnel requires a lot of customer service manpower and has low work order generation efficiency. Therefore, this application provides a method for automatically generating work orders based on email information.

[0039] Specifically, after receiving emails from the customer's system, the work order generation system parses the pending emails using an email protocol to obtain the email's body and sender information. For example, POP3 (Post Office Protocol) and IMAP (Internet Message Access Protocol) can retrieve email data from email servers. These protocols can be used to parse the email's body and sender information, and regular expressions can also be used to parse emails.

[0040] Since the user's description of the email body may contain abbreviations of products, or numbers instead of products, or the logic of the body may be disorganized, in order to accurately generate a work order based on the body of the email, the customer service staff can input the description information corresponding to the body of the email based on the body of the email. In order for the customer service staff to quickly input the description information of the body of the email, the customer service staff can input the voice description information corresponding to the body of the email into the work order generation system through voice input. The work order generation system can convert the voice description information into text information (recorded as target text information), wherein the target text information is associated with the semantics of the body of the email to be processed, that is, the target text information is the text with the same semantics as the body of the email input by the customer service staff based on their understanding of the semantics of the body of the email. This avoids the situation where the work order generation system has a large deviation in the recognition result of the semantics of the body of the email due to the presence of abbreviations of products, numbers instead of products, or disorganized logic in the body of the email.

[0041] After obtaining the target text information, based on the body information and the target text information, the semantic text of the email to be processed can be extracted using a historical work order text set or a pre-built semantic recognition model. Specifically, the step of "extracting the semantic text of the email to be processed based on the body information and the target text information" includes:

[0042] Matching the main text information and the target text information with the historical work order text set to obtain a matching score;

[0043] If the matching score is greater than a preset value, the historical work order text corresponding to the matching score is used as the semantic text of the email to be processed;

[0044] If the matching score is less than or equal to a preset value, the body text information and the target text information are input into a pre-built semantic recognition model to obtain the semantic text of the email to be processed.

[0045] The historical ticket text set refers to a collection of historical tickets, including those manually created by customer service personnel and those automatically generated by the ticket generation system. The body text of the pending email and the corresponding target text are matched against the historical ticket text set to obtain a matching score. This matching method can be performed by converting the text into vectors and calculating the similarity between the texts as the matching score. A higher matching score indicates a higher degree of similarity.

[0046] If the matching score is greater than the preset value, the historical work order text corresponding to the matching score will be used as the semantic text of the email to be processed, where the preset value can be 97%. In actual applications, the preset value can be adjusted. If the matching score is less than or equal to the preset value, it means that no suitable semantic text has been matched in the historical work order text set. At this time, the body information and target text information are input into the pre-built semantic recognition model to obtain the semantic text of the email to be processed. For example, the semantic recognition model can be trained based on a neural network model. Since the customer service system may receive multiple email feedbacks on the same product issue, the body information and target text information are matched with the historical work order text set respectively to quickly obtain the semantic text of this type of email.

[0047] Furthermore, the above step of "respectively matching the main text information and the target text information with the historical work order text set to obtain a matching score" includes:

[0048] Matching the text vectors corresponding to the main text information and the target text information with the text vectors corresponding to the historical work order text set to obtain a first score and a second score;

[0049] The average of the first score and the second score is used as the matching score.

[0050] The score obtained by matching the body text with the text vectors corresponding to the historical ticket text set is recorded as the first score, the score obtained by matching the target text with the text vectors corresponding to the historical ticket text set is recorded as the second score, and the average of the first and second scores is used as the matching score. Alternatively, weights can be assigned to the first and second scores, and the weighted sum of the first and second scores can be used to obtain the matching score. By adjusting the weights, the matched semantic text can be more biased towards the body text of the email being processed or the target text described by the customer service representative.

[0051] Furthermore, the step of “respectively matching the text vectors corresponding to the main text information and the target text information with the text vectors corresponding to the historical work order text set” specifically includes:

[0052] Separating the main text information and the target text information into sentences to obtain the sentences of the main text information and the target text information;

[0053] Converting the sentences of the main text information and the target text information into sentence vectors of the main text information and the target text information;

[0054] splicing the sentence vectors of the main text information to obtain a text vector of the main text information, and splicing the sentence vectors of the target text information to obtain a text vector of the target text information;

[0055] The text vector of the main text information and the text vector of the target text information are matched with the text vectors corresponding to the historical work order text set respectively.

[0056] Since sentence segmentation can capture the semantic information of each part of the text, the sentence vector at the sentence level can better reflect the meaning of each sentence in the text without causing information loss. Therefore, the main text information and the target text information are segmented separately, and the sentences of the main text information and the target text information are converted into sentence vectors of the main text information and sentence vectors of the target text information. This can accurately represent the sentence vectors of the main text information of the email to be processed and the corresponding features of the target text information. The sentence vectors of the main text information are concatenated to obtain the text vector of the main text information, and the sentence vectors of the target text information are concatenated to obtain the text vector of the target text information. The text vectors of the main text information and the text vectors of the target text information are matched with the text vectors corresponding to the historical work order text set. The specific matching method can be to calculate the cosine similarity between the text vectors, and use the calculated similarity as the matching score.

[0057] In one embodiment, the aforementioned "pre-built semantic recognition model includes an attention mechanism network, multiple re-parameterized networks, and a recognition network, and the inputting of the body information and the target text information into the pre-built semantic recognition model to obtain the semantic text of the email to be processed" includes:

[0058] Extracting features of the text information using multiple reparameterized networks and concatenating them to obtain a fusion feature matrix of the text information;

[0059] Extracting initial features of the target text information, and extending the initial features to obtain an extended feature matrix of the target text information;

[0060] Processing the extended feature matrix using an attention mechanism network to obtain an attention feature matrix of the target text information;

[0061] Fusing the fusion feature matrix of the main text information and the attention feature matrix of the target text information to obtain a fusion feature matrix;

[0062] The fused feature matrix is ​​input into the recognition network to obtain the semantic text of the email to be processed.

[0063] A reparameterized network combines convolution (Conv) and batch normalization (BN) to reduce computational complexity. Multiple reparameterized networks are used to extract features from the text message. The features extracted by each reparameterized network are then combined to create a fused feature matrix for the text message. Because multiple reparameterized networks can extract features from different perspectives, combining the features extracted by these networks provides a more comprehensive description of the text message in the email being processed.

[0064] By extending the initial features of the target text information, feature information of the target text information can be extracted from different scales. The extended features can fuse information of different dimensions (for example, the sentiment features of the text), thereby more effectively capturing the connection between the contexts of the target text information. The initial features can be extended by embedding the initial features into a graph structure through a graph neural network, and using the relationship between graph nodes as feature extension.

[0065] Since the attention mechanism network can learn and highlight the important parts of the text and suppress unimportant information, using the attention mechanism network to process the extended feature matrix can improve the subsequent model's ability to capture important features in the target text information, and enable the subsequent recognition network to extract more meaningful information when facing complex target text information.

[0066] After obtaining the fusion feature matrix of the main text information and the attention feature matrix of the target text information, the fusion feature matrix of the main text information and the attention feature matrix of the target text information are fused to obtain a fusion feature matrix. The fusion feature matrix can represent the features of the main text information of the email to be processed and the corresponding target text information. The fusion feature matrix is ​​input into the recognition network to obtain the semantic text of the email to be processed, wherein the recognition network can be trained based on the BERT model.

[0067] The step of “using multiple reparameterized networks to extract features of the text information and perform concatenation to obtain a fusion feature matrix of the text information” includes:

[0068] Extracting features of the text information using a first parameterized network to obtain a first text feature;

[0069] Inputting the first text feature into a second parameterized network and outputting a second text feature;

[0070] splicing the first text feature and the second text feature and inputting the concatenated features into a third parameterized network to obtain a third text feature;

[0071] The first text features, the second text features, and the third text features are fused to obtain a fused feature matrix for the body text information. The first, second, and third parameterized networks each comprise a network with a different convolution kernel. By extracting features from the body text information using reparameterized networks at different levels, a more comprehensive feature description of the body text information of the email being processed can be obtained.

[0072] After obtaining the semantic text of the email to be processed, the semantic keywords corresponding to the semantic text and the question type represented by the semantic text are extracted. Semantic keywords can be extracted using the TextRank algorithm or the TF-IDF algorithm, and semantic keywords indicate their importance to the semantic text. The question type represented by the semantic text refers to the question description information extracted from the semantic text. This question description information can be used as the subject of the target work order that is subsequently generated, for example, the subject of a work order related to a specific product or project.

[0073] Based on the sender's information and the type of issue, a blank work order template can be matched from the template library. For example, if the sender is typically a downstream channel partner, a corresponding work order template set is preconfigured for each channel partner, including work orders for each product and project. The target work order can be generated by filling the work order template with semantic keywords and semantic text.

[0074] In one embodiment, after generating the target work order, the method further includes:

[0075] The target work order is sent to a target user, and the target work order is associated with the email to be processed.

[0076] The target user can be a user of the company where the customer system is located, or an upstream supplier. The target work order is associated with the pending email through a unique identification number to facilitate subsequent tracing.

[0077] Reference Figure 2 , which is a functional module diagram of the work order generation device 100 based on email information of this application.

[0078] The work order generation device 100 based on email information described in this application can be installed in an electronic device. Depending on the functions implemented, the work order generation device 100 based on email information can include a parsing module 110, a conversion module 120, a first extraction module 130, a second extraction module 140, and a generation module 150. The module described in this application can also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can perform fixed functions, which are stored in the memory of the electronic device.

[0079] In this embodiment, the functions of each module / unit are as follows:

[0080] Parsing module 110: used to parse the email to be processed and obtain the body information and sender information of the email to be processed;

[0081] The conversion module 120 is configured to obtain the voice description information input by the customer service user for the main text information and convert the voice description information into target text information, wherein the target text information is semantically associated with the main text information;

[0082] A first extraction module 130 is configured to extract the semantic text of the email to be processed based on the body text information and the target text information;

[0083] The second extraction module 140 is used to extract the semantic keywords corresponding to the semantic text and the question type represented by the semantic text;

[0084] The generating module 150 is configured to determine a work order template based on the sender information and the problem type, and fill the semantic keywords and the semantic text into the work order template to generate a target work order.

[0085] In one embodiment, extracting the semantic text of the email to be processed based on the body text information and the target text information includes:

[0086] Matching the main text information and the target text information with the historical work order text set to obtain a matching score;

[0087] If the matching score is greater than a preset value, the historical work order text corresponding to the matching score is used as the semantic text of the email to be processed;

[0088] If the matching score is less than or equal to a preset value, the body text information and the target text information are input into a pre-built semantic recognition model to obtain the semantic text of the email to be processed.

[0089] In one embodiment, the matching of the main text information and the target text information with the historical work order text set to obtain a matching score includes:

[0090] Matching the text vectors corresponding to the main text information and the target text information with the text vectors corresponding to the historical work order text set to obtain a first score and a second score;

[0091] The average of the first score and the second score is used as the matching score.

[0092] In one embodiment, matching the text vectors corresponding to the main text information and the target text information with the text vectors corresponding to the historical work order text set includes:

[0093] Separating the main text information and the target text information into sentences to obtain the sentences of the main text information and the target text information;

[0094] Converting the sentences of the main text information and the target text information into sentence vectors of the main text information and the target text information;

[0095] splicing the sentence vectors of the main text information to obtain a text vector of the main text information, and splicing the sentence vectors of the target text information to obtain a text vector of the target text information;

[0096] The text vector of the main text information and the text vector of the target text information are matched with the text vectors corresponding to the historical work order text set respectively.

[0097] In one embodiment, the pre-built semantic recognition model includes an attention mechanism network, multiple re-parameterized networks, and a recognition network. Inputting the body information and the target text information into the pre-built semantic recognition model to obtain the semantic text of the email to be processed includes:

[0098] Extracting features of the text information using multiple reparameterized networks and concatenating them to obtain a fusion feature matrix of the text information;

[0099] Extracting initial features of the target text information, and extending the initial features to obtain an extended feature matrix of the target text information;

[0100] Processing the extended feature matrix using an attention mechanism network to obtain an attention feature matrix of the target text information;

[0101] Fusing the fusion feature matrix of the main text information and the attention feature matrix of the target text information to obtain a fusion feature matrix;

[0102] The fused feature matrix is ​​input into the recognition network to obtain the semantic text of the email to be processed.

[0103] In one embodiment, the extracting and concatenating features of the text information using multiple reparameterized networks to obtain a fused feature matrix of the text information includes:

[0104] Extracting features of the text information using a first parameterized network to obtain a first text feature;

[0105] Inputting the first text feature into a second parameterized network and outputting a second text feature;

[0106] splicing the first text feature and the second text feature and inputting the concatenated features into a third parameterized network to obtain a third text feature;

[0107] The first text feature, the second text feature, and the third text feature are fused to obtain a fused feature matrix of the text information.

[0108] In one embodiment, the work order generating apparatus 100 based on email information further includes an association module, which is configured to send the target work order to a target user and associate the target work order with the email to be processed.

[0109] Reference Figure 3 , which is a schematic diagram of a preferred embodiment of the electronic device of the present application.

[0110] The electronic device includes a processor 111, a communication interface 112, a memory 113 and a communication bus 114, wherein the processor 111, the communication interface 112, and the memory 113 communicate with each other via the communication bus 114;

[0111] Memory 113, for storing computer programs, for example, a work order generation program based on email information;

[0112] In some embodiments, the processor 111 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor 111 is generally used to control the overall operation of the electronic device, such as performing control and processing related to data interaction or communication. In this embodiment, the processor 111 is used to execute program code stored in the memory 113 or process data, such as executing the program code of a work order generation program based on email information.

[0113] The communication interface 112 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface). The communication interface 112 may also be used to establish a communication connection between the electronic device and other electronic devices.

[0114] The memory 113 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), 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 storage, magnetic disk, optical disk, etc. In some embodiments, the memory 113 may be an internal storage unit of the electronic device, such as the hard disk or memory of the electronic device. In other embodiments, the memory 113 may also be an external storage device of the electronic device, such as a plug-in hard disk, smart memory card (SMC), secure digital (SD) card, flash memory card, etc. equipped with the electronic device. Of course, the memory 113 may also include both the internal storage unit of the electronic device and its external storage device. In this embodiment, the memory 113 is generally used to store the operating system installed on the electronic device and various computer programs, such as the program code of a work order generation program based on email information. In addition, the memory 113 may also be used to temporarily store various types of data that have been output or are about to be output.

[0115] In one embodiment of the present application, the processor 111 is configured to execute a program stored in the memory 113 to implement the email-based work order generation method provided in any of the aforementioned method embodiments, including:

[0116] Parsing the email to be processed to obtain the body information and sender information of the email to be processed;

[0117] Acquiring voice description information input by a customer service user for the main text information, and converting the voice description information into target text information, wherein the target text information is semantically associated with the main text information;

[0118] Extracting the semantic text of the email to be processed based on the body text information and the target text information;

[0119] Extracting semantic keywords corresponding to the semantic text and the question type represented by the semantic text;

[0120] A work order template is determined based on the sender information and the problem type, and the semantic keywords and the semantic text are filled into the work order template to generate a target work order.

[0121] For a detailed description of the above steps, please refer to the Figure 1 Description of a flowchart of an embodiment of a method for generating a work order based on email information.

[0122] In addition, embodiments of the present application further provide a computer-readable storage medium, which may be non-volatile or volatile. The computer-readable storage medium includes a data storage area and a program storage area. The program storage area stores a work order generation program based on an email message. When the work order generation program based on an email message is executed by a processor, the following operations are performed:

[0123] Parsing the email to be processed to obtain the body information and sender information of the email to be processed;

[0124] Acquiring voice description information input by a customer service user for the main text information, and converting the voice description information into target text information, wherein the target text information is semantically associated with the main text information;

[0125] Extracting the semantic text of the email to be processed based on the body text information and the target text information;

[0126] Extracting semantic keywords corresponding to the semantic text and the question type represented by the semantic text;

[0127] A work order template is determined based on the sender information and the problem type, and the semantic keywords and the semantic text are filled into the work order template to generate a target work order.

[0128] The specific implementation of the computer-readable storage medium of the present application is roughly the same as the specific implementation of the above-mentioned work order generation method based on email information, and will not be repeated here.

[0129] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0130] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform, or of course, by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the relevant technology, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiment.

[0131] It should be noted that the descriptions of "first", "second", etc. in this application are for descriptive purposes only and should not be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" or "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but this must be based on the fact that they can be implemented by ordinary technicians in this field. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0132] It should be understood that the terms used herein are for the purpose of describing specific example embodiments only and are not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms "one", "an" and "said" as used herein may also be meant to include plural forms. The terms "comprise", "include", "contain" and "have" are inclusive and therefore specify the presence of stated features, steps, operations, elements and / or parts, but do not exclude the presence or addition of one or more other features, steps, operations, elements, parts, and / or combinations thereof. The method steps, processes, and operations described herein are not to be construed as necessarily requiring them to be performed in the specific order described or illustrated, unless the order of execution is clearly indicated. It should also be understood that additional or alternative steps may be used.

[0133] The foregoing is merely a list of specific embodiments of the present application, intended to enable those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the broadest scope consistent with the principles and novel features of the present application.

Claims

1. A work order generation method based on email information, characterized in that: The method comprises: Parsing the email to be processed to obtain the body information and sender information of the email to be processed; Acquiring voice description information input by a customer service user for the main text information, and converting the voice description information into target text information, wherein the target text information is semantically associated with the main text information; Extracting the semantic text of the email to be processed based on the body text information and the target text information; Extracting semantic keywords corresponding to the semantic text and the question type represented by the semantic text; Determine a work order template based on the sender information and the problem type, and fill the semantic keywords and the semantic text into the work order template to generate a target work order; The step of extracting the semantic text of the email to be processed based on the body text information and the target text information includes: Matching the main text information and the target text information with the historical work order text set to obtain a matching score; If the matching score is greater than a preset value, the historical work order text corresponding to the matching score is used as the semantic text of the email to be processed; If the matching score is less than or equal to a preset value, the body text information and the target text information are input into a pre-built semantic recognition model to obtain the semantic text of the email to be processed; The pre-built semantic recognition model includes an attention mechanism network, multiple re-parameterized networks, and a recognition network. Inputting the body information and the target text information into the pre-built semantic recognition model to obtain the semantic text of the email to be processed includes: Extracting features of the text information using multiple reparameterized networks and concatenating them to obtain a fusion feature matrix of the text information; Extracting initial features of the target text information, and extending the initial features to obtain an extended feature matrix of the target text information; Processing the extended feature matrix using an attention mechanism network to obtain an attention feature matrix of the target text information; Fusing the fusion feature matrix of the main text information and the attention feature matrix of the target text information to obtain a fusion feature matrix; Inputting the fused feature matrix into the recognition network to obtain the semantic text of the email to be processed; The method of extracting features of the text information using multiple reparameterized networks and concatenating the features to obtain a fusion feature matrix of the text information includes: Extracting features of the text information using a first parameterized network to obtain a first text feature; Inputting the first text feature into a second parameterized network and outputting a second text feature; splicing the first text feature and the second text feature and inputting the concatenated features into a third parameterized network to obtain a third text feature; The first text feature, the second text feature, and the third text feature are fused to obtain a fused feature matrix of the text information.

2. The method according to claim 1, characterized in that The matching of the main text information and the target text information with the historical work order text set to obtain a matching score includes: Matching the text vectors corresponding to the main text information and the target text information with the text vectors corresponding to the historical work order text set to obtain a first score and a second score; The average of the first score and the second score is used as the matching score.

3. The method according to claim 2, characterized in that Matching the text vectors corresponding to the main text information and the target text information with the text vectors corresponding to the historical work order text set includes: Separating the main text information and the target text information into sentences to obtain sentences of the main text information and the target text information; Converting the sentences of the main text information and the target text information into sentence vectors of the main text information and the target text information; splicing the sentence vectors of the main text information to obtain a text vector of the main text information, and splicing the sentence vectors of the target text information to obtain a text vector of the target text information; The text vector of the main text information and the text vector of the target text information are matched with the text vectors corresponding to the historical work order text set respectively.

4. The method according to claim 1, wherein The method further comprises: The target work order is sent to a target user, and the target work order is associated with the email to be processed.

5. A work order generation device based on email information, characterized in that: The device comprises: Parsing module: used to parse the email to be processed and obtain the body information and sender information of the email to be processed; A conversion module is configured to obtain voice description information input by a customer service user for the main text information and convert the voice description information into target text information, wherein the target text information is semantically associated with the main text information; A first extraction module: configured to extract the semantic text of the email to be processed based on the body information and the target text information; A second extraction module: used to extract semantic keywords corresponding to the semantic text and the question type represented by the semantic text; A generation module is configured to determine a work order template based on the sender information and the problem type, and fill the semantic keywords and the semantic text into the work order template to generate a target work order; The step of extracting the semantic text of the email to be processed based on the body text information and the target text information includes: Matching the main text information and the target text information with the historical work order text set to obtain a matching score; If the matching score is greater than a preset value, the historical work order text corresponding to the matching score is used as the semantic text of the email to be processed; If the matching score is less than or equal to a preset value, the body text information and the target text information are input into a pre-built semantic recognition model to obtain the semantic text of the email to be processed; The pre-built semantic recognition model includes an attention mechanism network, multiple re-parameterized networks, and a recognition network. Inputting the body information and the target text information into the pre-built semantic recognition model to obtain the semantic text of the email to be processed includes: Extracting features of the text information using multiple reparameterized networks and concatenating them to obtain a fusion feature matrix of the text information; Extracting initial features of the target text information, and extending the initial features to obtain an extended feature matrix of the target text information; Processing the extended feature matrix using an attention mechanism network to obtain an attention feature matrix of the target text information; Fusing the fusion feature matrix of the main text information and the attention feature matrix of the target text information to obtain a fusion feature matrix; Inputting the fused feature matrix into the recognition network to obtain the semantic text of the email to be processed; The method of extracting features of the text information using multiple reparameterized networks and concatenating the features to obtain a fusion feature matrix of the text information includes: Extracting features of the text information using a first parameterized network to obtain a first text feature; Inputting the first text feature into a second parameterized network and outputting a second text feature; splicing the first text feature and the second text feature and inputting the concatenated features into a third parameterized network to obtain a third text feature; The first text feature, the second text feature, and the third text feature are fused to obtain a fused feature matrix of the text information.

6. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; Memory for storing computer programs; The processor is configured to implement the email information-based work order generation method according to any one of claims 1 to 4 when executing the program stored in the memory.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the work order generation method based on email information according to any one of claims 1 to 4 is implemented.

Images