WO2020253388A1 - 基于机器学习的邮件处理方法、装置、介质及电子设备 - Google Patents

基于机器学习的邮件处理方法、装置、介质及电子设备 Download PDF

Info

Publication number
WO2020253388A1
WO2020253388A1 PCT/CN2020/087470 CN2020087470W WO2020253388A1 WO 2020253388 A1 WO2020253388 A1 WO 2020253388A1 CN 2020087470 W CN2020087470 W CN 2020087470W WO 2020253388 A1 WO2020253388 A1 WO 2020253388A1
Authority
WO
WIPO (PCT)
Prior art keywords
email
unread
machine learning
recipient
mail
Prior art date
Application number
PCT/CN2020/087470
Other languages
English (en)
French (fr)
Inventor
梁锦霞
Original Assignee
深圳壹账通智能科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳壹账通智能科技有限公司 filed Critical 深圳壹账通智能科技有限公司
Publication of WO2020253388A1 publication Critical patent/WO2020253388A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33Querying
    • G06F16/3331Query processing
    • G06F16/334Query execution
    • G06F16/3344Query execution using natural language analysis
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00Handling natural language data
    • G06F40/20Natural language analysis
    • G06F40/279Recognition of textual entities
    • G06F40/289Phrasal analysis, e.g. finite state techniques or chunking
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/045Combinations of networks
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L51/00User-to-user messaging in packet-switching networks, transmitted according to store-and-forward or real-time protocols, e.g. e-mail
    • H04L51/42Mailbox-related aspects, e.g. synchronisation of mailboxes

Definitions

  • This application relates to the field of artificial intelligence machine learning technology, and specifically to a method, device, medium, and electronic equipment for processing mail based on machine learning.
  • senders are generally grouped in the recipient’s mailbox. According to the different types of groups, the senders in different groups are set to accept emails only. Reminder or unread reminder.
  • the inventor realizes that when there is content in the email sent by the sender that requires the recipient to pay special attention and give priority to reply, but the sender is divided into groups that only accept no reminders, the recipient may not be able to read the email in time And miss the best response time.
  • the purpose of the embodiments of this application is to provide a mail processing method, device, medium, and electronic equipment based on machine learning, thereby at least to a certain extent, enabling the recipient to preferentially process the more important mail, and improving the recipient’s mail Processing efficiency, to prevent the recipient from being unable to read important emails in time and missing the best response time, improving the recipient’s email response efficiency.
  • a mail processing method based on machine learning includes:
  • a mail processing device based on machine learning including:
  • the first extraction unit is used to extract priority feature information from the first feature area of unread mail
  • a priority determining unit configured to input the priority feature information into a preset first machine learning model to obtain the priority number of the unread mail;
  • the second extraction unit is configured to extract template feature information from the first feature area of the unread mail whose priority number is greater than a preset threshold;
  • a template generating unit configured to input the template feature information into a second machine learning model to generate an email reply template corresponding to the unread email;
  • the processing unit is configured to obtain a mail reply instruction and display the corresponding mail reply template.
  • a computer program medium having computer-readable instructions stored thereon, and when the computer-readable instructions are executed by the processor of the computer, the computer executes a machine learning-based Mail processing method:
  • the mail processing method based on machine learning includes:
  • an electronic device including:
  • Memory storing computer readable instructions
  • the processor is used to read computer-readable instructions stored in the memory to execute a mail processing method based on machine learning:
  • the mail processing method based on machine learning includes:
  • the priority feature information is extracted from the first feature area of the unread mail, and the priority feature information is input into the preset first machine learning model to obtain The priority number of unread messages, so that the recipient can view unread messages with a higher priority number first, and avoid the recipient being unable to read important messages in time and miss the best response time.
  • an email reply template corresponding to the unread email is generated,
  • the recipient needs to reply to a mail with a higher priority number, by obtaining the mail reply instruction, the corresponding mail reply template is displayed, so that the recipient can quickly input the reply content and improve the mail reply efficiency of the recipient.
  • Fig. 1 schematically shows a flowchart of a mail processing method based on machine learning according to an embodiment of the present application
  • Fig. 2 schematically shows a flowchart of a process of obtaining priority feature information according to an embodiment of the present application
  • FIG. 3 schematically shows a flowchart of a process of obtaining priority feature information according to another embodiment of the present application
  • Fig. 4 schematically shows a flowchart of training a first machine learning model according to an embodiment of the present application
  • FIG. 5 schematically shows a flowchart of training a second machine learning model according to an embodiment of the present application
  • Fig. 6 schematically shows a block diagram of a mail processing device based on machine learning according to an embodiment of the present application
  • FIG. 7 shows a schematic structural diagram of a computer system suitable for implementing an electronic device of an embodiment of the present application
  • Fig. 8 shows a computer-readable storage medium diagram of mail processing based on machine learning according to an embodiment of the present application.
  • Fig. 1 schematically shows a flowchart of a mail processing method based on machine learning according to an embodiment of the present application.
  • the method for processing mail based on machine learning may include the following steps S110 to S150, which are described in detail as follows:
  • step S110 the priority feature information is extracted from the first feature area of the unread mail.
  • the aforementioned unread emails may be unread emails in a web mailbox, or unread emails in a client mailbox.
  • the priority characteristic information may be information used to indicate the importance of the unread mail in the unread mail.
  • the first characteristic area on the unread mail is the area where priority characteristic information is recorded.
  • the first characteristic area may have one or more places. When the first characteristic area of the unread mail has multiple places, the first characteristic area is different from the first characteristic area.
  • the priority feature information types in the feature area may also be different.
  • the first characteristic area of the unread email may include the subject of the unread email, the sender's email address, the sending time, the recipient's email address, and so on.
  • extracting priority feature information from the first feature area of an unread email may include the following steps S210 to S230, which are described in detail as follows:
  • step S210 the first text information in the subject of the unread email, the sender's email address, the sending time, and the recipient's email address are obtained.
  • the first text information obtained in the mail subject of the unread mail may be specific content in the mail subject; the first text obtained in the sender's email address and the recipient's email address
  • the information can be the name of the sender and the name of the recipient; the specific time when the sender sends the mail can be obtained in the sending time.
  • step S220 the first keyword is extracted from the first text information.
  • the first keywords extracted from different first text information are also different. For example: you can extract the item name in the subject of an unread email as the first keyword; or, when the sender remarks "urgent", “important”, and “reply as soon as possible” in the subject of an unread email When waiting for words, these words can be extracted as the first keywords; in addition, when the sender is a more important person, or there are more important people in the recipient, you can add the recipient or the sender The name of the more important person is extracted as the first keyword; in addition, the earlier the unread email is sent, the more the recipient needs to process it as soon as possible, so it can also be obtained from the first text information obtained from the sending time The specific sending time is extracted as the first keyword.
  • step S230 the priority feature information of the unread mail is determined according to the first keyword.
  • the first keyword extracted from the subject of the unread email, the sending time, the email address of the sender, and the email address of the recipient may be used as the priority feature information.
  • extracting priority feature information from the first feature area of an unread email may also include the following steps S310 to S350, which are described in detail as follows:
  • step S310 the characteristic region selection instruction input by the recipient is acquired.
  • the recipient in addition to the subject of the email, the time of sending, the sender’s email address, and the recipient’s email address in the first characteristic area of the unread email, the recipient can also specify the first feature area of the unread email. A characteristic area.
  • different characteristic area selection instructions correspond to different first characteristic areas of the unread mail, and the recipient designates a certain position in the unread mail as the first characteristic area by inputting the corresponding characteristic area selection instruction.
  • step S320 the position of the first characteristic area in the unread mail is determined according to the preset characteristic area selection instruction and the first characteristic area position comparison table.
  • the unread mail designated by the recipient may be determined according to the preset characteristic area selection instruction and the first characteristic area location comparison table The location of the first feature area.
  • the area designated by the recipient can be the subject of the unread email, the time of sending, the sender's email address and the recipient's email address, or an area in the body of the unread email, depending on the recipient It depends on people's needs.
  • step S330 the second text information at the first characteristic region corresponding to the characteristic region selection instruction is acquired.
  • the first feature region corresponding to the feature region selection instruction may be text or pictures.
  • the second text information can be directly obtained;
  • image recognition technology can be used to convert the text in the picture Converted into the second text message.
  • step S340 a second keyword is extracted from the second text information.
  • the second keywords extracted are also different. For example: you can specify the body of an unread email as the preset location. By obtaining the body of an unread email, when the recipient's name is mentioned in the unread email, the recipient's name can be extracted as the second keyword.
  • step S350 the priority feature information of the unread mail is determined according to the second keyword.
  • the second keyword extracted from the second text information at the preset position of the first feature area of the unread mail may be used as the priority feature information.
  • the second keyword may be preprocessed to form priority feature information.
  • the priority feature information may be input into a preset first machine learning model to obtain the priority number of the unread mail.
  • the priority feature information extracted from the first feature area such as the subject of the unread email, the sending time, the sender's email address, and the recipient's email address can be input into the pre-training
  • the priority number of the unread mail is obtained by the machine learning model.
  • the first machine learning model is obtained by training the machine learning model through sample training data.
  • the machine learning model may be a CNN (Convolutional Neural Network, convolutional neural network) model or a deep neural network model.
  • the process of training the first machine learning model includes the following steps:
  • Step S410 Obtain the email subject, sender's email address, sending time, and recipient's email address of the existing reference email.
  • the reference email is an existing email, that is, the existing email can be analyzed in the embodiment of this application to determine the subject of the existing email, the sending time of the existing email, and the existing email The recipient’s email address and the sender’s email address.
  • Step S420 Determine the priority feature information of the existing reference email according to the email subject, sender's email address, sending time, and keywords in the recipient's email address in the reference email, and generate a first training sample.
  • keywords extracted from the subject of the reference email, the sender's email address, the sending time, and the recipient's email address can be used as the first feature vector, and the priority number of the reference email As the second feature vector, training sample data is generated.
  • step S430 the machine learning model is trained through the first training sample data generated to obtain the first machine learning model.
  • the first feature vector may be input to the first machine learning model to obtain the reference email priority number output by the first machine learning model.
  • the priority number of the reference email is inconsistent with the second feature vector Adjust the first machine learning model until the priority number of the output reference mail is consistent with the second feature vector.
  • the priority number of unread mail can be obtained through the first machine learning model.
  • step S130 template feature information is extracted from the second feature area of the unread mail whose priority number is greater than a preset threshold.
  • the priority number of unread emails can be compared with a preset threshold.
  • the priority number of unread emails is When the value is greater than the preset threshold, it indicates that the importance of unread mail is higher, and the recipient needs to process it as soon as possible.
  • the template feature information can be extracted from the second feature area of the unread mail whose priority number is greater than the preset threshold.
  • the second feature area for extracting template feature information in the unread email may include the email subject of the unread email, the sender's email address, the recipient's email address, and the email body, etc.
  • the template feature information may be the item name in the subject of the email, the names of the sender and recipient, the content in the body of the email, and so on.
  • step S140 the template feature information is input into a second machine learning model to generate an email reply template corresponding to the unread email.
  • the recipient can directly reply to the email directly through the email reply template after reading the email, which improves Improve the recipient’s mail processing efficiency.
  • the process of training the second machine learning model includes the following steps:
  • Step S510 Obtain the email subject, sender's email address, recipient's email address, and email body of the existing reference email.
  • the reference email is an existing email, that is, an existing email can be analyzed in this embodiment of the application to determine the subject of the existing email, the recipient's email address, and the existing email.
  • Step S520 Determine the template feature information of the existing reference email according to the email subject, sender's email address and recipient's email address in the reference email, and keywords in the email body, and generate second training sample.
  • keywords extracted from the subject of the reference email, the sender's email address, the recipient's email address, and the body of the email can be used as the third feature vector, and the email reply template of the reference email
  • the second training sample data is generated as the fourth feature vector.
  • step S530 the machine learning model is trained through the generated second training sample data to obtain the second machine learning model.
  • the third feature vector may be input to the second machine learning model to obtain the email reply template of the reference email output by the second machine learning model.
  • the email reply template of the reference email and the fourth feature vector When they are inconsistent, adjust the second machine learning model until the email reply template of the output reference email is consistent with the fourth feature vector.
  • the mail reply template of the unread mail can be obtained through the second machine learning model.
  • step S150 a mail reply instruction is obtained, and the corresponding mail reply template is displayed.
  • the mail reply instruction may be automatically generated after the recipient reads the unread mail at a set time, or it may be generated by the recipient after clicking a certain function button in the mail.
  • training the machine learning model through the generated training sample data, and after obtaining the second machine learning model the method further includes:
  • the email reply template generated by the second machine learning model may not meet the requirements of the recipient.
  • the recipient that is, the user
  • a plurality of unread mails may be sequentially arranged in descending order of the priority number of the unread mails.
  • the priority feature information is extracted from the first feature area of the unread mail, and the priority feature information is input into the preset first machine learning model to obtain The priority number of unread messages, so that the recipient can view unread messages with a higher priority number first, and avoid the recipient being unable to read important messages in time and miss the best response time.
  • Fig. 6 schematically shows a block diagram of a mail processing apparatus based on machine learning according to an embodiment of the present application.
  • a mail processing apparatus 100 based on machine learning includes: a first extraction unit 110, a priority determination unit 120, a second extraction unit 130, a template generation unit 140, and a processing unit 150.
  • the first extraction unit 110 is configured to extract priority feature information from the first feature area of unread mail;
  • the priority determination unit 120 is configured to input the priority feature information into a preset first machine learning model, Obtain the priority number of the unread mail;
  • the second extraction unit 130 is configured to extract template feature information from the second feature area of the unread mail whose priority number is greater than a preset threshold;
  • the template generation unit 140 is configured to The template feature information is input into the second machine learning model to generate a mail reply template corresponding to the unread mail;
  • the processing unit 150 is configured to obtain a mail reply instruction and display the corresponding mail reply template.
  • the first characteristic area includes the subject of the unread email, the sender's email address, the sending time, and the recipient's email address; the first extraction unit 110 is configured to: State the subject of the unread email, the sender’s email address, the sending time, and the first text information in the recipient’s email address; extract the first keyword from the first text information; according to the first key The word determines the priority feature information of the unread mail.
  • the first extraction unit 110 is configured to: obtain a characteristic region selection instruction input by the recipient; and determine the unread according to a preset characteristic region selection instruction and the first characteristic region location comparison table The position of the first characteristic area in the mail; obtain the second text information at the first characteristic area corresponding to the characteristic area selection instruction; extract the second keyword from the second text information; according to the second The keyword determines the priority characteristic information of the unread mail.
  • the first characteristic area includes the subject of the unread email, the sender's email address, the sending time, and the recipient's email address;
  • the machine learning-based mail processing device 100 also includes: a first obtaining unit for obtaining the subject of the existing reference email, the sender's email address, the sending time, and the recipient's email address; the first generating unit is used for obtaining the email subject of the reference email;
  • the key words in the email subject, sender's email address, sending time, and recipient's email address determine the priority feature information of the existing reference email, and generate the first training sample data;
  • the first training unit is used for The machine learning model is trained through the generated first training sample data to obtain the first machine learning model.
  • the second characteristic area includes the subject of the unread email, the sender's email address, the recipient's email address, and the body of the email;
  • the email processing device based on machine learning 100 also includes: a second obtaining unit, used to obtain the subject of the existing reference email, the sender's email address, the recipient's email address, and the email body;
  • the subject of the email, the sender's email address, the recipient's email address, and the keywords in the email body determine the template feature information of the existing reference email, and generate the second training sample data;
  • the second training unit is used to pass The generated second training sample data trains a machine learning model to obtain the second machine learning model.
  • the mail processing device 100 based on machine learning further includes: a third acquiring unit, configured to acquire the recipient’s modification of the email reply template; and a third training unit, configured to modify the modification
  • the latter content is used as training sample data to train the second machine learning model.
  • the machine learning-based mail processing apparatus 100 further includes: a sorting unit configured to sequentially arrange a plurality of the unread mails in descending order of the priority number of the unread mails.
  • the electronic device 600 according to this embodiment of the present application will be described below with reference to FIG. 7.
  • the electronic device 600 shown in FIG. 7 is only an example, and should not bring any limitation to the functions and scope of use of the embodiments of the present application.
  • the electronic device 600 is represented in the form of a general-purpose computing device.
  • the components of the electronic device 600 may include, but are not limited to: the aforementioned at least one processing unit 610, the aforementioned at least one storage unit 620, and a bus 630 connecting different system components (including the storage unit 620 and the processing unit 610).
  • the storage unit stores program code, and the program code can be executed by the processing unit 610, so that the processing unit 610 executes the various exemplary methods described in the "exemplary method" section of this specification.
  • the processing unit 610 may perform step S110 as shown in FIG.
  • step S210 input the priority feature information into a preset first feature area A machine learning model to obtain the priority number of the unread mail; step S310, extract template feature information from the second feature area of the unread mail whose priority number is greater than a preset threshold; step S410, The template feature information is input into the second machine learning model to generate a mail reply template corresponding to the unread mail; step S510, a mail reply instruction is obtained, and the corresponding mail reply template is displayed.
  • the storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 6201 and/or a cache storage unit 6202, and may further include a read-only storage unit (ROM) 6203.
  • RAM random access storage unit
  • ROM read-only storage unit
  • the storage unit 620 may also include a program/utility tool 6204 having a set of (at least one) program module 6205.
  • program module 6205 includes but is not limited to: an operating system, one or more application programs, other program modules, and program data, Each of these examples or some combination may include the implementation of a network environment.
  • the bus 630 may represent one or more of several types of bus structures, including a storage unit bus or a storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local area using any bus structure among multiple bus structures. bus.
  • the electronic device 600 can also communicate with one or more external devices 700 (such as keyboards, pointing devices, Bluetooth devices, etc.), and can also communicate with one or more devices that enable a user to interact with the electronic device 600, and/or communicate with Any device (such as a router, modem, etc.) that enables the electronic device 600 to communicate with one or more other computing devices. This communication can be performed through an input/output (I/O) interface 650.
  • the electronic device 600 may also communicate with one or more networks (for example, a local area network (LAN), a wide area network (WAN), and/or a public network, such as the Internet) through the network adapter 660.
  • networks for example, a local area network (LAN), a wide area network (WAN), and/or a public network, such as the Internet
  • the network adapter 660 communicates with other modules of the electronic device 600 through the bus 630. It should be understood that although not shown in the figure, other hardware and/or software modules can be used in conjunction with the electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives And data backup storage system, etc.
  • the exemplary embodiments described herein can be implemented by software, or can be implemented by combining software with necessary hardware. Therefore, the technical solution according to the embodiments of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (can be CD-ROM, U disk, mobile hard disk, etc.) or on the network , Including several instructions to make a computing device (which may be a personal computer, server, terminal device, or network device, etc.) execute the method according to the embodiment of the present application.
  • a computing device which may be a personal computer, server, terminal device, or network device, etc.
  • a computer-readable storage medium is also provided.
  • the storage medium is a volatile storage medium or a non-volatile storage medium.
  • Program product In some possible implementation manners, each aspect of the present application can also be implemented in the form of a program product, which includes program code. When the program product runs on a terminal device, the program code is used to make the The terminal device executes the steps according to various exemplary embodiments of the present application described in the above-mentioned "Exemplary Method" section of this specification.
  • a program product 800 for implementing the above method according to an embodiment of the present application is described. It can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be stored in a terminal device, For example, running on a personal computer.
  • CD-ROM compact disk read-only memory
  • the program product of this application is not limited to this.
  • the readable storage medium can be any tangible medium that contains or stores a program, and the program can be used by or combined with an instruction execution system, device, or device.
  • the program product can use any combination of one or more readable media.
  • the readable medium may be a readable signal medium or a readable storage medium.
  • the readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or a combination of any of the above. More specific examples (non-exhaustive list) of readable storage media include: electrical connections with one or more wires, portable disks, hard disks, random access memory (RAM), read only memory (ROM), erasable Type programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
  • the computer-readable signal medium may include a data signal propagated in baseband or as a part of a carrier wave, and readable program code is carried therein. This propagated data signal can take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing.
  • the readable signal medium may also be any readable medium other than a readable storage medium, and the readable medium may send, propagate, or transmit a program for use by or in combination with the instruction execution system, apparatus, or device.
  • the program code contained on the readable medium can be transmitted by any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.
  • the program code used to perform the operations of this application can be written in any combination of one or more programming languages.
  • the programming languages include object-oriented programming languages-such as Java, C++, etc., as well as conventional procedural programming languages. Programming language-such as "C" language or similar programming language.
  • the program code can be executed entirely on the user's computing device, partly on the user's device, executed as an independent software package, partly on the user's computing device and partly executed on the remote computing device, or entirely on the remote computing device or server Executed on.
  • the remote computing device can be connected to a user computing device through any kind of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using Internet service providers) Business to connect via the Internet).
  • LAN local area network
  • WAN wide area network
  • Internet service providers Internet service providers

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Evolutionary Computation (AREA)
  • Databases & Information Systems (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

本申请实施例提供了一种基于机器学习的邮件处理方法、装置、介质及电子设备,涉及数据处理技术领域,该方法包括:从未读邮件的第一特征区域中提取出优先级特征信息;将优先级特征信息输入预设的第一机器学习模型,得到未读邮件的优先级数;从优先级数大于预设阈值的未读邮件的第二特征区域中提取出模板特征信息;将模板特征信息输入第二机器学习模型,生成与未读邮件对应的邮件回复模板;获取邮件回复指令,显示对应的邮件回复模板。本申请实施例的技术方案使收件人能够优先处理比较重要的邮件,提高了收件人的邮件处理效率,避免收件人无法及时阅读重要邮件而错过最佳回复时间。

Description

基于机器学习的邮件处理方法、装置、介质及电子设备 技术领域
本申请要求于2019年06月19日提交中国专利局、申请号为201910532896.0,发明名称为“基于机器学习的邮件处理方法、装置、介质及电子设备”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
本申请涉及人工智能的机器学习技术领域,具体而言,涉及一种基于机器学习的邮件处理方法、装置、介质及电子设备。
背景技术
目前,邮箱的邮件收发以及提醒等服务中,一般会在收件人邮箱中先对发件人进行分组,根据分组的不同类型,将不同分组中的发件人发送的邮件设置成只接受不提醒或者未读提醒。发明人意识到,当发件人发送的邮件中存在需要收件人特别关注并优先回复的内容,但发件人被分在只接受不提醒的分组时,收件人可能会无法及时阅读邮件而错过最佳回复时间。
需要说明的是,在上述背景技术部分公开的信息仅用于加强对本申请的背景的理解,因此可以包括不构成对本领域普通技术人员已知的现有技术的信息。
发明概述
技术问题
问题的解决方案
技术解决方案
本申请实施例的目的在于提供一种基于机器学习的邮件处理方法、装置、介质及电子设备,进而至少在一定程度上使收件人能够优先处理比较重要的邮件,提高了收件人的邮件处理效率,避免收件人无法及时阅读重要邮件而错过最佳回复时间,提高收件人的邮件回复效率。
本申请的其他特性和优点将通过下面的详细描述变得显然,或部分地通过本申请的实践而习得。
根据本申请实施例的第一方面,提供了一种基于机器学习的邮件处理方法,所 述方法包括:
从未读邮件的第一特征区域中提取出优先级特征信息;
将所述优先级特征信息输入预设的第一机器学习模型,得到所述未读邮件的优先级数;
从优先级数大于预设阈值的所述未读邮件的第二特征区域中提取出模板特征信息;
将所述模板特征信息输入第二机器学习模型,生成与所述未读邮件对应的邮件回复模板;
获取邮件回复指令,显示对应的所述邮件回复模板。
根据本申请实施例的第二方面,提供了一种基于机器学习的邮件处理装置,包括:
第一提取单元,用于从未读邮件的第一特征区域中提取出优先级特征信息;
优先级确定单元,用于将所述优先级特征信息输入预设的第一机器学习模型,得到所述未读邮件的优先级数;
第二提取单元,用于从优先级数大于预设阈值的所述未读邮件的第一特征区域中提取出模板特征信息;
模板生成单元,用于将所述模板特征信息输入第二机器学习模型,生成与所述未读邮件对应的邮件回复模板;
处理单元,用于获取邮件回复指令,显示对应的所述邮件回复模板。
根据本申请实施例的第三方面,提供了一种计算机程序介质,其上存储有计算机可读指令,当所述计算机可读指令被计算机的处理器执行时,使计算机执行一种基于机器学习的邮件处理方法:
其中,所述基于机器学习的邮件处理方法包括:
从未读邮件的第一特征区域中提取出优先级特征信息;
将所述优先级特征信息输入预设的第一机器学习模型,得到所述未读邮件的优先级数;
从优先级数大于预设阈值的所述未读邮件的第二特征区域中提取出模板特征信息;
将所述模板特征信息输入第二机器学习模型,生成与所述未读邮件对应的邮件回复模板;
获取邮件回复指令,显示对应的所述邮件回复模板。
根据本申请实施例的第四方面,提供了一种电子设备,包括:
存储器,存储有计算机可读指令;
处理器,用于读取存储器存储的计算机可读指令,以执行一种基于机器学习的邮件处理方法:
其中,所述基于机器学习的邮件处理方法包括:
从未读邮件的第一特征区域中提取出优先级特征信息;
将所述优先级特征信息输入预设的第一机器学习模型,得到所述未读邮件的优先级数;
从优先级数大于预设阈值的所述未读邮件的第二特征区域中提取出模板特征信息;
将所述模板特征信息输入第二机器学习模型,生成与所述未读邮件对应的邮件回复模板;
获取邮件回复指令,显示对应的所述邮件回复模板。
发明的有益效果
有益效果
本申请实施例提供的技术方案可以包括以下有益效果:
在本申请的一些实施例所提供的技术方案中,通过从未读邮件的第一特征区域中提取出优先级特征信息,并将该优先级特征信息输入预设的第一机器学习模型,得到未读邮件的优先级数,从而使收件人能够优先查看优先级数较高的未读邮件,避免收件人无法及时阅读重要的邮件而错过最佳回复时间。另外,通过从优先级数大于预设阈值的未读邮件的第一特征区域中提取出模板特征信息,并将模板特征信息输入第二机器学习模型,生成与未读邮件对应的邮件回复模板,当收件人需要对优先级数较高的邮件进行回复时,通过获取邮件回复指令,显示对应的邮件回复模板,使收件人能够快速输入回复内容,提高收件人的邮件回复效率。
应当理解的是,以上的一般描述和后文的细节描述仅是示例性和解释性的,并不能限制本申请。
对附图的简要说明
附图说明
此处的附图被并入说明书中并构成本说明书的一部分,示出了符合本申请的实施例,并与说明书一起用于解释本申请的原理。显而易见地,下面描述中的附图仅仅是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。在附图中:
图1示意性示出了根据本申请的一个实施例的基于机器学习的邮件处理方法的流程图;
图2示意性示出了根据本申请的一个实施例的获取优先级特征信息过程的流程图;
图3示意性示出了根据本申请的另一个实施例的获取优先级特征信息过程的流程图;
图4示意性示出了根据本申请的一个实施例的对第一机器学习模型进行训练的流程图;
图5示意性示出了根据本申请的一个实施例的对第二机器学习模型进行训练的流程图;
图6示意性示出了根据本申请的一个实施例的基于机器学习的邮件处理装置的框图;
图7示出了适于用来实现本申请实施例的电子设备的计算机系统的结构示意图;
图8示出了根据本申请的一个实施例的基于机器学习的邮件处理的计算机可读存储介质图。
发明实施例
本发明的实施方式
图1示意性示出了根据本申请的一个实施例的基于机器学习的邮件处理方法的流程图。
参照图1所示,根据本申请的一个实施例的基于机器学习的邮件处理方法,可以包括如下步骤S110至步骤S150,详细说明如下:
在步骤S110中,从未读邮件的第一特征区域中提取出优先级特征信息。
在本申请的一个实施例中,上述未读邮件可以是网页邮箱中的未读邮件,也可以是客户端邮箱中的未读邮件。优先级特征信息可以是未读邮件中用于表示未读邮件重要程度的信息。
其中,未读邮件上的第一特征区域为记录有优先级特征信息的区域,该第一特征区域可以有一处或多处,当未读邮件的第一特征区域具有多处时,不同第一特征区域内的优先级特征信息类型也可能不相同。可选地,未读邮件的第一特征区域可以包括未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址等等。
在本申请的一个实施例中,如图2所示,从未读邮件的第一特征区域中提取出优先级特征信息可以包括如下步骤S210至步骤S230,详细说明如下:
在步骤S210中,获取所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址中的第一文本信息。
在本申请的一个实施例中,在未读邮件的邮件主题中获取的第一文本信息可以是邮件主题中的具体内容;在发件人邮箱地址和收件人邮箱地址中获取的第一文本信息可以为发件人的姓名和收件人的姓名;发件时间中可以获取发件人发送邮件的具体时间。
在步骤S220中,从所述第一文本信息中提取出第一关键词。
在本申请的一个实施例中,由于第一文本信息的获取区域不同,因此,从不同的第一文本信息中提取出的第一关键词也不同。例如:可以将未读邮件的邮件主题中的项目名称作为第一关键词提取出来;或者,当发件人在未读邮件的邮件主题中备注有“紧急”、“重要”、“尽快回复”等词汇时,可以将这些词汇作为第一关键词提取出来;另外,当发件人为比较重要的人,或者,收件人中存在比较重要的人时,可以将收件人或发件人中比较重要的人的姓名作为第一关键词提取出来;再者,未读邮件的发件时间越早,越需要收件人尽快处理,因而也可以从发件时间中获取的第一文本信息中将具体发送时间作为第一关键词提取 出来。
在步骤S230中,根据所述第一关键词确定所述未读邮件的优先级特征信息。
在本申请的一个实施例中,可以将从未读邮件的邮件主题、发件时间、发件人邮箱地址和收件人邮箱地址中提取出来的第一关键词作为优先级特征信息。
具体地,可以从邮件主题中提取出的“项目名称”或在邮件主题中备注的“紧急”、“重要”、“尽快回复”;将发件人和收件人中比较重要的人的姓名,以及邮件发送的具体时间作为优先级特征信息。
在本申请的另一个实施例中,如图3所示,从未读邮件的第一特征区域提取出优先级特征信息也可以包括如下步骤S310至步骤S350,详细说明如下:
在步骤S310中,获取收件人输入的特征区域选择指令。
在本申请的一个实施例中,未读邮件的第一特征区域除了邮件主题、发件时间、发件人邮箱地址和收件人邮箱地址外,收件人还可以自行指定未读邮件的第一特征区域。
其中,不同的特征区域选择指令对应未读邮件的不同第一特征区域,收件人通过输入相应的特征区域选择指令指定未读邮件中的某个位置作为第一特征区域。
在步骤S320中,根据预设的特征区域选择指令与第一特征区域位置对照表确定所述未读邮件中的第一特征区域的位置。
在本申请的一个实施例中,在获取到收件人输入的特征区域选择指令后,可以按照预设的特征区域选择指令与第一特征区域位置对照表确定收件人所指定的未读邮件的第一特征区域的位置。
其中,收件人指定的区域可以为未读邮件的邮件主题、发件时间、发件人邮箱地址和收件人邮箱地址,也可以为未读邮件的正文某一区域,具体可根据收件人需求而定。
在步骤S330中,获取所述特征区域选择指令对应的第一特征区域处的第二文本信息。
在本申请的一个实施例中,特征区域选择指令对应的第一特征区域处可能是文字,也可能是图片。当特征区域选择指令对应的第一特征区域处为文字时,可 以直接获得第二文本信息;当特征区域选择指令对应的第一特征区域处是图片时,可以采用图像识别技术将图片中的文字转换成第二文本信息。
在步骤S340中,从所述第二文本信息中提取出第二关键词。
在本申请的一个实施例中,由于获取第二文本信息的第一特征区域不同,因此,提取出的第二关键词也不相同。例如:可以指定未读邮件的正文作为预设位置,通过获取未读邮件的正文,当未读邮件中提及收件人姓名时,可以将收件人的姓名作为第二关键词进行提取。
在步骤S350中,根据所述第二关键词确定所述未读邮件的优先级特征信息。
在本申请的一个实施例中,可以将从未读邮件的第一特征区域的预设位置处的第二文本信息中提取的第二关键词作为优先级特征信息。或者,也可以将该第二关键词进行预处理后形成优先级特征信息。
继续参照图1,在步骤S120中,可以将所述优先级特征信息输入预设的第一机器学习模型,得到所述未读邮件的优先级数。
在本申请的一个实施例中,可以将从未读邮件的邮件主题、发件时间、发件人邮箱地址和收件人邮箱地址等第一特征区域提取出的优先级特征信息输入到预先训练好的机器学习模型中,由机器学习模型得到该未读邮件的优先级数。
在本申请的一个实施例中,第一机器学习模型是通过样本训练数据对机器学习模型进行训练得到的。其中,机器学习模型可以是CNN(Convolutional Neural Network,卷积神经网络)模型,或者也可以是深度神经网络模型等。
如图4所示,根据本申请的一个实施例的对第一机器学习模型进行训练的过程,包括如下步骤:
步骤S410,获取已有的参考邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址。
在本申请的一个实施例中,参考邮件即已有的邮件,即本申请实施例中可以对已有邮件进行分析,以确定已有邮件的主题、已有邮件的发件时间、已有邮件的收件人邮箱地址和发件人邮箱地址。
步骤S420,根据所述参考邮件中的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址中的关键词确定所述已有的参考邮件的优先级特征信息,并生成 第一训练样本数据。
在本申请的一个实施例中,可以根据从参考邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址中提取的关键词作为第一特征向量,将参考邮件的优先级数作为第二特征向量来生成训练样本数据。
步骤S430,通过生成的所述第一训练样本数据对机器学习模型进行训练,得到所述第一机器学习模型。
在本申请的一个实施例中,可以将第一特征向量输入第一机器学习模型,得到第一机器学习模型输出的参考邮件优先级数,当该参考邮件的优先级数与第二特征向量不一致时,调整第一机器学习模型,直至输出的参考邮件的优先级数与第二特征向量一致。
由此,当第一机器学习模型训练完成后,可以通过第一机器学习模型得到未读邮件的优先级数。
继续参照图1,在步骤S130中,从优先级数大于预设阈值的所述未读邮件的第二特征区域中提取出模板特征信息。
在本申请的一个实施例中,可以在通过第一机器学习模型得到未读邮件的优先级数后,将未读邮件的优先级数与预设阈值进行比较,当未读邮件的优先级数大于预设阈值时,则说明未读邮件的重要程度较高,需要收件人尽快处理。此时,可以从优先级数大于预设阈值的未读邮件的第二特征区域中提取出模板特征信息。
可选地,未读邮件中用于提取模板特征信息的第二特征区域可以包括未读邮件的邮件主题、发件人邮箱地址、收件人邮箱地址和邮件正文等等。
其中,模板特征信息可以是邮件主题中的项目名称、发件人和收件人姓名、邮件正文中的内容等等。
在步骤S140中,将所述模板特征信息输入第二机器学习模型,生成与所述未读邮件对应的邮件回复模板。
可以理解的是,通过第二机器学习模型自动生成优先级数大于预设阈值的未读邮件的邮件回复模板后,收件人在阅读邮件后,能够直接通过该邮件回复模板快速回复邮件,提高了收件人的邮件处理效率。
如图5所示,根据本申请的一个实施例的对第二机器学习模型进行训练的过程,包括如下步骤:
步骤S510,获取已有的参考邮件的邮件主题、发件人邮箱地址、收件人邮箱地址和邮件正文。
在本申请的一个实施例中,参考邮件即已有的邮件,即本申请实施例中可以对已有邮件进行分析,以确定已有邮件的主题、已有邮件的收件人邮箱地址、已有邮件的发件人邮箱地址和已有邮件的邮件正文。
步骤S520,根据所述参考邮件中的邮件主题、发件人邮箱地址和收件人邮箱地址,以及邮件正文中的关键词确定所述已有的参考邮件的模板特征信息,并生成第二训练样本数据。
在本申请的一个实施例中,可以根据从参考邮件的邮件主题、发件人邮箱地址、收件人邮箱地址和邮件正文中提取的关键词作为第三特征向量,将参考邮件的邮件回复模板作为第四特征向量来生成第二训练样本数据。
步骤S530,通过生成的所述第二训练样本数据对机器学习模型进行训练,得到所述第二机器学习模型。
在本申请的一个实施例中,可以将第三特征向量输入第二机器学习模型,得到第二机器学习模型输出的参考邮件的邮件回复模板,当该参考邮件的邮件回复模板与第四特征向量不一致时,调整第二机器学习模型,直至输出的参考邮件的邮件回复模板与第四特征向量一致。
由此,当第二机器学习模型训练完成后,可以通过第二机器学习模型得到未读邮件的邮件回复模板。
继续参照图1,在步骤S150中,获取邮件回复指令,显示对应的所述邮件回复模板。
在本申请的一个实施例中,邮件回复指令可以是收件人阅读未读邮件后的设定时间自动生成,也可以是收件人通过点击邮件中的某个功能按钮后生成。
在本申请的一个实施例中,通过生成的训练样本数据对机器学习模型进行训练,得到第二机器学习模型之后还包括:
获取收件人对邮件回复模板的修改,并将修改后的内容作为训练样本数据对第 二机器学习模型进行训练。
在本申请的一个实施例中,第二机器学习模型生成的邮件回复模板可能并不符合收件人的要求,此时,可以将收件人(也即用户)对邮件回复模板的修改内容,并将修改的内容作为特征向量继续对第二机器学习模型进行训练,以使第二机器学习模型生成的邮件回复模板符合收件人的个性化要求。
在本申请的一个实施例中,可以按照未读邮件的优先级数的降序依次排列多个未读邮件。
由此,当收件人打开邮箱后,能够按照邮件的重要程度依次阅读邮件,使越重要的邮件越提前处理完成。
在本申请的一些实施例所提供的技术方案中,通过从未读邮件的第一特征区域中提取出优先级特征信息,并将该优先级特征信息输入预设的第一机器学习模型,得到未读邮件的优先级数,从而使收件人能够优先查看优先级数较高的未读邮件,避免收件人无法及时阅读重要的邮件而错过最佳回复时间。另外,通过从优先级数大于预设阈值的未读邮件的第二特征区域中提取出模板特征信息,并将模板特征信息输入第二机器学习模型,生成与未读邮件对应的邮件回复模板,当收件人需要对优先级数较高的邮件进行回复时,通过获取邮件回复指令,显示对应的邮件回复模板,使收件人能够快速输入回复内容,提高收件人的邮件回复效率。
以下介绍本申请的装置实施例,可以用于执行本申请上述的基于机器学习的邮件处理方法。
图6示意性示出了根据本申请的一个实施例的基于机器学习的邮件处理装置的框图。
参照图6所示,根据本申请的一个实施例的基于机器学习的邮件处理装置100,包括:第一提取单元110、优先级确定单元120、第二提取单元130、模板生成单元140和处理单元150。
其中,第一提取单元110用于从未读邮件的第一特征区域中提取出优先级特征信息;优先级确定单元120用于将所述优先级特征信息输入预设的第一机器学习模型,得到所述未读邮件的优先级数;第二提取单元130用于从优先级数大于预 设阈值的所述未读邮件的第二特征区域中提取出模板特征信息;模板生成单元140用于将所述模板特征信息输入第二机器学习模型,生成与所述未读邮件对应的邮件回复模板;处理单元150用于获取邮件回复指令,显示对应的所述邮件回复模板。
在本申请的一个实施例中,所述第一特征区域包括所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;第一提取单元110配置为:获取所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址中的第一文本信息;从所述第一文本信息中提取出第一关键词;根据所述第一关键词确定所述未读邮件的优先级特征信息。
在本申请的另一个实施例中,第一提取单元110配置为:获取收件人输入的特征区域选择指令;根据预设的特征区域选择指令与第一特征区域位置对照表确定所述未读邮件中的第一特征区域的位置;获取所述特征区域选择指令对应的第一特征区域处的第二文本信息;从所述第二文本信息中提取出第二关键词;根据所述第二关键词确定所述未读邮件的优先级特征信息。
在本申请的一个实施例中,所述第一特征区域包括所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;所述的基于机器学习的邮件处理装置100还包括:第一获取单元,用于获取已有的参考邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;第一生成单元,用于根据所述参考邮件中的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址中的关键词确定所述已有的参考邮件的优先级特征信息,并生成第一训练样本数据;第一训练单元,用于通过生成的所述第一训练样本数据对机器学习模型进行训练,得到所述第一机器学习模型。
在本申请的一个实施例中,所述第二特征区域包括所述未读邮件的邮件主题、发件人邮箱地址、收件人邮箱地址和邮件正文;所述的基于机器学习的邮件处理装置100还包括:第二获取单元,用于获取已有的参考邮件的邮件主题、发件人邮箱地址、收件人邮箱地址和邮件正文;第二生成单元,用于根据所述参考邮件中的邮件主题、发件人邮箱地址和收件人邮箱地址及邮件正文中的关键词确定所述已有的参考邮件的模板特征信息,并生成第二训练样本数据;第二训 练单元,用于通过生成的所述第二训练样本数据对机器学习模型进行训练,得到所述第二机器学习模型。
在本申请的一个实施例中,所述的基于机器学习的邮件处理装置100还包括:第三获取单元,用于获取收件人对邮件回复模板的修改;第三训练单元,用于将修改后的内容作为训练样本数据对所述第二机器学习模型进行训练。
在本申请的一个实施例中,所述的基于机器学习的邮件处理装置100还包括:排序单元,用于按照所述未读邮件的优先级数的降序依次排列多个所述未读邮件。
由于本申请的示例实施例的基于机器学习的邮件处理装置的各个功能模块与上述基于机器学习的邮件处理方法的示例实施例的步骤对应,因此对于本申请装置实施例中未披露的细节,请参照本申请上述的页面资源数据的处理方法的实施例。
在本申请的示例性实施例中,还提供了一种能够实现上述方法的电子设备。
所属技术领域的技术人员能够理解,本申请的各个方面可以实现为系统、方法或程序产品。因此,本申请的各个方面可以具体实现为以下形式,即:完全的硬件实施方式、完全的软件实施方式(包括固件、微代码等),或硬件和软件方面结合的实施方式,这里可以统称为“电路”、“模块”或“系统”。
下面参照图7来描述根据本申请的这种实施方式的电子设备600。图7显示的电子设备600仅仅是一个示例,不应对本申请实施例的功能和使用范围带来任何限制。
如图7所示,电子设备600以通用计算设备的形式表现。电子设备600的组件可以包括但不限于:上述至少一个处理单元610、上述至少一个存储单元620、连接不同系统组件(包括存储单元620和处理单元610)的总线630。
其中,所述存储单元存储有程序代码,所述程序代码可以被所述处理单元610执行,使得所述处理单元610执行本说明书上述“示例性方法”部分中描述的根据本申请各种示例性实施方式的步骤。例如,所述处理单元610可以执行如图1中所示步骤S110,从未读邮件的第一特征区域中提取出优先级特征信息;步骤S210,将所述优先级特征信息输入预设的第一机器学习模型,得到所述未读邮件的 优先级数;步骤S310,从优先级数大于预设阈值的所述未读邮件的第二特征区域中提取出模板特征信息;步骤S410,将所述模板特征信息输入第二机器学习模型,生成与所述未读邮件对应的邮件回复模板;步骤S510,获取邮件回复指令,显示对应的所述邮件回复模板。
存储单元620可以包括易失性存储单元形式的可读介质,例如随机存取存储单元(RAM)6201和/或高速缓存存储单元6202,还可以进一步包括只读存储单元(ROM)6203。
存储单元620还可以包括具有一组(至少一个)程序模块6205的程序/实用工具6204,这样的程序模块6205包括但不限于:操作系统、一个或者多个应用程序、其它程序模块以及程序数据,这些示例中的每一个或某种组合中可能包括网络环境的实现。
总线630可以为表示几类总线结构中的一种或多种,包括存储单元总线或者存储单元控制器、外围总线、图形加速端口、处理单元或者使用多种总线结构中的任意总线结构的局域总线。
电子设备600也可以与一个或多个外部设备700(例如键盘、指向设备、蓝牙设备等)通信,还可与一个或者多个使得用户能与该电子设备600交互的设备通信,和/或与使得该电子设备600能与一个或多个其它计算设备进行通信的任何设备(例如路由器、调制解调器等等)通信。这种通信可以通过输入/输出(I/O)接口650进行。并且,电子设备600还可以通过网络适配器660与一个或者多个网络(例如局域网(LAN),广域网(WAN)和/或公共网络,例如因特网)通信。如图所示,网络适配器660通过总线630与电子设备600的其它模块通信。应当明白,尽管图中未示出,可以结合电子设备600使用其它硬件和/或软件模块,包括但不限于:微代码、设备驱动器、冗余处理单元、外部磁盘驱动阵列、RAID系统、磁带驱动器以及数据备份存储系统等。
通过以上的实施方式的描述,本领域的技术人员易于理解,这里描述的示例实施方式可以通过软件实现,也可以通过软件结合必要的硬件的方式来实现。因此,根据本申请实施方式的技术方案可以以软件产品的形式体现出来,该软件产品可以存储在一个非易失性存储介质(可以是CD-ROM,U盘,移动硬盘等) 中或网络上,包括若干指令以使得一台计算设备(可以是个人计算机、服务器、终端装置、或者网络设备等)执行根据本申请实施方式的方法。
在本申请的示例性实施例中,还提供了一种计算机可读存储介质,所述存储介质为易失性存储介质或非易失性存储介质,其上存储有能够实现本说明书上述方法的程序产品。在一些可能的实施方式中,本申请的各个方面还可以实现为一种程序产品的形式,其包括程序代码,当所述程序产品在终端设备上运行时,所述程序代码用于使所述终端设备执行本说明书上述“示例性方法”部分中描述的根据本申请各种示例性实施方式的步骤。
参考图8所示,描述了根据本申请的实施方式的用于实现上述方法的程序产品800,其可以采用便携式紧凑盘只读存储器(CD-ROM)并包括程序代码,并可以在终端设备,例如个人电脑上运行。然而,本申请的程序产品不限于此,在本文件中,可读存储介质可以是任何包含或存储程序的有形介质,该程序可以被指令执行系统、装置或者器件使用或者与其结合使用。
所述程序产品可以采用一个或多个可读介质的任意组合。可读介质可以是可读信号介质或者可读存储介质。可读存储介质例如可以为但不限于电、磁、光、电磁、红外线、或半导体的系统、装置或器件,或者任意以上的组合。可读存储介质的更具体的例子(非穷举的列表)包括:具有一个或多个导线的电连接、便携式盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、光纤、便携式紧凑盘只读存储器(CD-ROM)、光存储器件、磁存储器件、或者上述的任意合适的组合。
计算机可读信号介质可以包括在基带中或者作为载波一部分传播的数据信号,其中承载了可读程序代码。这种传播的数据信号可以采用多种形式,包括但不限于电磁信号、光信号或上述的任意合适的组合。可读信号介质还可以是可读存储介质以外的任何可读介质,该可读介质可以发送、传播或者传输用于由指令执行系统、装置或者器件使用或者与其结合使用的程序。
可读介质上包含的程序代码可以用任何适当的介质传输,包括但不限于无线、有线、光缆、RF等等,或者上述的任意合适的组合。
可以以一种或多种程序设计语言的任意组合来编写用于执行本申请操作的程序 代码,所述程序设计语言包括面向对象的程序设计语言-诸如Java、C++等,还包括常规的过程式程序设计语言-诸如“C”语言或类似的程序设计语言。程序代码可以完全地在用户计算设备上执行、部分地在用户设备上执行、作为一个独立的软件包执行、部分在用户计算设备上部分在远程计算设备上执行、或者完全在远程计算设备或服务器上执行。在涉及远程计算设备的情形中,远程计算设备可以通过任意种类的网络,包括局域网(LAN)或广域网(WAN),连接到用户计算设备,或者,可以连接到外部计算设备(例如利用因特网服务提供商来通过因特网连接)。

Claims (20)

  1. 一种基于机器学习的邮件处理方法,其中,所述方法包括:
    从未读邮件的第一特征区域中提取出优先级特征信息;
    将所述优先级特征信息输入预设的第一机器学习模型,得到所述未读邮件的优先级数;
    从优先级数大于预设阈值的所述未读邮件的第二特征区域中提取出模板特征信息;
    将所述模板特征信息输入第二机器学习模型,生成与所述未读邮件对应的邮件回复模板;
    获取邮件回复指令,显示对应的所述邮件回复模板。
  2. 根据权利要求1所述的方法,其中,所述第一特征区域包括所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;所述从未读邮件的第一特征区域中提取出优先级特征信息包括:
    获取所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址中的第一文本信息;
    从所述第一文本信息中提取出第一关键词;
    根据所述第一关键词确定所述未读邮件的优先级特征信息。
  3. 根据权利要求1所述的方法,其中,所述从未读邮件的第一特征区域中提取出优先级特征信息包括:
    获取收件人输入的特征区域选择指令;
    根据预设的特征区域选择指令与第一特征区域位置对照表确定所述未读邮件中的第一特征区域的位置;
    获取所述特征区域选择指令对应的第一特征区域处的第二文本信息;
    从所述第二文本信息中提取出第二关键词;
    根据所述第二关键词确定所述未读邮件的优先级特征信息。
  4. 根据权利要求1所述的方法,其中,所述第一特征区域包括所述未 读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;
    所述第一机器学习模型通过以下方式得到:
    获取已有的参考邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;
    根据所述参考邮件中的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址中的关键词确定所述已有的参考邮件的优先级特征信息,并生成第一训练样本数据;
    通过生成的所述第一训练样本数据对机器学习模型进行训练,得到所述第一机器学习模型。
  5. 根据权利要求1所述的方法,其中,所述第二特征区域包括所述未读邮件的邮件主题、发件人邮箱地址、收件人邮箱地址和邮件正文;
    所述第二机器学习模型通过以下方法得到:
    获取已有的参考邮件的邮件主题、发件人邮箱地址、收件人邮箱地址和邮件正文;
    根据所述参考邮件中的邮件主题、发件人邮箱地址和收件人邮箱地址及邮件正文中的关键词确定所述已有的参考邮件的模板特征信息,并生成第二训练样本数据;
    通过生成的所述第二训练样本数据对机器学习模型进行训练,得到所述第二机器学习模型。
  6. 根据权利要求5所述的方法,其中,所述通过生成的所述第二训练样本数据对机器学习模型进行训练,得到所述第二机器学习模型之后还包括:
    获取收件人对邮件回复模板的修改;
    将修改后的内容作为训练样本数据对所述第二机器学习模型进行训练。
  7. 根据权利要求1所述的方法,其中,所述方法还包括:
    按照所述未读邮件的优先级数的降序依次排列多个所述未读邮件。
  8. 一种基于机器学习的邮件处理装置,其中,包括:
    第一提取单元,从未读邮件的第一特征区域中提取出优先级特征信息;
    优先级确定单元,用于将所述优先级特征信息输入预设的第一机器学习模型,得到所述未读邮件的优先级数;
    第二提取单元,用于从优先级数大于预设阈值的所述未读邮件的第二特征区域中提取出模板特征信息;
    模板生成单元,用于将所述模板特征信息输入第二机器学习模型,生成与所述未读邮件对应的邮件回复模板;
    处理单元,用于获取邮件回复指令,显示对应的所述邮件回复模板。
  9. 一种计算机程序介质,其上存储有计算机可读指令,当所述计算机可读指令被计算机的处理器执行时,使计算机执行一种基于机器学习的邮件处理方法:
    其中,所述基于机器学习的邮件处理方法包括:
    从未读邮件的第一特征区域中提取出优先级特征信息;
    将所述优先级特征信息输入预设的第一机器学习模型,得到所述未读邮件的优先级数;
    从优先级数大于预设阈值的所述未读邮件的第二特征区域中提取出模板特征信息;
    将所述模板特征信息输入第二机器学习模型,生成与所述未读邮件对应的邮件回复模板;
    获取邮件回复指令,显示对应的所述邮件回复模板。
  10. 根据权利要求9所述的计算机程序介质,其中,所述第一特征区域包括所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;所述从未读邮件的第一特征区域中提取出优先级 特征信息包括:
    获取所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址中的第一文本信息;
    从所述第一文本信息中提取出第一关键词;
    根据所述第一关键词确定所述未读邮件的优先级特征信息。
  11. 根据权利要求9所述的计算机程序介质,其中,所述从未读邮件的第一特征区域中提取出优先级特征信息包括:
    获取收件人输入的特征区域选择指令;
    根据预设的特征区域选择指令与第一特征区域位置对照表确定所述未读邮件中的第一特征区域的位置;
    获取所述特征区域选择指令对应的第一特征区域处的第二文本信息;
    从所述第二文本信息中提取出第二关键词;
    根据所述第二关键词确定所述未读邮件的优先级特征信息。
  12. 根据权利要求9所述的计算机程序介质,其中,所述第一特征区域包括所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;
    所述第一机器学习模型通过以下方式得到:
    获取已有的参考邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;
    根据所述参考邮件中的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址中的关键词确定所述已有的参考邮件的优先级特征信息,并生成第一训练样本数据;
    通过生成的所述第一训练样本数据对机器学习模型进行训练,得到所述第一机器学习模型。
  13. 根据权利要求9所述的计算机程序介质,其中,所述第二特征区域包括所述未读邮件的邮件主题、发件人邮箱地址、收件人邮箱地址和邮件正文;
    所述第二机器学习模型通过以下方法得到:
    获取已有的参考邮件的邮件主题、发件人邮箱地址、收件人邮箱地址和邮件正文;
    根据所述参考邮件中的邮件主题、发件人邮箱地址和收件人邮箱地址及邮件正文中的关键词确定所述已有的参考邮件的模板特征信息,并生成第二训练样本数据;
    通过生成的所述第二训练样本数据对机器学习模型进行训练,得到所述第二机器学习模型。
  14. 根据权利要求13所述的计算机程序介质,其中,所述通过生成的所述第二训练样本数据对机器学习模型进行训练,得到所述第二机器学习模型之后还包括:
    获取收件人对邮件回复模板的修改;
    将修改后的内容作为训练样本数据对所述第二机器学习模型进行训练。
  15. 根据权利要求9所述的计算机程序介质,其中,所述方法还包括:按照所述未读邮件的优先级数的降序依次排列多个所述未读邮件。
  16. 一种电子设备,其中,包括:
    存储器,存储有计算机可读指令;
    处理器,用于读取存储器存储的计算机可读指令,以执行基于机器学习的邮件处理方法:其中,所述基于机器学习的邮件处理方法包括:
    从未读邮件的第一特征区域中提取出优先级特征信息;
    将所述优先级特征信息输入预设的第一机器学习模型,得到所述未读邮件的优先级数;
    从优先级数大于预设阈值的所述未读邮件的第二特征区域中提取出模板特征信息;
    将所述模板特征信息输入第二机器学习模型,生成与所述未读邮 件对应的邮件回复模板;
    获取邮件回复指令,显示对应的所述邮件回复模板。
  17. 根据权利要求16所述的电子设备,其中,所述第一特征区域包括所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;所述从未读邮件的第一特征区域中提取出优先级特征信息包括:
    获取所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址中的第一文本信息;
    从所述第一文本信息中提取出第一关键词;
    根据所述第一关键词确定所述未读邮件的优先级特征信息。
  18. 根据权利要求16所述的电子设备,其中,所述从未读邮件的第一特征区域中提取出优先级特征信息包括:
    获取收件人输入的特征区域选择指令;
    根据预设的特征区域选择指令与第一特征区域位置对照表确定所述未读邮件中的第一特征区域的位置;
    获取所述特征区域选择指令对应的第一特征区域处的第二文本信息;
    从所述第二文本信息中提取出第二关键词;
    根据所述第二关键词确定所述未读邮件的优先级特征信息。
  19. 根据权利要求16所述的电子设备,其中,所述第一特征区域包括所述未读邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;
    所述第一机器学习模型通过以下方式得到:
    获取已有的参考邮件的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址;
    根据所述参考邮件中的邮件主题、发件人邮箱地址、发送时间和收件人邮箱地址中的关键词确定所述已有的参考邮件的优先级特征信息,并生成第一训练样本数据;
    通过生成的所述第一训练样本数据对机器学习模型进行训练,得到所述第一机器学习模型。
  20. 根据权利要求16所述的电子设备,其中,所述第二特征区域包括所述未读邮件的邮件主题、发件人邮箱地址、收件人邮箱地址和邮件正文;
    所述第二机器学习模型通过以下方法得到:
    获取已有的参考邮件的邮件主题、发件人邮箱地址、收件人邮箱地址和邮件正文;
    根据所述参考邮件中的邮件主题、发件人邮箱地址和收件人邮箱地址及邮件正文中的关键词确定所述已有的参考邮件的模板特征信息,并生成第二训练样本数据;
    通过生成的所述第二训练样本数据对机器学习模型进行训练,得到所述第二机器学习模型。
PCT/CN2020/087470 2019-06-19 2020-04-28 基于机器学习的邮件处理方法、装置、介质及电子设备 WO2020253388A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910532896.0 2019-06-19
CN201910532896.0A CN110348009A (zh) 2019-06-19 2019-06-19 基于机器学习的邮件处理方法、装置、介质及电子设备

Publications (1)

Publication Number Publication Date
WO2020253388A1 true WO2020253388A1 (zh) 2020-12-24

Family

ID=68182485

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/087470 WO2020253388A1 (zh) 2019-06-19 2020-04-28 基于机器学习的邮件处理方法、装置、介质及电子设备

Country Status (2)

Country Link
CN (1) CN110348009A (zh)
WO (1) WO2020253388A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113408281A (zh) * 2021-07-14 2021-09-17 北京天融信网络安全技术有限公司 邮箱账号异常检测方法、装置、电子设备及存储介质
CN113746814A (zh) * 2021-08-17 2021-12-03 上海硬通网络科技有限公司 邮件处理方法、装置、电子设备及存储介质
CN114970451A (zh) * 2022-04-13 2022-08-30 中国银行股份有限公司 一种邮件的标记方法、装置、设备及可读存储介质
CN116545976A (zh) * 2023-07-03 2023-08-04 深圳欧税通技术有限公司 业务邮件智能管理方法、系统和可读存储介质

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110348009A (zh) * 2019-06-19 2019-10-18 深圳壹账通智能科技有限公司 基于机器学习的邮件处理方法、装置、介质及电子设备
CN112784635A (zh) * 2019-11-07 2021-05-11 富士通株式会社 信息处理装置和方法以及利用模型分类的装置
CN112818109B (zh) * 2021-02-25 2022-09-16 网易(杭州)网络有限公司 邮件的智能回复方法、介质、装置和计算设备
CN113343682A (zh) * 2021-06-07 2021-09-03 中国工商银行股份有限公司 邮件处理方法、邮件处理装置、电子设备和存储介质
CN113592461A (zh) * 2021-08-10 2021-11-02 平安普惠企业管理有限公司 邮件处理方法、装置与存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6816885B1 (en) * 2000-09-21 2004-11-09 International Business Machines Corporation Method and system to handle large volume of E-mail received from a plurality of senders intelligently
US7406504B2 (en) * 2003-09-18 2008-07-29 Sbc Knowledge Ventures, L.P. Intelligent email detection and auto reply email technique to emails destined to no reply email addresses
US20150200899A1 (en) * 2013-01-29 2015-07-16 Google Inc. Automatic email reminder system based on machine learning
CN107147779A (zh) * 2017-04-21 2017-09-08 北京小米移动软件有限公司 信息提示方法及装置
US20180219817A1 (en) * 2017-01-30 2018-08-02 Futurewei Technologies, Inc. Personalized message priority classification
CN110348009A (zh) * 2019-06-19 2019-10-18 深圳壹账通智能科技有限公司 基于机器学习的邮件处理方法、装置、介质及电子设备

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6816885B1 (en) * 2000-09-21 2004-11-09 International Business Machines Corporation Method and system to handle large volume of E-mail received from a plurality of senders intelligently
US7406504B2 (en) * 2003-09-18 2008-07-29 Sbc Knowledge Ventures, L.P. Intelligent email detection and auto reply email technique to emails destined to no reply email addresses
US20150200899A1 (en) * 2013-01-29 2015-07-16 Google Inc. Automatic email reminder system based on machine learning
US20180219817A1 (en) * 2017-01-30 2018-08-02 Futurewei Technologies, Inc. Personalized message priority classification
CN107147779A (zh) * 2017-04-21 2017-09-08 北京小米移动软件有限公司 信息提示方法及装置
CN110348009A (zh) * 2019-06-19 2019-10-18 深圳壹账通智能科技有限公司 基于机器学习的邮件处理方法、装置、介质及电子设备

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113408281A (zh) * 2021-07-14 2021-09-17 北京天融信网络安全技术有限公司 邮箱账号异常检测方法、装置、电子设备及存储介质
CN113408281B (zh) * 2021-07-14 2024-02-09 北京天融信网络安全技术有限公司 邮箱账号异常检测方法、装置、电子设备及存储介质
CN113746814A (zh) * 2021-08-17 2021-12-03 上海硬通网络科技有限公司 邮件处理方法、装置、电子设备及存储介质
CN113746814B (zh) * 2021-08-17 2024-01-09 上海硬通网络科技有限公司 邮件处理方法、装置、电子设备及存储介质
CN114970451A (zh) * 2022-04-13 2022-08-30 中国银行股份有限公司 一种邮件的标记方法、装置、设备及可读存储介质
CN116545976A (zh) * 2023-07-03 2023-08-04 深圳欧税通技术有限公司 业务邮件智能管理方法、系统和可读存储介质
CN116545976B (zh) * 2023-07-03 2024-01-19 深圳欧税通技术有限公司 业务邮件智能管理方法、系统和可读存储介质

Also Published As

Publication number Publication date
CN110348009A (zh) 2019-10-18

Similar Documents

Publication Publication Date Title
WO2020253388A1 (zh) 基于机器学习的邮件处理方法、装置、介质及电子设备
US11134046B2 (en) Apparatus and method for maintaining a message thread with opt-in permanence for entries
JP6969653B2 (ja) 電子会議システム
JP6866860B2 (ja) 電子会議システム
US10873545B2 (en) Automatic action responses
US10205695B2 (en) Systems and methods for implementing intelligent chat communication within an email environment
US9414779B2 (en) Electronic communication warning and modification
US11086598B2 (en) Providing a communications channel between instances of automated assistants
US11558334B2 (en) Multi-message conversation summaries and annotations
US7818373B2 (en) Notifying co-recipients of others currently replying to communications
CN111742337A (zh) 使用机器学习模型的消息分析
US11943310B2 (en) Performing operations based upon activity patterns
US11418463B2 (en) Method and system of intelligently providing responses for a user in the user's absence
WO2022213943A1 (zh) 消息发送方法、消息发送装置、电子设备和存储介质
CN117971154A (zh) 多模态响应
US11656844B2 (en) Providing a communications channel between instances of automated assistants
US8768846B2 (en) System, method, and apparatus for management of media objects
CN108491471A (zh) 一种文本信息的处理方法、移动终端
JP5427669B2 (ja) サーバ装置、情報管理システムおよびプログラム
US11689482B2 (en) Dynamically generating a typing feedback indicator for recipient to provide context of message to be received by recipient
CN114124864B (zh) 消息处理方法、装置
CN117873619A (zh) 产品说明文档的生成方法、装置、存储介质及终端设备
CN116348922A (zh) 基于用户熟悉度终止执行图像分类

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20827924

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20827924

Country of ref document: EP

Kind code of ref document: A1