Mail processing method and device, storage medium and electronic equipment

Abstract

The application discloses a mail processing method and device, a storage medium and an electronic device, and relates to the technical field of artificial intelligence. The method comprises the following steps: obtaining N target information, wherein the N target information is at least one of the following: content information of M user historical processed mails, timeliness information of the M user historical processed mails, and frequency information of the M user historical processed mails; determining target data information of a target user processing mails according to the N target information; and generating a processing list for recommending the target user to process Z target mails based on the target data information of the target user processing mails and the Z target mails to be processed, wherein the processing list is used to determine a processing sequence of the Z target mails. Through the application, the problem that the user's working efficiency is low due to the long time for the user to process mails in the related art is solved, in which the mails are first manually classified and then processed.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and more specifically, to a method and apparatus for processing emails, a storage medium, and an electronic device. Background Technology

[0002] Currently, most email systems on the market lack automatic email categorization functionality. Furthermore, basic email categorization relies on manual methods such as pinning and highlighting important emails. Given the rapid development of communication technology, email system usage is declining; without a substantial volume of emails, email categorization becomes redundant. Additionally, when using internal email systems in large enterprises, the frequent communication between departments means that some employees may send and receive hundreds of emails daily. Users often lack the time to read and reply to these emails, potentially missing important information and crucial response deadlines, leading to a series of uncontrollable consequences.

[0003] Therefore, a common challenge faced by users of related technologies is how to efficiently extract important or timely emails from a large volume of emails. Furthermore, the use of internal email systems in large enterprises incurs excessive time costs for employees, impacting daily work efficiency and reducing effective working hours.

[0004] There is currently no effective solution to the problem that manual sorting and processing of emails in related technologies leads to excessively long processing times for users, resulting in low work efficiency. Summary of the Invention

[0005] The main objective of this application is to provide a method, apparatus, storage medium, and electronic device for processing emails, in order to solve the problem in related technologies where emails are first manually classified and then processed, resulting in excessively long processing times for users and consequently low user work efficiency.

[0006] To achieve the above objectives, according to one aspect of this application, a method for processing emails is provided. The method includes: acquiring N target information pieces, wherein the N target information pieces are at least one of the following: content information of emails historically processed by M users, timeliness information of the emails historically processed by the M users, and frequency information of the emails historically processed by the M users, wherein M and N are both positive integers greater than 1; determining target data information for a target user to process emails based on the N target information pieces, wherein the target user is a user among the M users, and the target data information is at least one of the following: timeliness information of the emails processed by the target user, frequency information of the emails processed by the target user, and content information of the emails processed by the target user; and generating a processing list for recommending the processing of the Z target emails to the target user based on the target data information for the emails processed by the target user and Z target emails to be processed, wherein the processing list is used to determine the processing order of the Z target emails, and Z is a positive integer greater than 1.

[0007] Further, if the N target information consists of the content information of the emails processed by the M users in the past, the timeliness information of the emails processed by the M users in the past, and the frequency information of the emails processed by the M users in the past, determining the target data information of the emails processed by the target user based on the N target information includes: processing the content information of the emails processed by the M users in the past to obtain a word vector set; encoding the timeliness information of the emails processed by the M users in the past to obtain a first vector set; encoding the frequency information of the emails processed by the M users in the past to obtain a second vector set; obtaining a target vector set based on the word vector set, the first vector set, and the second vector set; and determining the target data information of the emails processed by the target user based on the target vector set.

[0008] Further, obtaining the target vector set based on the word vector set, the first vector set, and the second vector set includes: determining a first matrix corresponding to each first vector in the first vector set, and determining a second matrix corresponding to each second vector in the second vector set; multiplying each first vector by the first matrix to obtain a third vector set; multiplying each second vector by the second matrix to obtain a fourth vector set; and combining the third vector set, the fourth vector set, and the word vector set to obtain the target vector set.

[0009] Further, determining the target data information for the target user to process emails based on the target vector set includes: obtaining the ID information of each user; determining the fifth vector corresponding to the target user from the target vector set based on the ID information of each user; and inputting the fifth vector into the target layer of the first model for processing to obtain the target data information for the target user to process emails, wherein the first model is used to generate the processing list.

[0010] Furthermore, based on the target data information of the target user's email processing and the Z target emails to be processed, generating a processing list for recommending the processing of the Z target emails to the target user includes: obtaining a target model, wherein the target model is a model obtained after training the first model; inputting the target data information of the target user's email processing and the Z target emails to be processed by the target user into the output layer of the target model for processing to obtain the processing list.

[0011] Further, obtaining the target model includes: obtaining a loss function; and training the first model using the loss function to obtain the target model.

[0012] Furthermore, after generating a processing list for recommending the processing of the Z target emails to the target user based on the target user's target data information for processing emails and the Z target emails to be processed, the method further includes: obtaining feedback information from the target user; and determining the accuracy of recommending the processing list of the Z target emails to the target user based on the feedback information from the target user.

[0013] To achieve the above objectives, according to another aspect of this application, an email processing apparatus is provided. The apparatus includes: a first acquisition unit, configured to acquire N target information, wherein the N target information is at least one of the following: content information of emails historically processed by M users, timeliness information of the emails historically processed by the M users, and frequency information of the emails historically processed by the M users, wherein M and N are both positive integers greater than 1; a first determination unit, configured to determine target data information for a target user to process emails based on the N target information, wherein the target user is a user among the M users, and the target data information is at least one of the following: timeliness information of the emails processed by the target user, frequency information of the emails processed by the target user, and content information of the emails processed by the target user; and a first generation unit, configured to generate a processing list for recommending the processing of the Z target emails to the target user based on the target data information for the emails processed by the target user and Z target emails to be processed, wherein the processing list is used to determine the processing order of the Z target emails, and Z is a positive integer greater than 1.

[0014] Further, if the N target information consists of the content information of the emails historically processed by the M users, the timeliness information of the emails historically processed by the M users, and the frequency information of the emails historically processed by the M users, the first determining unit includes: a first processing module, used to process the content information of the emails historically processed by the M users to obtain a set of word vectors; a second processing module, used to encode the timeliness information of the emails historically processed by the M users to obtain a first vector set; a third processing module, used to encode the frequency information of the emails historically processed by the M users to obtain a second vector set; a fourth processing module, used to obtain a target vector set based on the set of word vectors, the first vector set, and the second vector set; and a first determining module, used to determine the target data information of the emails processed by the target user based on the target vector set.

[0015] Further, the fourth processing module includes: a first determining submodule, used to determine a first matrix corresponding to each first vector in the first vector set, and to determine a second matrix corresponding to each second vector in the second vector set; a first processing submodule, used to multiply each first vector by the first matrix to obtain a third vector set; a second processing submodule, used to multiply each second vector by the second matrix to obtain a fourth vector set; and a third processing submodule, used to combine the third vector set, the fourth vector set, and the word vector set to obtain the target vector set.

[0016] Further, the first determining module includes: a first obtaining submodule, used to obtain the ID information of each user; a second determining submodule, used to determine the fifth vector corresponding to the target user from the target vector set based on the ID information of each user; and a fourth processing submodule, used to input the fifth vector into the target layer of the first model for processing to obtain the target data information of the target user's processed email, wherein the first model is used to generate the processing list.

[0017] Further, the first generation unit includes: a first acquisition module, used to acquire a target model, wherein the target model is a model obtained after training the first model; and a fifth processing module, used to input the target data information of the target user's processed emails and the Z target emails to be processed by the target user into the output layer of the target model for processing to obtain the processing list.

[0018] Furthermore, the first acquisition module includes: a second acquisition submodule for acquiring a loss function; and a first training submodule for training the first model using the loss function to obtain the target model.

[0019] Furthermore, the device further includes: a second acquisition unit, configured to acquire feedback information from the target user after generating a processing list for recommending the processing of the Z target emails to the target user based on the target data information of the target user processing emails and the Z target emails to be processed; and a second determination unit, configured to determine the accuracy of recommending the processing list of the Z target emails to the target user based on the feedback information from the target user.

[0020] To achieve the above objectives, according to another aspect of this application, a computer-readable storage medium is provided, the storage medium storing a program, wherein the program executes the email processing method described in any of the above claims.

[0021] To achieve the above objectives, according to another aspect of this application, an electronic device is provided, the electronic device including one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the email processing method described in any of the above.

[0022] This application employs the following steps: First, obtain N target information items, where each of the N target information items is at least one of the following: content information of emails processed by M users in the past, timeliness information of emails processed by M users in the past, and frequency information of emails processed by M users in the past, where M and N are both positive integers greater than 1. Second, based on the N target information items, determine the target data information for emails processed by the target user, where the target user is one of the M users, and the target data information is at least one of the following: timeliness information of emails processed by the target user, frequency information of emails processed by the target user, and content information of emails processed by the target user. Third, based on the target data information for emails processed by the target user and Z target emails to be processed, generate a processing list for recommending the processing of the Z target emails to the target user, where the processing list is used to determine the processing order of the Z target emails. This solves the problem in related technologies where emails are first manually classified and then processed, leading to excessively long processing times for users and consequently low user efficiency. Based on the obtained N target information, the target data information for the target user to process emails is determined, and the target data information is at least one of the following: the timeliness information of the target user's email processing, the frequency information of the target user's email processing, and the content information of the target user's email processing. Based on the target data information of the target user's email processing and Z target emails to be processed, a processing list for recommending the processing of Z target emails to the target user is generated. This avoids manually classifying emails before processing them, thereby saving the user's email processing time and improving the user's work efficiency. Attached Figure Description

[0023] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0024] Figure 1 This is a flowchart of an email processing method provided according to an embodiment of this application;

[0025] Figure 2 This is a flowchart of the email processing method provided in the embodiments of this application. Figure 1 ;

[0026] Figure 3 This is a flowchart of an optional email processing method provided according to an embodiment of this application;

[0027] Figure 4 This is a schematic diagram of an email processing apparatus provided according to an embodiment of this application;

[0028] Figure 5 This is a schematic diagram of an electronic device provided according to an embodiment of this application. Detailed Implementation

[0029] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0030] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0031] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0032] It should be noted that the user information (including but not limited to user device information, user personal information, content information of emails processed by multiple users in the past, timeliness information of emails processed by multiple users in the past, frequency information of emails processed by multiple users in the past, timeliness information of emails processed by the target user, frequency information of emails processed by the target user, content information of emails processed by the target user, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0033] For ease of description, the following explains some of the nouns or terms used in the embodiments of this application:

[0034] HR: Hit Rate, used to calculate the proportion of items in the test set out of the total number of items in the TopN recommendation list generated by the algorithm.

[0035] NDCG: Normalized Discount Cumulative Gain. The goal of NDCG is to make the ranking of the generated recommendation list as accurate as possible. Its essence is to measure the ranking accuracy of the algorithm. For example, if a user is not interested in an item, but it is at the top of the recommendation list, it will affect the user experience.

[0036] Optimizer: After an algorithm model is designed, it usually needs to be continuously optimized to update the parameters to minimize the value of the loss function, which is the optimal solution of the model. This optimization method is called an optimizer. Common optimizers include gradient descent, stochastic gradient descent, RMSprop (adaptive learning rate optimization), and Adam algorithm (a stochastic optimization algorithm), etc.

[0037] The present invention will now be described in conjunction with preferred implementation steps. Figure 1 This is a flowchart of an email processing method provided according to an embodiment of this application, such as... Figure 1 As shown, the method includes the following steps:

[0038] Step S101: Obtain N target information items, wherein the N target information items are at least one of the following: content information of M historical emails processed by users, timeliness information of M historical emails processed by users, and frequency information of M historical emails processed by users, wherein M and N are both positive integers greater than 1.

[0039] For example, the target information mentioned above could be data such as the content of emails processed by an employee in the past, processing time, processing frequency, and number of forwards. Furthermore, the M users mentioned above can refer to multiple employees.

[0040] Step S102: Based on N target information, determine the target data information for the target user to process emails, where the target user is one of the M users, and the target data information is at least one of the following: the timeliness information of the target user processing emails, the frequency information of the target user processing emails, and the content information of the target user processing emails.

[0041] For example, by preprocessing the content, processing time, and processing frequency of historical emails handled by employees, an employee's email handling preferences can be formed. Furthermore, the aforementioned target data information can be user email handling preference information; that is, the user's email handling preference information could include the processing time and frequency of such emails.

[0042] Step S103: Based on the target data information of the target user's emails and the Z target emails to be processed, generate a processing list to recommend the processing of the Z target emails to the target user. The processing list is used to determine the processing order of the Z target emails, where Z is a positive integer greater than 1.

[0043] For example, emails received by employees can be categorized based on their email handling preferences, and a corresponding recommendation list (the aforementioned processing list) can be generated for each user. This serves to remind employees to prioritize their emails, while emails that are historically frequently ignored can be placed at the bottom of the recommendation list as unimportant emails, allowing employees to delay their processing.

[0044] Through the above steps S101 to S103, based on the N target information obtained, the target data information for the target user to process emails is determined, and the target data information is at least one of the following: the timeliness information of the target user to process emails, the frequency information of the target user to process emails, and the content information of the target user to process emails. Based on the target data information for the target user to process emails and the Z target emails to be processed, a processing list for recommending the processing of the Z target emails to the target user is generated. This avoids manually classifying emails before processing them, thereby saving the user's time in processing emails and thus improving the user's work efficiency.

[0045] Figure 2 This is a flowchart of the email processing method provided in the embodiments of this application. Figure 1 ,like Figure 2 As shown, in the email processing method provided in this application embodiment, if the N target information consists of the content information of M emails processed by M users in the past, the timeliness information of M emails processed by M users in the past, and the frequency information of M emails processed by M users in the past, the target data information for the target user's email processing is determined based on the N target information, including:

[0046] Step S201: Process the content information of the emails processed by M users in the past to obtain a set of word vectors;

[0047] Step S202: Encode the timeliness information of the historical processed emails of M users to obtain the first vector set;

[0048] Step S203: Encode the frequency information of historical emails processed by M users to obtain a second vector set;

[0049] Step S204: Based on the word vector set, the first vector set, and the second vector set, obtain the target vector set;

[0050] Step S205: Based on the target vector set, determine the target data information for the target user to process the email.

[0051] For example, in terms of data acquisition, data on the processing time and frequency of various user emails are obtained through the email system's backend. Therefore, the three effective features to focus on in the data preprocessing part are email content, processing time, and processing frequency. Among them, email content is a text-based feature, and text-based features can be processed into word vectors using the Word2vec model (a related model used to generate word vectors). In layman's terms, the importance of an email is mostly reflected in its content, and the degree of importance will vary depending on different occupational types and work focuses. For example, emails we receive in our daily office work often indicate "important," "please complete before **date**," etc. Or, "**course training, required personnel," "email from department email or department on-call email," etc. These types of emails usually require our timely processing, but not necessarily. Therefore, the above-mentioned word vector set can be a collection of multiple word vectors obtained by processing email content using the Word2vec model. Processing timeliness and frequency are numerical features, which can be processed using One-hot vector encoding (an encoding method) to convert them into binary vectors of 0 and 1. A 1 indicates the presence of the feature in the encoded vector, while a 0 indicates its absence. Therefore, the first vector set mentioned above can be a set of multiple binary vectors obtained by processing the timeliness of a user's historical email processing using One-hot vector encoding, and the second vector set mentioned above can be a set of multiple binary vectors obtained by processing the frequency of a user's historical email processing using One-hot vector encoding. Then, based on the obtained word vector set, the first vector set, and the second vector set, the final vector set to be used (the aforementioned target vector set) is obtained. Finally, based on the final vector set to be used, the user's email processing preference information (the aforementioned target data information) is determined.

[0052] In summary, data such as the content, processing time, and frequency of emails processed by users in the past can be obtained through the email system. Compared with other recommendation algorithms, this greatly reduces the cost of data acquisition and improves the efficiency of data acquisition, thereby enabling the rapid and accurate determination of user preference information.

[0053] Optionally, in the email processing method provided in this application embodiment, obtaining the target vector set based on the word vector set, the first vector set, and the second vector set includes: determining the first matrix corresponding to each first vector in the first vector set, and determining the second matrix corresponding to each second vector in the second vector set; multiplying each first vector by the first matrix to obtain a third vector set; multiplying each second vector by the second matrix to obtain a fourth vector set; and combining the third vector set, the fourth vector set, and the word vector set to obtain the target vector set.

[0054] For example, vectors processed by one-hot encoding are typically very sparse. Furthermore, since the processing frequency and timeliness are subjective factors in determining user email preferences, the data sparsity problem common in recommendation algorithms may occur, making subsequent model training difficult. Since dimensionality reduction can address the sparsity issue to some extent, the processed one-hot vectors can be dimensionality-reduced. Specifically, the one-hot encoded vector can be multiplied by the corresponding dimension's weight matrix to obtain a relatively dense embedding vector (the vectors in the third or fourth vector set mentioned above). This is then combined with the word vectors obtained from the text feature processing above to obtain the complete embedding vector (the vectors in the target vector set mentioned above).

[0055] The above approach increases the amount of data, addresses the problem of data sparsity, and allows the model to be trained using this data, thereby improving the accuracy of model training and reducing its difficulty.

[0056] Optionally, in the email processing method provided in this application embodiment, determining the target data information for the target user to process emails based on the target vector set includes: obtaining the ID information of each user; determining the fifth vector corresponding to the target user from the target vector set based on the ID information of each user; and inputting the fifth vector into the target layer of the first model for processing to obtain the target data information for the target user to process emails, wherein the first model is used to generate a processing list.

[0057] For example, by assigning a user a unique identifier (ID) and interacting with the resulting complete embedding vector, we can obtain the user's email processing preferences under different email content, processing time, and processing frequency. That is, based on the user ID, we can determine the user's corresponding embedding vector (i.e., the embedding vector corresponding to the user's historical email content, processing time, and processing frequency) from the final complete embedding vector set. Then, we input the user's corresponding embedding vector into the preference layer (the target layer) of the email processing recommendation model (the first model mentioned above) for processing to obtain the user's email processing preferences (the target data information mentioned above).

[0058] The above method can quickly and accurately obtain user preference information.

[0059] Optionally, in the email processing method provided in this application embodiment, generating a processing list for recommending the processing of Z target emails to the target user based on the target data information of the target user processing emails and Z target emails to be processed includes: obtaining a target model, wherein the target model is a model obtained after training a first model; and inputting the target data information of the target user processing emails and the Z target emails to be processed by the target user into the output layer of the target model for processing to obtain the processing list.

[0060] For example, a recommendation list (the processing list mentioned above) can be generated based on preferences for the corresponding user. Specifically, the user's email processing preferences (the target data information mentioned above) and multiple emails received by the user that need to be processed are input into the output layer of a trained email processing recommendation model (the target model mentioned above) for processing, resulting in the recommendation list (the processing list mentioned above).

[0061] In conclusion, processing emails based on automatically generated recommendation lists can effectively reduce the time employees spend on emails in their daily work, thereby improving work efficiency.

[0062] Optionally, in the email processing method provided in this application embodiment, obtaining the target model includes: obtaining a loss function; and training a first model using the loss function to obtain the target model.

[0063] For example, when training an email processing recommendation model, the training process can use the binary cross-entropy loss function (the loss function mentioned above) as the objective function, and select the Adam optimizer, which has the best overall performance. The loss function is typically used to measure the performance of the model, that is, the degree of difference between the predicted result and the actual result. Its value is non-negative and a real number. During model training, the loss function is used, and the obtained loss value is used to update the parameters through backpropagation, thereby reducing the loss. The smaller the loss value, the better the model's performance. By continuously updating the weights and parameters, and under the constraint of a certain number of iterations, the optimal weights and parameters, as well as the optimal email processing recommendation list, are saved.

[0064] In summary, by iterating through training the model, a well-trained model can be easily obtained.

[0065] Optionally, in the email processing method provided in this application embodiment, after generating a processing list for recommending the processing of Z target emails to the target user based on the target data information of the target user and Z target emails to be processed, the method further includes: obtaining feedback information from the target user; and determining the accuracy of recommending the processing list of Z target emails to the target user based on the feedback information from the target user.

[0066] For example, in a typical recommendation algorithm process, the final generated recommendation list uses HR and NDCG values ​​as evaluation metrics for algorithm accuracy. However, in this embodiment, employees can directly provide feedback on the recommendation accuracy after using the product.

[0067] The above approach can quickly improve the accuracy of the processing list recommended to users.

[0068] For example, Figure 3 This is a flowchart of an optional email processing method provided according to an embodiment of this application, such as... Figure 3 As shown, the optional email processing methods include the following steps:

[0069] Firstly, regarding data acquisition, data on the timeliness and frequency of various email processing tasks are obtained from the email system's backend and input into the input layer of the email processing recommendation model. Furthermore, the three key features for data preprocessing are email content, processing timeliness, and processing frequency. Email content is a text-based feature. In this invention, text-based features are processed into word vectors using a Word2vec model. Simply put, the importance of an email is largely reflected in its content, and this importance varies depending on different professions and work priorities. For example, emails received in our daily work often specify "Important," "Please complete by **date**," etc. Or, emails such as "**Course Training, Required Personnel," or "Email from the department email address or department on-call email address" usually require timely processing, but this doesn't guarantee immediate attention. Processing timeliness and frequency are numerical features. This invention intends to process them using One-hot vector encoding, converting them into binary vectors of 0 and 1. A 1 indicates the presence of the feature in the encoded vector, while a 0 indicates its absence. The processed One-hot vectors are then subjected to dimensionality reduction. Typically, vectors processed by one-hot encoding are very sparse. Furthermore, the processing frequency and timeliness are subjective factors used to judge user email preferences, potentially leading to data sparsity, a common problem in recommendation algorithms. This can make subsequent model training difficult. Dimensionality reduction can address this issue to some extent. Specifically, the one-hot encoded vector is multiplied by the corresponding weight matrix to obtain a relatively dense embedding vector. This is then combined with the word vectors obtained from the text feature processing described above to obtain the complete embedding vector. This layer is the embedding layer of this model.

[0070] Next is the model's preference layer, which, as the name suggests, obtains the user's email processing preferences at this layer. Unlike other types of recommendation algorithms, the processing time and frequency features obtained in this invention are sufficient to reflect user preferences. To demonstrate the rationality and completeness of this invention, users are assigned a number, user IDs are generated, and these are interacted with the embedding vectors mentioned above to obtain the user's email processing preferences under different email content, processing time, and processing frequency.

[0071] Based on preferences, a recommendation list for the corresponding user is generated, followed by training iterations. During training, the binary cross-entropy loss function is used as the objective function, and the Adam optimizer, which offers the best overall performance, is selected. The loss function is typically used to measure the performance of a model, that is, the degree of difference between the predicted and actual results. Its value is non-negative and a real number. During model training, the loss function is used, and the obtained loss value is updated through backpropagation to reduce the loss. The smaller the loss value, the better the model's performance. Weights and parameters are continuously updated, and under the constraint of a certain number of iterations, the optimal weights and parameters, along with the optimal email processing recommendation list, are saved. In a typical recommendation algorithm, the generated recommendation list uses HR and NDCG values ​​as evaluation metrics for algorithm accuracy. Due to the convenience of this invention, employees can directly provide feedback on the recommendation accuracy after using the product.

[0072] In addition, feature extraction of email data can be manifested in the analysis and processing of email content to form corresponding features. It can also be achieved through deep learning-level feature extraction methods, such as convolutional neural networks, to extract features from emails and realize email processing and recommendation. Furthermore, emails can be classified by processing frequency and processing timeliness to achieve more efficient recommendation.

[0073] By automating the categorization of emails based on employees' processing habits, the above-mentioned solutions can significantly reduce the time employees spend processing emails daily, thereby reducing their workload and allowing them to focus on core tasks and improve work efficiency.

[0074] In summary, the email processing method provided in this application embodiment obtains N target information, wherein the N target information is at least one of the following: content information of emails processed by M users in the past, timeliness information of emails processed by M users in the past, and frequency information of emails processed by M users in the past, wherein M and N are both positive integers greater than 1; based on the N target information, the target data information for the target user to process emails is determined, wherein the target user is a user among the M users, and the target data information is at least one of the following: timeliness information of emails processed by the target user, frequency information of emails processed by the target user, and content information of emails processed by the target user; based on the target data information for the target user to process emails and Z target emails to be processed, a processing list for recommending the processing of Z target emails to the target user is generated, wherein the processing list is used to determine the processing order of Z target emails. This solves the problem in related technologies where emails are first manually classified and then processed, resulting in excessively long processing time for users and consequently low user work efficiency. Based on the obtained N target information, the target data information for the target user to process emails is determined, and the target data information is at least one of the following: the timeliness information of the target user's email processing, the frequency information of the target user's email processing, and the content information of the target user's email processing. Based on the target data information of the target user's email processing and Z target emails to be processed, a processing list for recommending the processing of Z target emails to the target user is generated. This avoids manually classifying emails before processing them, thereby saving the user's email processing time and improving the user's work efficiency.

[0075] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0076] This application also provides an email processing apparatus. It should be noted that the email processing apparatus of this application can be used to execute the email processing method provided in this application. The email processing apparatus provided in this application will be described below.

[0077] Figure 4 This is a schematic diagram of an email processing apparatus according to an embodiment of this application. Figure 4 As shown, the device includes: a first acquisition unit 401, a first determination unit 402, and a first generation unit 403.

[0078] Specifically, the first acquisition unit 401 is used to acquire N target information, wherein the N target information is at least one of the following: content information of M emails processed by users in the past, timeliness information of M emails processed by users in the past, and frequency information of M emails processed by users in the past, wherein M and N are both positive integers greater than 1.

[0079] The first determining unit 402 is used to determine the target data information of the target user processing the email based on N target information, wherein the target user is a user among M users, and the target data information is at least one of the following: the timeliness information of the target user processing the email, the frequency information of the target user processing the email, and the content information of the target user processing the email.

[0080] The first generation unit 403 is used to generate a processing list for recommending the processing of Z target emails to the target user based on the target data information of the target user's email processing and Z target emails to be processed. The processing list is used to determine the processing order of the Z target emails, where Z is a positive integer greater than 1.

[0081] In summary, the email processing apparatus provided in this application embodiment acquires N target information through the first acquisition unit 401, wherein the N target information is at least one of the following: content information of emails processed by M users in the past, timeliness information of emails processed by M users in the past, and frequency information of emails processed by M users in the past, wherein M and N are both positive integers greater than 1; the first determination unit 402 determines the target data information of the target user's email processing based on the N target information, wherein the target user is a user among the M users, and the target data information is at least one of the following: timeliness information of the target user's email processing, frequency information of the target user's email processing, and content information of the target user's email processing; the first generation unit 403 generates a processing list for recommending the processing of Z target emails to the target user based on the target data information of the target user's email processing and Z target emails to be processed, wherein the processing list is used to determine the processing order of the Z target emails, and Z is a positive integer greater than 1. This solves the problem in related technologies where emails are first manually classified and then processed, resulting in excessively long processing time for users and consequently low user work efficiency. Based on the obtained N target information, the target data information for the target user to process emails is determined, and the target data information is at least one of the following: the timeliness information of the target user's email processing, the frequency information of the target user's email processing, and the content information of the target user's email processing. Based on the target data information of the target user's email processing and Z target emails to be processed, a processing list for recommending the processing of Z target emails to the target user is generated. This avoids manually classifying emails before processing them, thereby saving the user's email processing time and improving the user's work efficiency.

[0082] Optionally, in the email processing apparatus provided in this application embodiment, if the N target information consists of the content information of M emails processed by M users in the past, the timeliness information of M emails processed by M users in the past, and the frequency information of M emails processed by M users in the past, the first determining unit includes: a first processing module, used to process the content information of the M emails processed by M users in the past to obtain a set of word vectors; a second processing module, used to encode the timeliness information of the M emails processed by M users in the past to obtain a first vector set; a third processing module, used to encode the frequency information of the M emails processed by M users in the past to obtain a second vector set; a fourth processing module, used to obtain a target vector set based on the set of word vectors, the first vector set, and the second vector set; and a first determining module, used to determine the target data information of the emails processed by the target user based on the target vector set.

[0083] Optionally, in the email processing apparatus provided in this application embodiment, the fourth processing module includes: a first determining submodule, used to determine a first matrix corresponding to each first vector in a first vector set, and to determine a second matrix corresponding to each second vector in a second vector set; a first processing submodule, used to multiply each first vector by the first matrix to obtain a third vector set; a second processing submodule, used to multiply each second vector by the second matrix to obtain a fourth vector set; and a third processing submodule, used to combine the third vector set, the fourth vector set, and the word vector set to obtain a target vector set.

[0084] Optionally, in the email processing apparatus provided in this application embodiment, the first determining module includes: a first obtaining submodule, used to obtain the ID information of each user; a second determining submodule, used to determine the fifth vector corresponding to the target user from the target vector set based on the ID information of each user; and a fourth processing submodule, used to input the fifth vector into the target layer of the first model for processing to obtain the target data information of the target user's email processing, wherein the first model is used to generate a processing list.

[0085] Optionally, in the email processing apparatus provided in this application embodiment, the first generation unit includes: a first acquisition module, used to acquire a target model, wherein the target model is a model obtained after training the first model; and a fifth processing module, used to input the target data information of the target user's email processing and the Z target emails to be processed by the target user into the output layer of the target model for processing to obtain a processing list.

[0086] Optionally, in the email processing apparatus provided in this application embodiment, the first acquisition module includes: a second acquisition submodule for acquiring a loss function; and a first training submodule for training a first model using the loss function to obtain a target model.

[0087] Optionally, in the email processing apparatus provided in the embodiments of this application, the apparatus further includes: a second acquisition unit, configured to acquire feedback information from the target user after generating a processing list for recommending the processing of Z target emails to the target user based on the target data information of the target user processing emails and Z target emails to be processed; and a second determination unit, configured to determine the accuracy of recommending the processing list of Z target emails to the target user based on the feedback information from the target user.

[0088] The email processing device includes a processor and a memory. The first acquisition unit 401, the first determination unit 402, and the first generation unit 403 are all stored in the memory as program units. The processor executes the program units stored in the memory to achieve the corresponding functions.

[0089] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and user efficiency can be improved by adjusting kernel parameters.

[0090] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0091] This invention provides a computer-readable storage medium storing a program thereon, which, when executed by a processor, implements the email processing method.

[0092] This invention provides a processor for running a program, wherein the program executes a method for processing the email.

[0093] like Figure 5As shown, this embodiment of the invention provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs the following steps: acquiring N target information, wherein the N target information is at least one of the following: content information of emails historically processed by M users, timeliness information of the emails historically processed by the M users, and frequency information of the emails historically processed by the M users, wherein M and N are both positive integers greater than 1; determining target data information for a target user to process emails based on the N target information, wherein the target user is a user among the M users, and the target data information is at least one of the following: timeliness information of the emails processed by the target user, frequency information of the emails processed by the target user, and content information of the emails processed by the target user; generating a processing list for recommending the processing of the Z target emails to the target user based on the target data information for the emails processed by the target user and Z target emails to be processed, wherein the processing list is used to determine the processing order of the Z target emails, and Z is a positive integer greater than 1.

[0094] When the processor executes the program, it also performs the following steps: If the N target information consists of the content information of the emails processed by the M users in the past, the timeliness information of the emails processed by the M users in the past, and the frequency information of the emails processed by the M users in the past, determining the target data information of the emails processed by the target user based on the N target information includes: processing the content information of the emails processed by the M users in the past to obtain a set of word vectors; encoding the timeliness information of the emails processed by the M users in the past to obtain a first set of vectors; encoding the frequency information of the emails processed by the M users in the past to obtain a second set of vectors; obtaining a target vector set based on the set of word vectors, the first set of vectors, and the second set of vectors; and determining the target data information of the emails processed by the target user based on the target vector set.

[0095] When the processor executes the program, it also performs the following steps: obtaining a target vector set based on the word vector set, the first vector set, and the second vector set includes: determining a first matrix corresponding to each first vector in the first vector set, and determining a second matrix corresponding to each second vector in the second vector set; multiplying each first vector by the first matrix to obtain a third vector set; multiplying each second vector by the second matrix to obtain a fourth vector set; and combining the third vector set, the fourth vector set, and the word vector set to obtain the target vector set.

[0096] When the processor executes the program, it also performs the following steps: determining the target data information for the target user to process emails based on the target vector set includes: obtaining the ID information of each user; determining the fifth vector corresponding to the target user from the target vector set based on the ID information of each user; inputting the fifth vector into the target layer of the first model for processing to obtain the target data information for the target user to process emails, wherein the first model is used to generate the processing list.

[0097] When the processor executes the program, it also performs the following steps: Based on the target data information of the target user's email processing and the Z target emails to be processed, generating a processing list for recommending the processing of the Z target emails to the target user includes: obtaining a target model, wherein the target model is a model obtained after training the first model; inputting the target data information of the target user's email processing and the Z target emails to be processed by the target user into the output layer of the target model for processing to obtain the processing list.

[0098] When the processor executes the program, it also performs the following steps: obtaining the target model includes: obtaining a loss function; training the first model using the loss function to obtain the target model.

[0099] When the processor executes the program, it also performs the following steps: after generating a processing list for recommending the processing of the Z target emails to the target user based on the target data information of the target user's email processing and the Z target emails to be processed, the method further includes: obtaining feedback information from the target user; and determining the accuracy of recommending the processing list of the Z target emails to the target user based on the feedback information from the target user.

[0100] The devices mentioned in this article can be servers, PCs, tablets, mobile phones, etc.

[0101] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having the following method steps: acquiring N target information, wherein the N target information is at least one of the following: content information of emails historically processed by M users, timeliness information of the emails historically processed by the M users, and frequency information of the emails historically processed by the M users, wherein M and N are both positive integers greater than 1; determining target data information for a target user to process emails based on the N target information, wherein the target user is a user among the M users, and the target data information is at least one of the following: timeliness information of the emails processed by the target user, frequency information of the emails processed by the target user, and content information of the emails processed by the target user; generating a processing list for recommending the processing of the Z target emails to the target user based on the target data information for the emails processed by the target user and Z target emails to be processed, wherein the processing list is used to determine the processing order of the Z target emails, and Z is a positive integer greater than 1.

[0102] When executed on a data processing device, it is also suitable to execute an initialization program with the following steps: if the N target information consists of the content information of the emails historically processed by the M users, the timeliness information of the emails historically processed by the M users, and the frequency information of the emails historically processed by the M users, determining the target data information of the emails processed by the target user based on the N target information includes: processing the content information of the emails historically processed by the M users to obtain a set of word vectors; encoding the timeliness information of the emails historically processed by the M users to obtain a first set of vectors; encoding the frequency information of the emails historically processed by the M users to obtain a second set of vectors; obtaining a target vector set based on the set of word vectors, the first set of vectors, and the second set of vectors; and determining the target data information of the emails processed by the target user based on the target vector set.

[0103] When executed on a data processing device, it is also suitable to execute an initialization program with the following method steps: obtaining a target vector set based on the word vector set, the first vector set, and the second vector set includes: determining a first matrix corresponding to each first vector in the first vector set, and determining a second matrix corresponding to each second vector in the second vector set; multiplying each first vector by the first matrix to obtain a third vector set; multiplying each second vector by the second matrix to obtain a fourth vector set; and combining the third vector set, the fourth vector set, and the word vector set to obtain the target vector set.

[0104] When executed on a data processing device, it is also suitable to execute an initialization program with the following method steps: determining the target data information for the target user to process emails based on the target vector set, including: obtaining the ID information of each user; determining the fifth vector corresponding to the target user from the target vector set based on the ID information of each user; inputting the fifth vector into the target layer of the first model for processing to obtain the target data information for the target user to process emails, wherein the first model is used to generate the processing list.

[0105] When executed on a data processing device, it is also suitable to execute an initialization program with the following method steps: generating a processing list for recommending the processing of the Z target emails to the target user based on the target user's target data information for processing emails and Z target emails to be processed, including: obtaining a target model, wherein the target model is a model obtained after training the first model; inputting the target user's target data information for processing emails and the Z target emails to be processed by the target user into the output layer of the target model for processing to obtain the processing list.

[0106] When executed on a data processing device, it is also suitable to execute an initialization program with the following method steps: obtaining the target model includes: obtaining a loss function; training the first model using the loss function to obtain the target model.

[0107] When executed on a data processing device, the method is also adapted to execute an initialization procedure having the following steps: after generating a processing list for recommending the processing of the Z target emails to the target user based on the target data information for processing emails by the target user and the Z target emails to be processed, the method further includes: obtaining feedback information from the target user; and determining the accuracy of recommending the processing list of the Z target emails to the target user based on the feedback information from the target user.

[0108] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0109] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0110] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0111] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0112] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0113] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0114] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0115] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0116] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0117] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for processing emails, characterized in that, include: Obtain N target information items, wherein the N target information items include: content information of M emails processed by users in the past, timeliness information of the M emails processed by users in the past, and frequency information of the M emails processed by users in the past, wherein M and N are both positive integers greater than 1; Based on the N target information, target data information for the target user's email processing is determined, wherein the target user is one of the M users, and the target data information includes: the timeliness information of the target user's email processing, the frequency information of the target user's email processing, and the content information of the target user's email processing. Based on the target data information of the target user's email processing and Z target emails to be processed, a processing list is generated to recommend the processing of the Z target emails to the target user, wherein the processing list is used to determine the processing order of the Z target emails, and Z is a positive integer greater than 1; If the N target information consists of the content information of the emails processed by the M users in the past, the timeliness information of the emails processed by the M users in the past, and the frequency information of the emails processed by the M users in the past, then based on the N target information, the target data information for the emails processed by the target user includes: The content information of the emails processed by the M users in the past is processed to obtain a set of word vectors; The timeliness information of the historical processed emails of the M users is encoded to obtain a first vector set; The frequency information of the historical processed emails of the M users is encoded to obtain a second vector set; Based on the word vector set, the first vector set, and the second vector set, a target vector set is obtained; Based on the target vector set, determine the target data information for the target user to process emails; Based on the word vector set, the first vector set, and the second vector set, the target vector set is obtained as follows: Determine the first matrix corresponding to each first vector in the first vector set, and determine the second matrix corresponding to each second vector in the second vector set, wherein the first matrix and the second matrix are weight matrices; Multiply each first vector by the first matrix to obtain the third set of vectors; Multiply each second vector by the second matrix to obtain the fourth set of vectors; The third vector set, the fourth vector set, and the word vector set are combined to obtain the target vector set; Based on the target vector set, the target data information for the target user to process emails includes: Get the ID information for each user; Based on each user's ID information, determine the fifth vector corresponding to the target user from the target vector set; The fifth vector is input into the target layer of the first model for processing to obtain the target data information of the target user for processing emails. The first model is used to generate the processing list. The first model is an email processing recommendation model, and the target layer is a preference layer.

2. The method according to claim 1, characterized in that, Based on the target data information of the target user's email processing and Z target emails to be processed, a processing list is generated to recommend processing of the Z target emails to the target user, including: Obtain a target model, wherein the target model is a model obtained by training the first model; The target data information of the target user's emails and the Z target emails to be processed by the target user are input into the output layer of the target model for processing to obtain the processing list.

3. The method according to claim 2, characterized in that, Obtaining the target model includes: Obtain the loss function; The first model is trained using the loss function to obtain the target model.

4. The method according to claim 1, characterized in that, After generating a processing list to recommend processing of the Z target emails to the target user based on the target user's target email data information and the Z target emails to be processed, the method further includes: Obtain feedback information from the target user; Based on the feedback information from the target user, determine the accuracy of recommending the Z target emails to the target user.

5. An email processing apparatus, characterized in that, include: The first acquisition unit is used to acquire N target information, wherein the N target information includes: content information of M emails processed by users in the past, timeliness information of the M emails processed by users in the past, and frequency information of the M emails processed by users in the past, wherein M and N are both positive integers greater than 1; The first determining unit is used to determine the target data information of the target user processing the email based on the N target information, wherein the target user is a user among the M users, and the target data information includes: the timeliness information of the target user processing the email, the frequency information of the target user processing the email, and the content information of the target user processing the email. The first generation unit is used to generate a processing list for recommending the processing of the Z target emails to the target user based on the target data information of the target user's email processing and the Z target emails to be processed, wherein the processing list is used to determine the processing order of the Z target emails, and Z is a positive integer greater than 1. If the N target information consists of the content information, timeliness information, and frequency information of emails processed by M users in the past, the first determining unit comprises: a first processing module for processing the content information of the emails processed by M users in the past to obtain a set of word vectors; a second processing module for encoding the timeliness information of the emails processed by M users in the past to obtain a first vector set; a third processing module for encoding the frequency information of the emails processed by M users in the past to obtain a second vector set; a fourth processing module for obtaining a target vector set based on the word vector set, the first vector set, and the second vector set; and a first determining module for determining the target data information of the emails processed by the target user based on the target vector set. The fourth processing module includes: a first determining submodule, used to determine a first matrix corresponding to each first vector in a first vector set, and to determine a second matrix corresponding to each second vector in a second vector set, wherein the first matrix and the second matrix are weight matrices; a first processing submodule, used to multiply each first vector by the first matrix to obtain a third vector set; a second processing submodule, used to multiply each second vector by the second matrix to obtain a fourth vector set; and a third processing submodule, used to combine the third vector set, the fourth vector set, and the word vector set to obtain a target vector set. The first determining module includes: a first obtaining submodule, used to obtain the ID information of each user; a second determining submodule, used to determine the fifth vector corresponding to the target user from the target vector set based on the ID information of each user; and a fourth processing submodule, used to input the fifth vector into the target layer of the first model for processing to obtain the target data information of the target user's email processing, wherein the first model is used to generate a processing list, the first model is an email processing recommendation model, and the target layer is a preference layer.

6. A computer-readable storage medium, characterized in that, The storage medium stores a program, wherein the program executes the email processing method according to any one of claims 1 to 4.

7. An electronic device, characterized in that, It includes one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the email processing method according to any one of claims 1 to 4.

Images