Message processing methods, apparatus, storage media, computer equipment and software products
By displaying messages to be replied to in the message interface and combining intelligent and manual replies, the problem of low message reply efficiency in existing technologies is solved, and the ability to quickly generate reply messages that meet user needs is achieved.
Patent Information
- Application Number
- CN202510622362.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-14
- Publication Date
- 2026-11-17
AI Technical Summary
In existing technologies, computer devices require the relevant parties to input the reply message word by word when responding to a message, resulting in low processing efficiency.
The message interface displays the message to be replied to, and in response to the smart reply trigger operation, it enters the smart reply mode. Multiple first message units are displayed in sequence in the message editing area, allowing users to control the processing of the smart reply model through editing operations, and combine manual reply to generate the final reply message.
By combining intelligent and manual replies, it is possible to quickly generate response messages that meet user needs and improve message processing efficiency.
Smart Images

Figure CN123114346A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of computers, and in particular to a message processing method, apparatus, storage medium, computer equipment, and program product. Background Technology
[0002] With the rapid development of computer technology, various communication software has become an important part of daily life. By using communication software to send messages (such as emails, instant messages, or text messages), information can be transmitted between different devices.
[0003] However, when computer devices respond to received messages, they typically rely on manual intervention from the relevant parties to obtain a reply. This means the parties need to type the reply word by word and adjust the wording according to specific reply requirements. Therefore, this approach is time-consuming and inefficient. Summary of the Invention
[0004] This application provides a message processing method, apparatus, storage medium, computer equipment, and program product that can combine intelligent and manual replies to improve message processing efficiency.
[0005] To solve the above-mentioned technical problems, this application provides the following technical solution: According to one aspect of this application, a message processing method is provided, comprising: Display pending messages in the message interface; In response to the intelligent reply trigger operation of the message to be replied to, a message editing area is displayed in the message interface; In the message editing area, multiple first message units are displayed sequentially, wherein the multiple first message units are obtained sequentially by processing the message to be replied to through an intelligent reply model; During the sequential display of the plurality of first message units, in response to the first editing operation on the plurality of first message units, the intelligent reply model is controlled to stop processing the message to be replied to; The reply message for the message to be replied to is determined based on the first editing operation.
[0006] In the above scheme, before displaying the message to be replied to in the message interface, the message processing method further includes: A first retrieval request is triggered based on the message source to retrieve a portion of the message to be replied to, wherein the portion of the message to be replied to is used to display in the message interface; Before sequentially displaying multiple first message units in the message editing area, the message processing method further includes: A second retrieval request is triggered based on the message source to obtain the full content of the message to be replied to, wherein the full content of the message to be replied to is used for processing by the intelligent reply model.
[0007] According to one aspect of this application, a message processing apparatus is provided, comprising: The first display module is used to display messages awaiting reply in the message interface; The second display module is used to display a message editing area in the message interface in response to the intelligent reply trigger operation of the message to be replied to; The third display module is used to sequentially display multiple first message units in the message editing area, wherein the multiple first message units are obtained sequentially by processing the message to be replied to through an intelligent reply model; The control module is used to control the intelligent reply model to stop processing the message to be replied to in response to a first editing operation on the plurality of first message units during the sequential display of the plurality of first message units. The first determining module is used to determine a reply message for the message to be replied to based on the first editing operation.
[0008] In the above scheme, the first determining module is further configured to: In the message editing area, a first candidate reply message is displayed, wherein the first candidate reply message is obtained by editing the plurality of first message units according to the first editing operation; Based on the first candidate reply message, a reply message is generated for the message to be replied to.
[0009] In the above scheme, the first determining module is further configured to: In response to the intelligent reply triggering operation on the message to be replied to and the first candidate reply message, a plurality of second message units are displayed sequentially after the first candidate reply message displayed in the message editing area. The plurality of second message units are obtained sequentially by processing the message to be replied to and the first candidate reply message through the intelligent reply model. Based on the first candidate reply message and the plurality of second message units, a reply message is generated for the message to be replied to.
[0010] In the above scheme, the message processing device further includes a fourth display module, used for: In response to an edit trigger operation on the message editing area, a message editing control is displayed on the message interface; The message editing control is used to receive a first editing operation on the plurality of first message units.
[0011] In the above scheme, the third display module is further used for: In the message editing area, multiple first message units are displayed sequentially until the intelligent reply model finishes processing the message to be replied to; The message processing device further includes a second determining module, used for: In response to a second edit operation on the plurality of first message units, a reply message for the message to be replied to is determined based on the second edit operation.
[0012] In the above scheme, the second determining module is further configured to: In the message editing area, a second candidate reply message is displayed, wherein the second candidate reply message is obtained by editing the plurality of first message units according to the second editing operation; Based on the second candidate reply message, a reply message is generated for the message to be replied to.
[0013] In the above scheme, the message processing device further includes a fifth display module, used for: The message editing control is displayed in the message interface; The message editing control is used to receive a second editing operation on the plurality of first message units.
[0014] In the above scheme, the message processing device further includes a third determining module, used for: In response to a manual reply to the message to be replied to, a message editing area is displayed in the message interface; In response to the third editing operation, a third candidate reply message edited according to the third editing operation is displayed in the message editing area; In response to the intelligent reply triggering operation on the message to be replied to and the third candidate reply message, a plurality of third message units are displayed sequentially after the third candidate reply message displayed in the message editing area. The plurality of third message units are obtained sequentially by processing the message to be replied to and the third candidate reply message through the intelligent reply model. A reply message is generated for the message to be replied to based on the third candidate reply message and the plurality of third message units.
[0015] In the above scheme, the message processing device further includes a sliding display module, used for: In response to a swipe operation on the message interface, the message to be displayed in the message interface is swiped to show the message in the direction indicated by the swipe operation. The message to be displayed includes the plurality of first message units and the message to be replied to.
[0016] In the above scheme, the message processing device further includes a style determination module, used for: A smart reply style is determined for the message to be replied to, so that the smart reply model processes the message to be replied to according to the smart reply style, thereby obtaining the plurality of first message units in sequence.
[0017] In the above scheme, the style determination module is also used for: The message interface displays multiple smart reply styles, each with its own options. In response to a selection operation of these options, the smart reply style corresponding to the selected option is determined as the smart reply style for the message to be replied to; or... Analyze the response styles of multiple historical response messages to obtain an intelligent response style for the message to be responded to.
[0018] In the above scheme, the message processing device further includes an extraction module, used for: The message to be replied to is processed by extracting text to obtain the text to be processed in the message to be replied to; The text to be processed in the message to be replied to is processed by the intelligent reply model to obtain the plurality of first message units in sequence.
[0019] In the above scheme, the message to be replied to includes the text to be replied to and the image to be replied to. The extraction module is also used for: The image to be replied to is processed by text recognition to obtain the text in the image to be replied to; The text to be replied to is merged with the text in the image to be replied to, resulting in the text to be processed.
[0020] In the above scheme, the extraction module is also used for: Determine the number of characters in the text of the image to be replied to; When the number of characters exceeds the threshold, the text to be replied to is merged with the text in the image to be replied to to obtain the text to be processed. The extraction module is also used for: When the number of characters is less than or equal to the number threshold, the text to be replied to is determined as text to be processed.
[0021] In the above scheme, the message processing device further includes a compression module, used for: The message to be replied to is split into multiple sub-messages; The multiple sub-messages are compressed separately to obtain compressed sub-messages corresponding to the multiple sub-messages; The compressed sub-messages corresponding to the multiple sub-messages are merged to obtain a merged message; The merged message is processed by the intelligent reply model to obtain the plurality of first message units in sequence.
[0022] In the above scheme, the compression module is also used for: Determine the amount of information in the message to be replied to; When the amount of information exceeds the information threshold, the message to be replied to is split into multiple sub-messages.
[0023] In the above scheme, the message processing device further includes a first request module, used for: A first retrieval request is triggered based on the message source to retrieve a portion of the message to be replied to, wherein the portion of the message to be replied to is used to display in the message interface; The message processing device further includes a second request module, used for: A second retrieval request is triggered based on the message source to obtain the full content of the message to be replied to, wherein the full content of the message to be replied to is used for processing by the intelligent reply model.
[0024] According to one aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program adapted for loading by a processor to perform the above-described message processing method.
[0025] According to one aspect of this application, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described message processing method.
[0026] According to one aspect of this application, a computer program product is provided, the computer program product comprising a computer program stored in a storage medium. A processor of a computer device reads the computer program from the storage medium, and the processor executes the computer program to implement the above-described message processing method.
[0027] In related technologies, message replies require the user to manually input the message word by word and adjust the wording according to specific reply needs. This results in a significant time consumption to obtain a reply, leading to low processing efficiency. To address this, this application displays the message to be replied to in the message interface. Then, in response to the intelligent reply trigger operation of the message to be replied to, it enters the intelligent reply mode. Specifically, in the intelligent reply mode, a message editing area is displayed in the message interface, and multiple first message units are displayed sequentially in the message editing area. These multiple first message units are obtained sequentially by the intelligent reply model processing the message to be replied to. This improves the real-time display and avoids blocking the waiting for all first message units to be obtained before displaying them. Simultaneously, during the sequential display of multiple first message units, in response to the first editing operation on the multiple first message units, the intelligent reply model stops processing the message to be replied to, allowing the user to intervene at any time during the sequential display of multiple first message units to ensure that their needs are met. Finally, the reply message for the message to be replied to is determined based on the first editing operation, achieving a combination of manual and intelligent reply. This application combines manual and intelligent responses to obtain reply messages that better meet the needs of the relevant recipients more quickly, thereby improving message processing efficiency.
[0028] Other features and advantages of this application will be set forth in the following description and will be apparent in part from the description or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the description, claims and drawings. Attached Figure Description
[0029] The accompanying drawings are used to provide a further understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application. Figure 1 This is a schematic diagram of the architecture of the message processing system provided in an embodiment of this application; Figure 2 This is a schematic diagram of an interface illustrating the message processing method provided in this application embodiment applied to a short message reply scenario; Figure 3 This is a schematic diagram of an interface illustrating the message processing method provided in this application embodiment applied to an email reply scenario; Figure 4 This is a schematic diagram of an interface illustrating the message processing method provided in this application embodiment applied to an instant messaging message reply scenario; Figure 5 This is a flowchart illustrating a message processing method provided in an embodiment of this application; Figure 6 This is a schematic diagram of an interface of the message processing method provided in an embodiment of this application; Figure 7 This is another schematic diagram of the message processing method provided in the embodiments of this application; Figure 8 This is another schematic diagram of the message processing method provided in the embodiments of this application; Figure 9 This is another schematic diagram of the interface of the message processing method provided in the embodiments of this application; Figure 10 This is another schematic diagram of the message processing method provided in the embodiments of this application; Figure 11 This is another schematic diagram of the message processing method provided in the embodiments of this application; Figure 12 This is another schematic diagram of the interface of the message processing method provided in the embodiments of this application; Figure 13 This is another schematic diagram of the interface of the message processing method provided in the embodiments of this application; Figure 14 This is another schematic diagram of the interface of the message processing method provided in the embodiments of this application; Figure 15 This is another schematic diagram of the interface of the message processing method provided in the embodiments of this application; Figure 16 This is a schematic diagram of a text extraction process provided in an embodiment of this application; Figure 17 This is another schematic diagram of the text extraction process provided in the embodiments of this application; Figure 18 This is a flowchart illustrating the splitting, compression, and merging of messages to be replied to, provided in an embodiment of this application. Figure 19 This is another flowchart illustrating the splitting, compression, and merging of messages to be replied to, provided in an embodiment of this application. Figure 20 This is another schematic diagram of the interface of the message processing method provided in the embodiments of this application; Figure 21 This is another flowchart illustrating the message processing method provided in the embodiments of this application; Figure 22 This is a flowchart illustrating the process of splitting, compressing, and merging text to be processed, provided in an embodiment of this application. Figure 23 This is a schematic diagram of the message processing apparatus provided in an embodiment of this application; Figure 24This is a schematic diagram of the structure of a terminal provided in an embodiment of this application; Figure 25 This is a schematic diagram of the structure of a server provided in an embodiment of this application. Detailed Implementation
[0030] To enable those skilled in the art to better understand the solutions of this application, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0031] It should be noted that while some processes described in the specification, claims, and accompanying drawings include multiple steps appearing in a specific order, it should be clearly understood that these steps may not be performed in the order they appear herein, or may be performed in parallel. The step numbers are merely used to distinguish different steps and do not themselves represent any execution order. Furthermore, descriptions such as "first," "second," or "objective" in this document are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. "Multiple" in this document refers to at least two.
[0032] It is worth noting that in the specific embodiments of this application, messages and other related data are involved. When the above embodiments of this application are applied to specific products or technologies, permission or consent from the target is required, and the collection, use, and processing of related data must comply with relevant laws, regulations, and standards. For example, when an embodiment of this application needs to obtain message data, separate permission or consent from the target can be obtained through pop-ups or redirection to a confirmation page. After obtaining the target's separate permission or consent, the necessary message data for enabling the embodiment of this application to operate normally can then be obtained.
[0033] Before providing a further detailed description of the embodiments of this application, the nouns and terms used in the embodiments of this application are explained, and the nouns and terms used in the embodiments of this application shall be interpreted as follows: 1) Message: Content transmitted through various communication methods (such as written communication, email, instant messaging software, etc.). Messages may include text, sound, or images, and may cover various topics, such as notifications, warnings, and suggestions. This application embodiment can process various types of messages transmitted through online communication methods (such as email, instant messaging software, or SMS).
[0034] 2) Intelligent Response Model: A model that can automatically analyze and understand the semantics of input information based on various algorithms, technologies, and logical processing mechanisms, and generate corresponding responses accordingly. For example, an intelligent response model could be a Large Language Model (MLM). A MLM is an artificial intelligence model based on deep learning technology, possessing a large number of parameters and powerful language understanding and generation capabilities. Through unsupervised or semi-supervised learning on massive amounts of data, it automatically captures grammatical, semantic, and pragmatic patterns in language, learns rich language knowledge, and can effectively achieve intelligent responses.
[0035] For message replies, the solutions provided by relevant technologies usually require the relevant parties to input the reply message word by word, and also require them to adjust the expression according to the specific reply requirements. This requires the relevant parties to spend a lot of time processing the messages, resulting in low processing efficiency.
[0036] To address this, this application proposes a message processing method, apparatus, storage medium, computer device, and program product. It proposes displaying a message to be replied to in a message interface and, in response to a smart reply trigger operation on the message to be replied to, displaying a message editing area in the message interface. Then, in the message editing area, multiple first message units obtained by the smart reply model after processing the message to be replied to are displayed sequentially. This improves the real-time performance of the display. Simultaneously, during the sequential display of multiple first message units, in response to a first edit operation on the multiple first message units, the smart reply model is controlled to stop processing the message to be replied to, allowing relevant parties to intervene at any time during the sequential display of multiple first message units. Finally, a reply message is determined based on the first edit operation. Compared to relevant parties manually inputting reply messages, this application's embodiment, through a combination of smart and manual replies, obtains reply messages that better meet the reply needs of relevant parties more quickly, thus improving message processing efficiency.
[0037] Please see Figure 1 , Figure 1 This is a schematic diagram of a message processing system provided in an embodiment of this application. It includes a terminal 140, an Internet 130, a gateway 120, a server 110, etc.
[0038] Terminal 140 can take various forms, including desktop computers, laptops, PDAs (personal digital assistants), mobile phones, vehicle terminals, and dedicated terminals. Furthermore, it can be a single device or a collection of multiple devices. For example, multiple desktop computers can be interconnected via a local area network, sharing a single monitor to work collaboratively, forming a single terminal 140. Terminal 140 can communicate with the Internet 130 via wired or wireless means to exchange data.
[0039] Server 110 refers to a computer system capable of providing certain services to terminal 140. Compared to ordinary terminal 140, server 110 has higher requirements in terms of stability, security, and performance. Server 110 can be a single high-performance computer in a network platform, a cluster of multiple high-performance computers, a portion of a single high-performance computer (e.g., a virtual machine), or a combination of portions of multiple high-performance computers (e.g., virtual machines). Server 110 can also communicate with the Internet 130 via wired or wireless means to exchange data.
[0040] Gateway 120, also known as an internetwork connector or protocol converter, is a computer system or device that acts as a translator, enabling network interconnection at the transport layer. It bridges the gap between two systems using different communication protocols, data formats, languages, or even completely different architectures. Gateways can also provide filtering and security functions. Messages sent from terminal 140 to server 110 are forwarded to the corresponding server 110 via gateway 120. Messages sent from server 110 to terminal 140 are also forwarded to the corresponding terminal 140 via gateway 120.
[0041] This embodiment can be implemented by terminal 140. In this architecture, the intelligent reply model can be deployed locally on terminal 140. When the intelligent reply model is deployed locally on terminal 140, since cross-terminal communication is not required during reply message generation, latency in the reply message generation process can be reduced, improving real-time response speed. Furthermore, it reduces reliance on network connectivity; for example, terminal 140 can generate reply messages offline and send the generated reply messages when connected to the network.
[0042] Alternatively, this embodiment can be implemented jointly by terminal 140 and server 110. In this architecture, the intelligent reply model can be deployed on server 110. When the intelligent reply model is deployed on server 110, since server 110 is equipped with more powerful computing resources, the efficiency of reply message generation can be improved, especially when the message to be replied to is relatively complex. At the same time, the computing resources occupied by terminal 140 can be reduced, allowing terminal 140 to avoid consuming too many computing resources to perform message processing, making this embodiment applicable to lightweight terminals with poor performance (such as mobile phones).
[0043] The embodiments of this application can be applied in a variety of scenarios. Examples of these scenarios will be provided below.
[0044] (a) SMS reply scenario As an example, the following are provided: Figure 2 The message interface of terminal 140 is shown. The message interface of terminal 140 displays, as follows: Figure 2 The diagram shows SMS 1, SMS 2, and SMS 3. When terminal 140 receives a smart reply trigger operation for a message to be replied to in the message interface, it can process the message to be replied to using the smart reply model, thereby obtaining multiple first message units in sequence. Taking SMS 2 as an example, when a smart reply trigger operation is received for SMS 2, a message editing area can be displayed in the message interface. Simultaneously, SMS 2 is processed by the smart reply model deployed on terminal 140 to obtain multiple first message units in sequence, as shown... Figure 2 (1) and Figure 2 As shown in (2), multiple first message units are displayed sequentially in the message editing area to achieve real-time display of multiple first message units. Thus, during the sequential display of multiple first message units, if a first editing operation is received on multiple first message units, the intelligent reply model can be controlled to stop processing the message to be replied to in response to the first editing operation, and the reply message for the message to be replied to can be determined according to the first editing operation.
[0045] The above method can generate corresponding reply messages for short messages in the message interface, reducing the time spent by relevant parties in processing short messages, and realizing a combination of automatic and manual processing of short messages, thereby improving the processing efficiency of short messages.
[0046] (ii) Email reply scenario For example, server 110 is a backend server for an email platform, and terminal 140 runs a client for the email platform. Terminal 140 can obtain relevant data for messages awaiting reply from server 110 and display the messages in the message interface. Figure 3 As shown, the messages displayed in the message interface are emails, specifically including the email body and email attachments. For email attachments, only the file name, attachment size, and other data can be displayed to keep the message interface concise.
[0047] Then, in response to the intelligent reply trigger operation of the message to be replied to, terminal 140 can display a message editing area in the message interface. The message editing area includes, for example: Figure 3 The diagram shows a first sub-area 310, a second sub-area 320, and a third sub-area 330. The first sub-area displays recipient information for the reply message, the second sub-area displays the subject of the reply message, and the third sub-area sequentially displays multiple first message units and the reply message for the message to be replied to. Simultaneously, terminal 140 sends the message to be replied to to server 110. Server 110 processes the message using an intelligent reply model, sequentially obtaining multiple first message units corresponding to the message to be replied to. Each time server 110 obtains a first message unit, it sends it to terminal 140, allowing terminal 140 to display it in the message editing area. Subsequently, the relevant party can edit the multiple first message units in the message editing area, allowing terminal 140 to determine the reply message for the message to be replied to based on the editing operation (i.e., the first editing operation).
[0048] The above method enables the generation of corresponding reply messages for emails, improving email processing efficiency. Furthermore, since the intelligent reply model processes the content of both the email body and attachments, it can improve the accuracy of the multiple first message units generated by the intelligent reply model.
[0049] (III) Instant Messaging Message Reply Scenarios For example, server 110 is a backend server of an instant messaging platform, and terminal 140 runs a client of the instant messaging platform. At least one instant messaging message is displayed in the message interface of terminal 140. In response to a selection operation on an instant messaging message, terminal 140 can designate the instant messaging message indicated by the selection operation as a message to be replied to. When terminal 140 receives a smart reply trigger operation (such as a smart reply option being triggered) for the message to be replied to in the message interface, terminal 140 can display a message editing area in the message interface. Simultaneously, terminal 140 sends the message to be replied to to server 110. Server 110 processes the message to be replied to using a smart reply model, sequentially obtaining multiple first message units corresponding to the message to be replied to. Each time server 110 obtains a first message unit, it sends it to terminal 140 so that terminal 140 can display it in the message editing area. Terminal 140 can receive a first edit operation on the multiple first message units, and in response to the first edit operation, controls the smart reply model to stop processing the message to be replied to, and determines the reply message for the message to be replied to based on the first edit operation. Figure 4 As shown, the message interface displays instant messaging message 1 and instant messaging message 2. If instant messaging message 1 is selected as the message to be replied to, the reply message 410 corresponding to instant messaging message 1 can be displayed in the message editing area.
[0050] By using the above methods, corresponding reply messages can be generated for instant messaging messages, enabling intelligent replies to instant messaging messages on the instant messaging platform and improving the processing efficiency of instant messaging messages.
[0051] It is worth noting that, Figure 1 The message processing system shown and Figures 2 to 4 The scenarios shown are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of message processing technology and the emergence of new business scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.
[0052] In this application embodiment, the description will be from the perspective of a message processing device. Specifically, the message processing device can be integrated into a computer device that has a storage unit and is equipped with a microprocessor and has computing capabilities. The computer device can be a server and / or a terminal, and there is no limitation on this.
[0053] Please see Figure 5 , Figure 5 This is a flowchart illustrating a message processing method provided in an embodiment of this application. The message processing method can be implemented by a terminal or jointly by a terminal and a server. Figure 5 The message processing methods shown include: Step 510: Display the messages to be replied to in the message interface; Step 520: In response to the intelligent reply trigger operation of the message to be replied to, display the message editing area in the message interface; Step 530: In the message editing area, multiple first message units are displayed sequentially. These multiple first message units are obtained sequentially by the intelligent reply model processing the message to be replied to. Step 540: During the process of sequentially displaying multiple first message units, in response to the first editing operation on the multiple first message units, control the intelligent reply model to stop processing the message to be replied to; Step 550: Determine the reply message for the message to be replied to based on the first editing operation.
[0054] Steps 510 to 550 are described in detail below.
[0055] In step 510, the message to be replied to is displayed in the message interface.
[0056] In this context, "pending reply messages" refer to messages that have been received but have not yet been replied to. The message interface is an interactive interface used for message communication, allowing relevant parties to view detailed information about received pending reply messages and to reply to them by entering text, voice, or sending images. For example... Figure 6 As shown, the message interface includes at least an area for displaying messages to be replied to.
[0057] For example, messages received in real time can be designated as messages awaiting a reply. Alternatively, the message interface can also display N historical messages. Historical messages refer to messages received in the past or replies sent in the past. The N historical messages include at least one of the following: replies sent in the past, messages received in the past and replied to, or messages received in the past but not replied to. For example, when N historical messages are displayed in the message interface, in response to a selection operation on the N historical messages, one historical message indicated by the selection operation (i.e., the selected historical message) can be designated as a message awaiting a reply. Alternatively, batch selection is supported, that is, when a batch selection operation on N historical messages is received, all M selected historical messages are designated as messages awaiting a reply. Here, N and M are both integers greater than 0, and N is greater than or equal to M. Or, in response to a content selection operation on a historical message, the selected content (partial or full content) of that historical message can be designated as a message awaiting a reply.
[0058] When multiple messages awaiting reply exist, their display order can be set according to the actual situation, without any restrictions. For example, the display order can be determined based on the receipt time of the messages. Alternatively, the display order can be determined based on the level corresponding to the message type (such as notification, alert, advertisement, and suggestion). Based on the display order of the multiple messages, they are then displayed in the message interface.
[0059] In step 520, in response to the smart reply trigger operation of the message to be replied to, the message editing area is displayed in the message interface.
[0060] The specific triggering method for smart reply actions is not limited. It can be triggered by smart reply options displayed in the message interface, by recognizing specific gesture operations (i.e., recognizing specific shape changes or positional changes of the relevant object's hand), or by recognizing specific voice input. Taking the triggering of smart reply actions based on smart reply options displayed in the message interface as an example... Figure 7 As shown in Figure (1), a smart reply option 710 is displayed in the message interface (the smart reply option 710 can be a response to a response to a message). Figure 6 The response option shown is triggered by the action of the message to be replied to (or it may be displayed together with the message to be replied to in the message interface). Therefore, when the smart response option 710 is triggered, it is considered that a smart response trigger action for the message to be replied to has been received, and the message interface displays the response option as shown in the example. Figure 7 The message editing area 720 is shown in Figure (2).
[0061] The message editing area is an interactive area for editing replies to messages awaiting a response. The content edited in the message editing area can be text messages, emoticons, images, or files. In some embodiments, content can be entered or modified in the message editing area using an input device (e.g., keyboard, mouse, or touch panel). Furthermore, audio data can be collected using a recording device and converted to text to display the resulting text content in the message editing area. The message editing area can also be equipped with a formatting tool. This tool allows adjustment of the formatting of the content in the message editing area, such as font, color, alignment, bold, or italics. Thus, content can be entered in the message editing area as a reply to a message in various ways. It should be noted that, unlike methods that rely solely on editing operations in the message editing area to obtain a reply, this embodiment uses an intelligent reply model to obtain the reply.
[0062] When there are multiple messages to be replied to, the system can receive a smart reply trigger operation for the selected message after any message to be replied to is selected, and in response to the smart reply trigger operation for the selected message, display the message editing area in the message interface.
[0063] In step 530, multiple first message units are displayed sequentially in the message editing area. These multiple first message units are obtained sequentially by processing the message to be replied to using the intelligent reply model.
[0064] Intelligent response models utilize various algorithms, technologies, and logical processing mechanisms to analyze and understand input information and generate corresponding responses. For example, an intelligent response model can be a large language model or a generative language model. Intelligent response models are obtained by training large language models or generative language models with a large number of training samples. Training a language model with a large number of training samples improves its ability to analyze input content, performs tasks such as text summarization and sentiment analysis, and generates output content with certain semantic meaning. Once the language model converges, the training process is complete, resulting in the intelligent response model.
[0065] By deploying an intelligent reply model on a terminal or server, the message to be replied to can be processed, thereby obtaining multiple first message units sequentially for the message to be replied to. For example, the intelligent reply model is fixedly configured to implement intelligent reply tasks. In this case, the message to be replied to can be directly input into the intelligent reply model, meaning the intelligent reply model will default to the message to be replied to as the input information. Alternatively, the intelligent reply model is a general model for implementing natural language processing (such as a large language model). In this case, the message to be replied to needs to be substituted into a specified template to construct a prompt, and then the prompt is input into the intelligent reply model. Only then will the intelligent reply model execute the intelligent reply task. The template needs to explicitly indicate the message to be replied to and provide instructions for executing the intelligent reply task. For example, a template like "The message to be replied to is: xxxx. Please help me generate a reply message," and a message to be replied to like "Welcome to read and share 'XX Newsletter'," will generate the prompt "The message to be replied to is: Welcome to read and share 'XX Newsletter'. Please help me generate a reply message" after substituting the message to be replied to into the template.
[0066] The first message unit refers to the smallest semantic unit, such as a word, subword, or character. For example, during the continuous processing of messages to be replied to using an intelligent reply model, the message editing area can display items such as... Figure 7The content shown in Figure (2) is as follows: each time a first message unit is generated, the message interface can be refreshed to display the newly generated first message unit in the message editing area. Sequential display refers to displaying the multiple first message units according to their generation order. This improves the real-time performance of the display and makes it easier for relevant objects to view the multiple first message units obtained through the intelligent reply model. Furthermore, if relevant objects are dissatisfied with the content of the multiple first message units obtained, they can intervene at any time, avoiding resource waste caused by the intelligent reply model continuing to process them.
[0067] In some embodiments, there is a possibility that the first message unit output by the intelligent reply model may not be displayed in the message editing area because the language used is incompatible with the common language of the interface display. Therefore, in this case, the corresponding format conversion logic can be invoked to convert the format of each first message unit output by the intelligent reply model, resulting in a first message unit that can be displayed in the message editing area. For example, if the first message unit output by the intelligent reply model uses Markdown, the format conversion logic between Markdown and Hypertext Markup Language can be invoked to convert the format of each first message unit output by the intelligent reply model, resulting in a first message unit using Hypertext Markup Language. Then, each first message unit using Hypertext Markup Language is rendered through a WebView component, a local rendering library, or server rendering to display each first message unit in the message editing area. This method effectively solves the display problem caused by language (format) mismatch, ensuring that the first message unit output by the intelligent reply model can be accurately and normally presented in the message editing area, improving the compatibility and stability of message display, and enabling relevant objects to view the first message unit in the message editing area.
[0068] In some embodiments, the number of first message units displayed can be limited. For example, a maximum display number can be set according to actual conditions. When the number of first message units is less than the maximum display number, multiple first message units can be displayed sequentially in the message editing area. When the number of first message units reaches the maximum display number, the intelligent reply model is controlled to stop processing the message to be replied to, or the processing of the intelligent reply model is not interfered with (i.e., the intelligent reply model continues to process the message to be replied to), but the first message units subsequently obtained by the intelligent reply model are no longer displayed in the message editing area. The above method of limiting the number of first message units displayed can avoid the message editing area from becoming cluttered due to too much information, enabling relevant objects to accurately and quickly understand the semantics contained therein based on the displayed first message units, and to perform subsequent operations accordingly, such as triggering editing operations.
[0069] It should be noted that the message editing area can be fixed or can adaptively increase as the number of first message units increases.
[0070] In some embodiments, during the sequential display of a plurality of first message units, the message processing method further includes: In response to a swipe operation on the message interface, the message to be displayed in the message interface is displayed according to the direction indicated by the swipe operation. The message to be displayed includes multiple first message units and a message to be replied to.
[0071] During the sequential display of multiple first message units, swiping operations on the message interface can be received. A swiping operation refers to an operation achieved through continuous movement. For example, a swiping operation on the message interface can be achieved through specific gestures. Alternatively, it can be achieved through an input device (such as a touch panel, mouse, etc.). Responding to the swiping operation on the message interface, the messages to be displayed are swiped according to the direction indicated by the swiping operation (e.g., from the top to the bottom of the message interface, from the bottom to the top of the message interface, or other directions). The messages to be displayed include multiple first message units and messages to be replied to. Due to the limited display space of the message interface, only a portion of the content of the messages to be displayed can be shown during the swiping display process. It should be noted that the swiping operation does not affect the intelligent reply to messages to be replied to; that is, while swiping to display the messages to be displayed on the message interface, the intelligent reply model can still continue to process the messages to be replied to. This allows relevant users to selectively view specific content; for example, users can refer to the messages to be replied to to determine whether the newly displayed first message unit meets their needs.
[0072] For example, see Figure 14 It can receive swipe gestures on the message interface, with the swipe gesture indicating the direction as follows: Figure 14 As shown in Figure (1), the direction is from the bottom to the top of the message interface. Therefore, following the direction indicated by the sliding operation, the message to be displayed in the message interface can be slid to change the content displayed on the message interface. Figure 14 The content shown in Figure (1) is transformed into Figure 14 The contents shown in Figure (2).
[0073] In some embodiments, in response to the triggering operation of the sliding option in the message interface, the message to be displayed in the message interface can be slid out at a certain sliding speed.
[0074] In some embodiments, the message processing method further includes: The intelligent reply style for the message to be replied to is determined so that the intelligent reply model processes the message according to the intelligent reply style, thereby obtaining multiple first message units in sequence.
[0075] For example, intelligent reply styles can be concise, detailed, formal, or friendly. Thus, the intelligent reply model can process the message to be replied to according to a specific intelligent reply style, obtaining multiple first message units that match that style. Specifically, this process can be implemented in several ways. For instance, the intelligent reply style can be incorporated when constructing the prompt, such as "Given the message to be replied to is: xxxx. Please generate a reply message for me using a concise reply style." Another example is the construction of an intelligent reply model with multiple branch networks, each corresponding to a specific intelligent reply style, and each branch network trained using a dataset corresponding to that style. After determining the intelligent reply style for the message to be replied to, the branch network corresponding to that style is selected to process the message.
[0076] For example, when the intelligent reply style for a message to be replied to is concise, the number of first message units can be limited, ensuring that the number of first message units generated by the intelligent reply model is less than or equal to a certain limit. When the intelligent reply style for a message to be replied to is detailed, the intelligent reply model can generate detailed content corresponding to each question or request in the message to be replied to. If the intelligent reply style for a message to be replied to is formal and rigorous, the intelligent reply model can choose highly professional words as first message units when generating them, and limit the number of first message units to ensure the accuracy and professionalism of multiple first message units, and to ensure that the overall content structure composed of multiple first message units is clear. When the intelligent reply style for a message to be replied to is friendly, the intelligent reply model can choose a more friendly expression (such as using caring language) when processing the message to be replied to, avoiding the use of overly formal words. In this way, the intelligent reply model can obtain multiple first message units that are more in line with the intelligent reply style of the message to be replied to, to meet the reply needs in different scenarios. Of course, the classification of intelligent reply styles is not limited to the above examples and can be set according to the actual application scenario.
[0077] On the one hand, the multiple intelligent reply styles greatly expand the applicability of intelligent replies, accurately matching different scenarios and needs. Whether it's a scenario that seeks to efficiently obtain key information, or a scenario that requires in-depth and detailed answers, rigorous and professional responses, or friendly and cordial communication, it can cover all of these. On the other hand, the intelligent reply model can process reply messages according to specific intelligent reply styles, which can significantly improve the quality of multiple first message units obtained, making them as consistent as possible with the reply style of the relevant object, and minimizing the need for the relevant object to trigger editing operations due to dissatisfaction with multiple first message units, thus improving the efficiency of generating reply messages.
[0078] In some embodiments, determining the smart reply style for the message to be replied to includes: The message interface displays multiple smart reply styles, each with its own selection option. In response to a selection of these options, the smart reply style corresponding to the selected option is chosen as the smart reply style for the message to be replied to; or... The system analyzes and processes multiple historical replies to determine the intelligent reply style for each message to be replied to.
[0079] There are several ways to determine the smart reply style for a message. One method is to display multiple smart reply style options in the message interface (e.g., ...). Figure 15 The options shown ("Concise Reply Option 1510," "Detailed Reply Option 1520," "Formal and Rigorous Reply Option 1530," and "Friendly Reply Option 1540") are displayed in various positions without restriction. When a selection operation is received for each of these smart reply styles, the smart reply style corresponding to the selected option is determined as the smart reply style for the message to be replied to. In this way, users can quickly select the appropriate smart reply style directly in the message interface based on their current needs, allowing the smart reply model to generate multiple first message units that meet their expectations. This greatly enhances users' sense of participation and control over the smart reply process, increasing their autonomy; simultaneously, it produces multiple high-quality, demand-responsive first message units, thereby significantly improving user satisfaction and the efficiency of reply generation, making the message processing process more practical and user-friendly.
[0080] The second approach determines the intelligent reply style for the message to be replied to based on multiple historical reply messages (referring to multiple previously sent reply messages). This involves analyzing and statistically processing the tone, emotion, and word usage style of multiple historical reply messages to arrive at the intelligent reply style for the message to be replied to. These multiple historical reply messages can be sent to the same recipient, to different recipients, or sent within a certain time period; there are no restrictions. This method eliminates the need for selection of the relevant recipients, simplifying the intelligent reply process, improving the intelligence and automation of message processing, and ensuring that the generated first message units conform to the reply style of the relevant recipients.
[0081] In some embodiments, multiple historical messages awaiting reply (referring to multiple previously received messages awaiting reply) can be categorized. For example, they can be categorized according to dimensions such as message source, message topic, sentiment category, receiving time, or language used. For each category, the response style of multiple historical replies corresponding to the historical messages awaiting reply belonging to that category is analyzed to obtain the intelligent response style corresponding to that category. In this way, upon receiving a message awaiting reply, the category to which the message belongs can be determined, and the intelligent response style corresponding to that category can be taken as the intelligent response style for the message to be replied to. Through the above method, the targeting and adaptability of intelligent replies can be significantly improved, avoiding a "one-size-fits-all" approach to intelligent replies. Whether facing formal messages in a work scenario or friendly communication messages in a life scenario, multiple appropriate first message units can be generated.
[0082] In step 540, during the sequential display of multiple first message units, in response to the first editing operation on the multiple first message units, the intelligent reply model is controlled to stop processing the message to be replied to.
[0083] See Figure 7 As shown in Figure (1), when the smart reply option 710 is triggered, it is considered that a smart reply trigger operation for the message to be replied to has been received. At this time, the message to be replied to is processed by the smart reply model, and multiple first message units are obtained in sequence, as shown in Figure (1). Figure 7 As shown in Figure (2), multiple first message units are displayed sequentially in the message editing area. During the process of displaying multiple first message units sequentially in the message editing area, if no first edit operation is received on the already displayed multiple first message units, new first message units will continue to be displayed in the message editing area until the intelligent reply model finishes processing the message to be replied to. At this point, the result is as follows: Figure 7 The multiple first message units 730 shown in Figure (3) are as follows.
[0084] See Figure 8 As shown in Figure (1), when the smart reply option 710 is triggered, it is considered that a smart reply trigger operation for the message to be replied to has been received. At this time, the message to be replied to is processed by the smart reply model, and multiple first message units are obtained in sequence, as shown in Figure (1). Figure 8 As shown in Figure (2), multiple first message units are displayed sequentially in the message editing area. During the process of displaying multiple first message units sequentially in the message editing area, if a first edit operation is received on the already displayed multiple first message units, the intelligent reply model is controlled to stop processing the message to be replied to. For example, see [reference needed]. Figure 8 As shown in Figure (2), the first editing operation is received when the first message unit "help" is first displayed. At this time, the intelligent reply model stops processing the message to be replied to, that is, the intelligent reply model stops generating subsequent first message units.
[0085] The first editing operation is an editing operation received during the sequential display of multiple first message units in the message editing area. The first editing operation may include at least one of a first deletion operation, a first modification operation, a first addition operation, a first formatting operation, or a first reorganization operation. For example, when the first editing operation includes a first deletion operation, in response to the first deletion operation on multiple first message units, the first message unit indicated by the first deletion operation can be deleted. When the first editing operation includes a first modification operation, in response to the first modification operation on multiple first message units, the first message unit indicated by the first modification operation can be modified. When the first editing operation includes a first addition operation, in response to the first addition operation on multiple first message units, a specific message unit indicated by the first addition operation can be inserted at a specific position indicated by the first addition operation on multiple first message units. When the first editing operation includes a first formatting operation, in response to the first formatting operation on multiple first message units, the alignment, line spacing, or character spacing of the multiple first message units can be modified according to the format indicated by the first formatting operation; or, according to the style indicated by the first formatting operation, the style of at least one of the multiple first message units can be modified (e.g., applying bold or italic). When the first editing operation includes a first reorganization operation, in response to the first reorganization operation on multiple first message units, the overall structure of the multiple first message units can be adjusted, such as adjusting the order of paragraphs formed by the multiple first message units. The first editing operation may also include other types of operations, and is not limited to the foregoing examples.
[0086] By following the steps described above, when a first edit operation is received on multiple first message units, the intelligent reply model can be controlled to stop processing the messages to be replied to (i.e., stop generating more first message units), thus avoiding conflicts between subsequently generated first message units and editing operations on related objects. Furthermore, it can promptly prevent the intelligent reply model from generating first message units that do not meet the requirements, avoiding waste of computing resources; simultaneously, it prevents content chaos in the message editing area caused by excessive generation by the intelligent reply model, improving editing efficiency.
[0087] In step 550, a reply message for the message to be replied to is determined based on the first editing operation.
[0088] After editing the multiple first message units in response to a first edit operation, a reply message for the message to be replied to can be determined. Specifically, the multiple first message units can be edited according to the first edit operation to obtain the edited multiple first message units. For example, see [link to example]. Figure 8 Taking the first edit operation as the first delete operation as an example, in response to the first delete operation, the first message unit indicated by the first delete operation can be deleted, that is, deleted as follows: Figure 8 (2) As shown in the figure, "electronic journal", "content", "very" and the subsequent first message unit, multiple first message units are obtained after editing. The multiple first message units after editing are as follows: Figure 8 As shown in Figure (3).
[0089] Then, based on the edited multiple first message units, a reply message for the message to be replied to can be obtained. For example, the entire set of edited multiple first message units can be directly used as the reply message for the message to be replied to. Alternatively, after editing the multiple first message units according to the first editing operation to obtain the edited multiple first message units, in response to a confirmation operation on the edited multiple first message units (e.g., the confirmation option in the message interface is triggered, the send option in the message interface is triggered, a specific gesture operation or specific voice input is received, etc.), the edited multiple first message units are determined as the reply message for the message to be replied to. Or, if no editing operation is received on the edited multiple first message units within a certain period after they are displayed, the edited multiple first message units are determined as the reply message for the message to be replied to. It should be noted that the edited multiple first message units here are obtained after the intelligent reply model has edited the multiple first message units in response to the first editing operation before completing the processing of the message to be replied to.
[0090] In this way, multiple first message units generated based on the intelligent reply model can obtain reply messages that better meet the reply needs of the relevant objects based on the received first editing operation.
[0091] It should be noted that the reply message can be displayed in the message editing area even before it has been sent. This allows the recipient to confirm the reply message correctly via gestures, voice input, or by triggering relevant options (e.g.,...). Figure 7 Sending reply messages using options such as (3) shown in the figure improves message processing efficiency.
[0092] In steps 510 to 550 above, the message to be replied to is displayed in the message interface. Then, in response to the intelligent reply trigger operation of the message to be replied to, the intelligent reply mode is entered. Specifically, in the intelligent reply mode, a message editing area is displayed in the message interface, and multiple first message units are displayed sequentially in the message editing area. These multiple first message units are obtained sequentially by the intelligent reply model processing the message to be replied to. This improves the real-time performance of the display and avoids blocking the waiting for all first message units to be obtained before displaying them. At the same time, in response to the first editing operation on the multiple first message units, the intelligent reply model is controlled to stop processing the message to be replied to, so that relevant objects can intervene at any time during the sequential display of multiple first message units to ensure that the needs of relevant objects are met. Finally, the reply message for the message to be replied to is determined according to the first editing operation, realizing the combination of manual reply and intelligent reply. This application, by combining manual reply and intelligent reply, can obtain reply messages that meet the reply needs of relevant objects more quickly, that is, it can improve message processing efficiency.
[0093] In some embodiments, determining a reply message for the message to be replied to based on the first editing operation includes: The message editing area displays the first candidate reply message, which is obtained by editing multiple first message units based on the first editing operation; Based on the first candidate reply message, generate a reply message for the message to be replied to.
[0094] If a first edit operation is received during the sequential display of multiple first message units, then in response to the first edit operation, the multiple first message units can be edited according to the first edit operation to obtain the edited multiple first message units. For ease of explanation, the edited multiple first message units as a whole are referred to as the first candidate reply message. It can be understood that the first candidate reply message will be displayed in the message editing area. The first edit operation can include at least one of the following: a first delete operation, a first modify operation, a first add operation, a first formatting operation, or a first reorganization operation. For example, if the message composed of multiple first message units is "Hello, email received", and if the first edit operation is used to instruct to change "Hello" to "Hello" and "Received" to "Read", then a first candidate reply message with content such as "Hello, email read" can be obtained and displayed in the message editing area. Or, see Figure 8 Figure (2) in Figure 8 Based on the multiple first message units shown in Figure (2), a first deletion operation on the multiple first message units can be received, and in response to the first deletion operation on the multiple first message units, the first message unit indicated by the first deletion operation in the multiple first message units can be deleted, resulting in the following: Figure 8 The first candidate reply message 810 shown in Figure (3) completes the editing of multiple first message units at this time.
[0095] Then, based on the first candidate reply message, a reply message for the message to be replied to can be generated. For example, the first candidate reply message can be directly identified as the reply message for the message to be replied to. For instance, in response to a confirmation operation on the first candidate reply message (e.g., the confirmation option in the message interface is triggered, the send option in the message interface is triggered, a specific gesture operation or specific voice input is received, etc.), the first candidate reply message is identified as the reply message for the message to be replied to. Alternatively, if no edit operation on the first candidate reply message is received within a certain period after it is displayed, the first candidate reply message is identified as the reply message for the message to be replied to.
[0096] Through the above steps, relevant parties are allowed to intervene at any time during the sequential display of multiple first message units. In other words, relevant parties are allowed to edit the multiple first message units that have been displayed during the sequential display of multiple first message units, thereby achieving a combination of manual reply and intelligent reply, and obtaining reply messages that better meet the reply needs of relevant parties.
[0097] In some embodiments, generating a response message for the message to be responded to based on the first candidate response message includes: In response to the intelligent reply trigger operation of the message to be replied to and the first candidate reply message, multiple second message units are displayed sequentially after the first candidate reply message displayed in the message editing area. These multiple second message units are obtained sequentially by processing the message to be replied to and the first candidate reply message through the intelligent reply model. Based on the first candidate reply message and multiple second message units, a reply message is generated for the message to be replied to.
[0098] Based on the first candidate reply message, a reply message can be generated for the message to be replied to. For example, the smart reply option can be continuously displayed in the message interface. Alternatively, the smart reply option can be displayed in the message interface after receiving the first edit operation. Or, the smart reply option can be displayed in the message interface after receiving the first candidate reply message. In response to the triggering operation of the smart reply option, the message to be replied to and the first candidate reply message can be processed by the smart reply model to obtain multiple second message units sequentially, which are then displayed sequentially after the first candidate reply message displayed in the message editing area. The second message unit is the smallest semantic unit obtained by the smart reply model after comprehensively analyzing the message to be replied to and the first candidate reply message. It should be noted that, in addition to triggering the smart reply option, the smart reply triggering operation can also be achieved through specific gesture operations, specific voice input, etc.
[0099] When inputting the message to be replied to and the first candidate reply message into the intelligent reply model, it is necessary to distinguish between them. For example, the message to be replied to and the first candidate reply message can be substituted into a specified template to construct a prompt, which is then input into the intelligent reply model. The intelligent reply model will then continue to reply based on the first candidate reply message. The template needs to clearly distinguish between the message to be replied to and the first candidate reply message, and also needs to provide instructions for executing the intelligent reply task. For example, the template could be "The message to be replied to is: xxxx. The current reply message is: xxxx. Please help me continue generating a reply message," where the message to be replied to is such as "Welcome to read and share 'XX Newsletter'," and the first candidate reply message is such as "Hello! Thank you for sharing the reading link of 'XX Newsletter'." Substituting the message to be replied to and the first candidate reply message into the template will construct the prompt "The message to be replied to is: Welcome to read and share 'XX Newsletter'. The current reply message is: Hello! Thank you for sharing the reading link of 'XX Newsletter'. Please help me continue generating a reply message."
[0100] For example, such as Figure 9 As shown, after receiving the first candidate reply message 810, if a smart reply trigger operation is received for both the message to be replied to and the first candidate reply message (e.g., ... Figure 9the smart reply option 710 in Figure (2) is triggered), then after the first candidate reply message displayed in the message editing area, a plurality of second message units may be displayed sequentially, that is Figure 9 "Wish", "work", "smooth", ",", "all the best", "and" in Figure (3). Then, based on the first candidate reply message 810 and the plurality of second message units, a reply message 910 for the to-be-replied message can be obtained.
[0101] In the process of sequentially displaying the plurality of second message units, editing operations on the first candidate reply message and the plurality of second message units may be received again. In response to the editing operations on the first candidate reply message and the plurality of second message units, the smart reply model is controlled to stop processing the to-be-replied message and the first candidate reply message. Moreover, after the editing of the first candidate reply message and the plurality of second message units is completed, it is also possible to perform smart reply again based on the edited content in response to a smart reply trigger operation, so as to carry out multiple rounds until a reply message for the to-be-replied message is obtained.
[0102] In the above steps, after the first candidate reply message is obtained, the smart reply model continues to generate the smart reply content (i.e., the second message units) for the to-be-replied message after the first candidate reply message based on the first candidate reply message and the to-be-replied message. In this way, the processing process of the smart reply model can refer to the first candidate reply message. Since the first candidate reply message is obtained based on the first editing operation and reflects the reply style of the relevant object to a certain extent, the plurality of second message units generated by the smart reply model can be more close to the reply style of the relevant object, that is, the quality of the second message units can be improved, making the second message units more in line with requirements, which is conducive to quickly generating a reply message.
[0103] In some embodiments, before controlling the smart reply model to stop processing the to-be-replied message in response to a first editing operation on a plurality of first message units, the message processing method further includes: In response to an editing trigger operation on the message editing area, displaying a message editing control in the message interface; The message editing control is configured to receive the first editing operation on the plurality of first message units.
[0104] An edit trigger operation is an operation that activates the message editing area. For example, it could be clicking anywhere within the message editing area, a specific gesture, or a specific voice input operation; there are no restrictions here. In response to the edit trigger operation on the message editing area, a message editing control is displayed in the message interface. Message editing controls are typically used for text editing, formatting, and interaction. Furthermore, attachments such as images, videos, or files, as well as emoticons, can be inserted through the message editing control. The message editing control can provide voice input functionality; that is, when the voice input option in the message editing control is triggered, it can receive voice input from the relevant user. After recognizing the voice input, the corresponding text can be obtained and displayed in the message editing area. For example, such as... Figure 10 As shown, in response to an edit trigger operation in the message editing area, a message editing control, such as 1010, is displayed in the message interface. This allows relevant objects to easily trigger the first editing operation through the message editing control displayed in the message interface, realizing interaction between the relevant objects and the message interface. Furthermore, in addition to text editing, it also allows the insertion of attachments such as images, videos, files, and emoticons, enriching the editing methods and meeting diverse needs. Moreover, by integrating voice input functionality, voice input can be converted into text, making the editing process more convenient.
[0105] In some embodiments, a plurality of first message units are displayed sequentially in the message editing area, including: In the message editing area, multiple first message units are displayed sequentially until the intelligent reply model finishes processing the message to be replied to; Message processing methods also include: In response to a second edit operation on multiple first message units, a reply message for the message to be replied to is determined based on the second edit operation.
[0106] Multiple first message units are displayed sequentially in the message editing area until the intelligent reply model finishes processing the message to be replied to. At this point, the intelligent reply model stops generating first message units, resulting in a message like this: Figure 7The figure (3) shows multiple first message units 730. Then, the relevant object is allowed to edit the multiple first message units. Specifically, after receiving the second edit operation on the multiple first message units, the multiple first message units can be edited according to the second edit operation to obtain the edited multiple first message units. It should be noted that the edited multiple first message units here are different from the edited multiple first message units mentioned above. The edited multiple first message units here are obtained after the intelligent reply model has finished processing the message to be replied to, in response to the second edit operation and after the multiple first message units are edited.
[0107] The second editing operation is performed after the intelligent reply model has completed processing the message to be replied to. The second editing operation may include at least one of a second deletion operation, a second modification operation, a second addition operation, a second formatting operation, or a second reorganization operation. For example, when the second editing operation includes a second deletion operation, in response to the second deletion operation on multiple first message units, the first message unit indicated by the second deletion operation can be deleted. When the second editing operation includes a second modification operation, in response to the second modification operation on multiple first message units, the first message unit indicated by the second modification operation on multiple first message units can be modified. When the second editing operation includes a second addition operation, in response to the second addition operation on multiple first message units, a specific message unit indicated by the second addition operation can be inserted at a specific position indicated by the second addition operation on multiple first message units. When the second editing operation includes a second formatting operation, in response to the second formatting operation on multiple first message units, the alignment, line spacing, or character spacing of the multiple first message units can be modified according to the format indicated by the second formatting operation, or the style of at least one of the multiple first message units can be modified (e.g., applying bold or italics) according to the style indicated by the second formatting operation. When the second editing operation includes a second reorganization operation, in response to the second reorganization operation on multiple first message units, the overall structure of the multiple first message units can be adjusted, such as adjusting the order of the paragraphs formed by the multiple first message units. It should be noted that the second editing operation can also include other types of operations and is not limited to the examples described above.
[0108] Subsequently, the reply message for the message to be replied to can be determined based on the second editing operation. For example, the edited multiple first message units can be directly used as the reply message for the message to be replied to. Alternatively, after receiving a confirmation operation on the edited multiple first message units, the edited multiple first message units can be used as the reply message for the message to be replied to, where the specific form of the confirmation operation is not limited. In one example, after obtaining the edited multiple first message units, a confirmation option or a send option can be displayed in the message interface. When the confirmation option or send option is triggered, the edited multiple first message units can be used as the reply message for the message to be replied to. In another example, when a specific gesture operation or a specific voice input is received, the edited multiple first message units can be used as the reply message for the message to be replied to. After the intelligent reply model completes the processing of the message to be processed, the relevant object is allowed to edit the obtained multiple first message units, thus obtaining a reply message that better meets the reply needs of the relevant object.
[0109] In some embodiments, determining a reply message for the message to be replied to based on the second editing operation includes: In the message editing area, a second candidate reply message is displayed, which is obtained by editing multiple first message units according to the second editing operation; Based on the second candidate reply message, generate a reply message for the message to be replied to.
[0110] Multiple first message units are displayed sequentially in the message editing area until the intelligent reply model finishes processing the message to be replied to. At this point, the intelligent reply model stops generating first message units, resulting in a message like this: Figure 7 The multiple first message units 730 shown in Figure (3) are then allowed to be edited by the relevant objects. Specifically, after receiving a second edit operation on the multiple first message units, the multiple first message units can be edited according to the second edit operation to obtain a second candidate reply message, and the second candidate reply message is displayed in the message editing area. The second edit operation may include at least one of a second delete operation, a second modify operation, a second add operation, a second formatting operation, or a second reorganization operation. In one example, such as Figure 11 As shown in Figure (1), in Figure 11 Based on the content shown in Figure (1), a second deletion operation on multiple first message units 730 can be received, and in response to the second deletion operation on multiple first message units 730, the first message unit indicated by the second deletion operation in the multiple first message units is deleted, resulting in the following: Figure 11 The second candidate response message 1110 is shown in Figure (2).
[0111] After obtaining the second candidate reply message, in response to the intelligent reply trigger operation on the second candidate reply message, the message units obtained sequentially after processing the message to be replied to and the second candidate reply message by the intelligent reply model can be displayed in the message editing area. During this process, when inputting the message to be replied to and the second candidate reply message into the intelligent reply model, it is necessary to distinguish between them. For example, the message to be replied to and the second candidate reply message can be substituted into a specified template to construct a prompt, and then the prompt can be input into the intelligent reply model. In this way, the intelligent reply model will continue to reply based on the second candidate reply message. The template could be something like, "The message to be replied to is known to be: xxxx. The current reply message is: xxxx. Please help me continue to generate a reply message." Thus, after the intelligent reply model completes the processing of the message to be replied to, it can process it again based on the content edited by the relevant object (i.e., the second candidate reply message), allowing the intelligent reply model to refer to the edited content of the relevant object, thereby improving the quality of the message units output by the intelligent reply model. Of course, the edit-intelligent reply process can be executed repeatedly in multiple rounds without limitation.
[0112] Alternatively, after receiving a second candidate reply message, it can be directly used as the reply to the message to be replied to. For example, after receiving a confirmation operation for the second candidate reply message, it can be used as the reply to the message to be replied to, where the specific form of the confirmation operation is not limited. In one example, after receiving the second candidate reply message, a confirmation option or a send option can be displayed in the message interface. When the confirmation option or send option is triggered, the second candidate reply message can be used as the reply to the message to be replied to. In another example, when a specific gesture operation or specific voice input is received, the second candidate reply message can be used as the reply to the message to be replied to. After the intelligent reply model completes the processing of the message to be processed, the relevant object is allowed to edit the multiple first message units obtained, thus obtaining a reply message that better meets the reply needs of the relevant object.
[0113] In some embodiments, in response to a second edit operation on a plurality of first message units, before determining a reply message for the message to be replied to based on the second edit operation, the message processing method further includes: Display message editing controls in the message interface; The message editing control is used to receive second editing operations on multiple first message units.
[0114] Before determining a reply message for the message to be replied to based on the second editing operation on multiple first message units, the interface of the message editing control can be called to display the message editing control in the message interface. For example, such as... Figure 12 As shown, the result is as follows Figure 12 After multiple first message units 730 shown in Figure (1), a message editing control 1210 can be displayed in the message interface. The above method can indicate that the processing of the intelligent reply model has been completed by displaying the message editing control. At this time, the second editing operation can be triggered through the message editing control. At the same time, it is convenient for relevant objects to trigger the second editing operation through the message editing control, so as to realize the interaction between relevant objects and the message interface. In addition to text editing, attachments such as pictures, videos, files, and emoticons can be inserted, which enriches the editing method and meets diverse needs. At the same time, voice input function can be integrated into the message editing control, which can convert voice input into text, making the editing process more convenient.
[0115] In some embodiments, a message insertion point can also be displayed at the target location (editing location) in the message editing area. The message insertion point, like a cursor, serves to indicate the target location for the editing operation. The target location can be determined based on the position of the last first message unit displayed in the message editing area. For example, the boundary position of the last first message unit in a preset direction can be determined, and the target location is determined based on the boundary position of the last first message unit in the preset direction and a preset distance. The preset direction can be determined based on the reading habits of the relevant object; for example, if the relevant object reads from left to right, then the preset direction is from left to right. Through the above steps, after all the obtained first message units are displayed, the message insertion point can be automatically inserted at the target location. This eliminates the need for the relevant object to manually select the target location to insert the message insertion point, facilitating the editing of multiple first message units and improving the editing efficiency.
[0116] In some embodiments, after displaying the message to be replied to in the message interface, the message processing method further includes: In response to a manual reply to a message, a message editing area is displayed in the message interface; In response to the third editing operation, the third candidate reply message edited according to the third editing operation is displayed in the message editing area; In response to the intelligent reply trigger operation of the message to be replied to and the third candidate reply message, multiple third message units are displayed sequentially after the third candidate reply message displayed in the message editing area. These multiple third message units are obtained sequentially by processing the message to be replied to and the third candidate reply message through the intelligent reply model. Based on the third candidate reply message and multiple third message units, generate a reply message for the message to be replied to.
[0117] After a message awaiting a reply is displayed in the message interface, manual replies are allowed. Specifically, in response to a manual reply trigger, a message editing area is displayed in the message interface. This manual reply trigger can be a selection of manual reply options displayed in the message interface, a specific gesture, or a specific voice input; there are no restrictions here. For example, it could be triggered by... Figure 13 The manual reply option 1310 is shown in Figure (1). Then, in response to the third editing operation, the third candidate reply message edited according to the third editing operation can be displayed in the message editing area, for example, Figure 13 The third candidate reply message 1320 is shown in Figure (2). A message editing control can be displayed in the message interface to receive the third editing operation.
[0118] Then, in response to the intelligent reply trigger operation of the message to be replied to and the third candidate reply message, the intelligent reply model can process the message to be replied to and the third candidate reply message to obtain multiple third message units in sequence. The third message unit is the smallest semantic unit obtained after processing the message to be replied to and the third candidate reply message by the intelligent reply model. When inputting the message to be replied to and the third candidate reply message into the intelligent reply model, it is necessary to distinguish between them. For example, the message to be replied to and the third candidate reply message can be substituted into a specified template to construct a prompt, and then the prompt can be input into the intelligent reply model. In this way, the intelligent reply model will continue to reply based on the third candidate reply message. The template could be something like, "The message to be replied to is known to be: xxxx. The current reply message is: xxxx. Please help me continue to generate a reply message." Thus, after the third candidate reply message is displayed in the message editing area, multiple obtained third message units can be displayed sequentially, for example, as shown below. Figure 13 The (3) figure shows multiple third message units 1330.
[0119] Based on the third candidate reply message and multiple third message units, a reply message for the message to be replied to can be obtained. For example, the entirety of the third candidate reply message and multiple third message units can be directly used as the reply message for the message to be replied to. Alternatively, upon receiving a confirmation operation on the third candidate reply message and multiple third message units (e.g., the confirmation option in the message interface is triggered, the send option in the message interface is triggered, a specific gesture operation is received, or a specific voice input is received), in response to the confirmation operation on the third candidate reply message and multiple third message units, the entirety of the third candidate reply message and multiple third message units is used as the reply message for the message to be replied to. Or, if no editing operation on the third candidate reply message and multiple third message units is received within a certain time after the intelligent reply model has completed processing, the entirety of the third candidate reply message and multiple third message units is used as the reply message for the message to be replied to.
[0120] During the sequential display of multiple third message units, editing operations on the third candidate reply message and the multiple third message units can be received. In response to these editing operations, the intelligent reply model can be controlled to stop processing the message to be replied to and the third candidate reply message. Furthermore, after editing the third candidate reply message and the multiple third message units is completed, an intelligent reply can be triggered again based on the edited content in response to an intelligent reply trigger operation. This process is repeated multiple times until a reply message for the message to be replied to is obtained.
[0121] Through the steps described above, after the relevant party manually inputs part of the reply content to the message to be replied to (i.e., the third candidate reply message), the intelligent reply model processes the message to be replied to and the third candidate reply message. Multiple third message units can then be displayed after the third candidate reply message. This combines manual reply with intelligent reply, allowing the intelligent reply model to refer to the third candidate reply message, which to some extent reflects the relevant party's reply style, and obtain multiple third message units that better match the relevant party's reply style. The above process is equivalent to the relevant party setting the tone for the reply style, and then the intelligent reply model continuing to reply based on this, effectively improving the accuracy of the intelligent reply.
[0122] In some embodiments, in response to a manual reply triggering operation of the message to be replied to, a message editing area can be displayed in the message interface. In response to a third editing operation, a third candidate reply message edited according to the third editing operation can be displayed in the message editing area, and the third candidate reply message can be used as a reply message to the message to be replied to.
[0123] In some embodiments, the message processing method further includes: The text extraction process is performed on the message to be replied to, resulting in the text to be processed in the message to be replied to. The unprocessed text in the message to be replied to is processed by the intelligent reply model to obtain multiple first message units in sequence.
[0124] Since intelligent reply models may only support text content as input, text extraction processing can be performed on the message to be replied to to obtain the text to be processed. This text can then be input into the intelligent reply model so that it can process the text and sequentially obtain multiple first message units. For example, text extraction processing can be performed on the text-formatted content of the message to be replied to to obtain the text to be processed, thus improving the efficiency of text extraction. Alternatively, text extraction processing can be performed on both text-formatted content and non-text-formatted content (if any) of the message to be replied to to obtain the text to be processed. This results in a more comprehensive extraction of the text to be processed, including non-text formats such as image and video formats.
[0125] The above method addresses the scenario where the intelligent reply model only supports text content as input. Converting the message to be replied to into a text format that the intelligent reply model can process ensures its proper handling, thus guaranteeing the normal execution of the intelligent reply task. It should be noted that if the message to be replied to contains a table, text extraction can be performed on the table to avoid omitting any information contained within it.
[0126] In one example, the message to be replied to is an email, which includes an email body and an attachment. In this case, text extraction can be performed on the email body and attachments to obtain the text from the email body and attachments respectively. Then, an intelligent reply model can be used to process the text from the email body and attachments (the text from the email body and attachments together constitutes the text to be processed), sequentially obtaining multiple first message units. In this way, the intelligent reply model can process all the content of the message to be replied to, making the obtained multiple first message units more accurate.
[0127] In some embodiments, the message to be replied to includes the text to be replied to and the image to be replied to. Text extraction processing is performed on the message to be replied to to obtain the text to be processed in the message to be replied to, including: Step 1610: Perform text recognition processing on the image to be replied to to obtain the text in the image; Step 1620: Merge the text to be replied to with the text in the image to be replied to to obtain the text to be processed.
[0128] In step 1610, the text to be replied to is the text-formatted content of the message to be replied to, and the image to be replied to is the non-text-formatted (specifically, image-formatted) content of the message to be replied to. When the message to be replied to includes both text and an image, text recognition processing can be performed on the image to be replied to obtain the text within the image. The image to be replied to can be a photo, screenshot, line chart, bar chart, or pie chart, etc. In one example, by rendering the hypertext markup language file corresponding to the message to be replied to, the message to be replied to can be displayed in the message interface. At this time, the hypertext markup language file includes text elements (such as paragraph elements, heading elements, etc.) and image elements. The image element can be an `img` tag, which includes information such as the source address of the image to be replied to, the display width of the image to be replied to, the display height of the image to be replied to, the border style of the image to be replied to, and the specified alternative text corresponding to the image to be replied to. The elements in the hypertext markup language file are traversed; if an image element exists, the source address of the image to be replied to is obtained from the image element, and the image to be replied to is retrieved based on the source address and cached locally. Subsequently, Optical Character Recognition (OCR) technology can be used to perform text recognition processing on the image to be replied to, obtaining the text within the image. Alternatively, a convolutional neural network model or a multimodal model can also be used to perform text recognition processing on the image to be replied to. For example, if the image to be replied to contains text such as "cutting-edge artificial intelligence technology," then the aforementioned text recognition processing methods can be used to extract the text contained in the image to be replied to, i.e., the text containing "cutting-edge artificial intelligence technology."
[0129] For text-formatted content in the message to be replied to, the text content can be extracted by traversing the text elements (such as paragraph elements (p tags), heading elements (h tags), etc.) in the HTML file from beginning to end, following a depth-first approach. Similarly, when table elements exist in the HTML file, their text content can be extracted and added to the reply text to avoid missing information contained within the table. Alternatively, table elements can be considered a special type of text element.
[0130] In step 1620, the text to be replied to and the text in the image to be replied to can be merged to obtain the text to be processed. Specifically, the relative positions of the image and the text to be replied to in the message to be replied to can be determined (such as the relative positions of different elements in the corresponding Hypertext Markup Language file of the message to be replied to), and then the text to be replied to and the text in the image to be replied to can be merged according to the relative positions. For example, if the image to be replied to is located before all the text to be replied to in the message to be replied to, then the text in the image to be replied to can be placed before the text to be replied to, so that the text to be replied to and the text in the image to be replied to can be merged. In this way, the comprehensiveness and accuracy of the reply message obtained by the intelligent reply method can be improved.
[0131] It should be noted that when the message to be replied to only includes an image, the text recognition processing method described above can be used to perform text recognition processing on the image to obtain the text within the image as the text to be processed. When the message to be replied to includes multiple images, text recognition processing can be performed on each image separately to obtain the text within each image. The text from the multiple images is then merged to obtain the text to be processed. These steps ensure that even when the message to be replied to only includes an image, the text within the image can be extracted and processed using an intelligent reply model to achieve an intelligent reply to the message. Furthermore, when the message to be replied to only includes the text to be replied to, the text to be replied to can be directly identified as the text to be processed.
[0132] Steps 1610-1620 above can perform text extraction processing based on the hypertext markup language file corresponding to the message to be replied to. This decouples the text extraction process from the display process of the message to be replied to, avoiding the impact of errors or incomplete display of the message on the text extraction process, and improving the accuracy of text extraction processing. Of course, in addition to performing text extraction processing based on the hypertext markup language file corresponding to the message to be replied to, a screenshot of the displayed message to be replied to can also be taken, and text recognition processing can be performed on the screenshot to obtain the text to be processed.
[0133] In some embodiments, the text to be replied to is merged with the text in the image to be replied to to obtain the text to be processed, including: Step 1710: Determine the number of characters in the text in the image to be replied to; Step 1720: When the number of characters exceeds the threshold, merge the text to be replied to with the text in the image to be replied to to obtain the text to be processed. Message processing methods also include: Step 1730: When the number of characters is less than or equal to the number threshold, the text to be replied to is determined as the text to be processed.
[0134] The character count is the total number of characters contained in the text. After obtaining the text in the image to be replied to, the character count of the text in the image to be replied to can be determined. For example, the total number of all letters, numbers, punctuation marks, spaces, and line breaks in the text in the image to be replied to can be determined. Alternatively, non-essential content such as spaces and line breaks can be ignored, and only the total number of letters, numbers, and punctuation marks can be determined.
[0135] It allows comparison of the relative relationship between the number of characters and a threshold value. The threshold value can be a value set based on the actual application scenario, or it can be an adaptive threshold. For example, the threshold value can be 50, 60, or other values; there are no restrictions here.
[0136] When the number of characters exceeds the threshold, the text in the image to be replied to is considered to have practical value. Therefore, the text to be replied to is merged with the text in the image to be replied to, resulting in the text to be processed.
[0137] When the number of characters is less than or equal to a threshold, the text in the image to be replied to is considered to have no practical value. Therefore, the text in the image to be replied to is filtered (ignored), and the text to be replied to is determined as the text to be processed, which can improve the accuracy and conciseness of the text to be processed. For example, if the image to be replied to is a landscape picture, which usually mainly shows natural landscapes or cityscapes, it may contain a small amount of text, such as picture descriptions, copyright information, etc. This text usually does not need to be replied to. If this text is included as part of the text to be processed, it may introduce noise, leading to a decrease in the accuracy of intelligent reply. As another example, if the image to be replied to is blurry, the accuracy of text recognition and processing will be greatly affected due to the low clarity. It may only be able to recognize some scattered and incomplete text, which does not provide valuable content. If this text is included as part of the text to be processed, it may introduce noise, leading to a decrease in the accuracy of intelligent reply.
[0138] In some embodiments, if the message to be replied to includes multiple images to be replied to, the number of characters in the text of each of the multiple images to be replied to is determined, and the text with a character count greater than the count threshold is merged with the text to be replied to to obtain the text to be processed.
[0139] In summary, the embodiments of this application can determine whether the text in the image to be replied to has practical value based on the number of characters, and then determine whether to include the text in the image to be replied to as part of the text to be processed. This reduces the possibility that images such as landscape pictures or blurry pictures will interfere with the subsequent processing, improves the accuracy of the text to be processed, and thus improves the accuracy of intelligent replies.
[0140] In some embodiments, the message processing method further includes: Step 1810: Divide the message to be replied to into multiple sub-messages; Step 1820: Compress the multiple sub-messages separately to obtain the compressed sub-messages corresponding to the multiple sub-messages; Step 1830: Merge the compressed sub-messages corresponding to the multiple sub-messages to obtain the merged message; The merged message is processed by the intelligent response model to obtain multiple first message units in sequence.
[0141] In step 1810, to facilitate processing by the intelligent reply model, the message to be replied to can be split into multiple sub-messages. The specific method of splitting the message to be replied to is not limited. For example, the message to be replied to can be split into multiple sub-messages with equal character counts based on the number of characters in the message. Alternatively, the message to be replied to can be split based on punctuation marks (such as commas, semicolons, or periods) and special characters (such as spaces or newlines) in the message to ensure that the sentences or paragraphs in each sub-message are complete.
[0142] In step 1820, multiple sub-messages can be compressed separately to obtain compressed sub-messages corresponding to each sub-message. For example, if the intelligent reply model has semantic compression capabilities (e.g., the intelligent reply model is a large language model), semantic compression of multiple sub-messages can be performed using the intelligent reply model to obtain compressed sub-messages corresponding to each sub-message. Alternatively, long phrases in multiple sub-messages can be replaced with synonymous abbreviations to achieve compression of multiple sub-messages. Alternatively, for any sub-message including an image, the image can be removed from the sub-message or cropped to achieve compression of the sub-message.
[0143] In step 1830, the compressed sub-messages corresponding to the multiple sub-messages are merged to obtain a merged message. Specifically, the order of the multiple sub-messages in the message to be replied to can be determined, and the compressed sub-messages corresponding to the multiple sub-messages are merged according to this order to obtain a merged message. Compared to the message to be replied to, the merged message has less information, thus reducing the processing burden on the intelligent reply model and improving its processing efficiency. At the same time, the merged message contains the more critical and important content of the message to be replied to, enabling the intelligent reply model to accurately understand the semantics of the merged message and provide a targeted reply, resulting in more accurate and targeted first message units, avoiding misunderstandings caused by the large amount of content in the message to be processed.
[0144] It should be noted that, in this embodiment of the application, the message to be replied to can also be directly compressed, and the compressed message can be input into the intelligent reply model for processing. Compared with this method, the advantage of splitting the message to be replied to into multiple sub-messages and then compressing each sub-message is that the sub-messages contain less information, and the semantics of the sub-messages can be understood more accurately and effectively to achieve compression processing, that is, the accuracy of compression processing can be improved.
[0145] In some embodiments, step 1810, splitting the message to be replied to into multiple sub-messages, includes: Step 1910: Determine the amount of information in the message to be replied to; Step 1920: When the amount of information exceeds the information threshold, split the message to be replied to into multiple sub-messages.
[0146] Information volume can refer to the number of characters in the message to be replied to, or it can refer to the storage space occupied by the message to be replied to (e.g., in the case of an email message). The information volume threshold can be a value set according to the actual application scenario, or it can be an adaptive threshold. When the information volume is the number of characters in the message to be replied to, the information volume threshold can be 3000 or 3500, without restriction. When the information volume is the storage space occupied by the message to be replied to, the information volume threshold can be 10 megabytes, or it can be other values. For ease of understanding, the following explanation uses the number of characters as the information volume.
[0147] When the amount of information in the message to be replied to (referring to the total amount of information in the message to be replied to) exceeds the information amount threshold, the content of the message to be replied to is relatively complex. When the intelligent reply model processes the message to be replied to, there is a possibility that it is difficult to determine the key content in the message to be replied to. Therefore, the message to be replied to is split into multiple sub-messages to improve the reliability of the multiple first message units obtained, and at the same time improve the processing efficiency when processing the message to be replied to.
[0148] In some embodiments, the message to be replied to can be divided into multiple sub-messages based on the amount of information to be split, wherein the amount of information in each sub-message does not exceed the amount of information to be split. The amount of information to be split can be set according to the capability of compression processing, that is, the amount of information to be split can refer to the amount of information applicable to compression processing. For example, if compression processing works well for texts of 2000 characters or less, then the amount of information to be split can be set to 2000. The amount of information to be split can be the same as or different from the information threshold, and there is no limitation on this. For example, if the amount of information in the message to be replied to is 5000 and the amount of information to be split is 2000, then the first 2000 characters of the message to be replied to are taken as the first sub-message, the middle 2000 characters of the message to be replied to are taken as the second sub-message, and the last 1000 characters of the message to be replied to are taken as the third sub-message, that is, it is finally split into 3 sub-messages.
[0149] It should be noted that when the amount of information in the message to be replied to is small, because the intelligent reply model can effectively analyze, understand, and extract key information (semantics) from the message, there is no need to split the message when the amount of information is less than or equal to the information threshold. The intelligent reply model can directly process the message to be replied to, sequentially obtaining multiple first message units. This simplifies the processing of messages with small amounts of information and improves the efficiency of intelligent replies.
[0150] In some embodiments, before displaying the message to be replied to in the message interface, the message processing method further includes: A first retrieval request is triggered based on the message source to retrieve a portion of the message to be replied to, wherein the portion of the message to be replied to is used to display in the message interface; Before displaying multiple first message units sequentially in the message editing area, the message processing method also includes: A second retrieval request is triggered based on the source of the message, in order to obtain the full content of the message to be replied to through the second retrieval request, wherein the full content of the message to be replied to is used for processing by the intelligent reply model.
[0151] Before displaying a message awaiting a reply in the message interface, a first retrieval request can be triggered based on the message source (e.g., server address). This first retrieval request retrieves a portion of the message's content from its source (e.g., a local database or server), and then displays that portion in the message interface. Specifically, the first retrieval request is used to retrieve a specified amount of information from the message awaiting a reply. Understandably, for messages with more information than the specified amount, the first retrieval request can retrieve a portion of the message's content. For messages with less than or equal to the specified amount, the first retrieval request can retrieve the entire message's content. This reduces the waiting time for relevant parties and enables rapid display of messages awaiting a reply in the message interface.
[0152] Typically, if a message to be replied to is not fully loaded (meaning only a portion of its content is retrieved), the unloaded portion needs to be retrieved only upon detecting a specific swipe event in the message interface, resulting in low loading efficiency. Furthermore, it's understandable that during the intelligent reply process, to ensure the accuracy of multiple first message units, the intelligent reply model needs to process the full content of the message to be replied to, enabling it to generate multiple corresponding first message units based on the complete content. Therefore, upon receiving an intelligent reply trigger operation, and only a portion of the message to be replied to is retrieved at the time of the trigger operation, a second retrieval request can be triggered based on the message source, thereby retrieving the full content of the message to be replied to. This second retrieval request does not limit the amount of information to be retrieved, thus allowing the retrieval of the full content of the message to be replied to. In this way, even when the message to be replied to is not fully loaded in the message interface, the full content can be retrieved via the second retrieval request upon receiving an intelligent reply trigger operation, ensuring the accuracy of subsequent intelligent replies.
[0153] It's important to note that before triggering the second retrieval request, the loading progress of the message to be replied to can be determined to decide whether to trigger the second retrieval request. The loading progress represents the ratio between the amount of information in the retrieved content of the message to be replied to, and the total amount of information in the entire message to be replied to. For example, a loading progress of 100% indicates that the message to be replied to has been fully loaded and its entire content has been retrieved. A loading progress greater than or equal to 0 but less than 100% indicates that the message to be replied to has not been fully loaded, and only a portion of its content has been retrieved.
[0154] The following will describe an exemplary application of the embodiments of this application in a real-world application scenario. For ease of understanding, an example will be given using a large language model as the intelligent response model.
[0155] Figure 20 This is another schematic diagram of message processing provided in the embodiments of this application, see [link to diagram]. Figure 20 First, from the product side, the message processing method provided in this application embodiment will be explained. Taking the message to be replied to as an email as an example, the message processing method mainly includes the following steps, which can be implemented by the terminal: 1) Display any email in the message interface, and respond to such... Figure 20 The triggering operation of the "Reply" option at the bottom of the message interface shown in Figure (1) will display the message interface as shown in Figure (2). Figure 20 The “Smart Reply” and “Manual Reply” options shown in Figure (2) allow you to start smart replying to emails when the “Smart Reply” option is triggered (i.e., when a smart reply trigger operation is received for an email). 2) During the intelligent reply process, in the message editing area, see as follows: Figure 20 As shown in Figure (3), the first message units obtained sequentially by the large language model can be displayed in sequence. Furthermore, during the intelligent reply process, in response to a swipe operation on the message interface (including an upward or downward swipe operation), the messages to be displayed in the message interface (including multiple already displayed first message units and emails) are displayed by swiping. The swipe operation does not affect the intelligent reply to emails. Then, step 3) or step 4) can be executed. 3) During the intelligent reply process, if the first edit operation on multiple first message units is received, the large language model is controlled to stop processing the email. This resolves the conflict between the intelligent reply process and the editing operations of related objects. 4) After the large language model completes the processing of the email, it can obtain, for example: Figure 20 The multiple first message units are shown in Figure (4). Furthermore, a message editing control can be raised in the message interface to prompt the relevant object to continue editing the obtained multiple first message units.
[0156] Figure 21 This is another flowchart illustrating the message processing method provided in this application embodiment; see below. Figure 21 The following example illustrates the message processing method provided in this application from the perspective of underlying implementation, taking the pending message as an email and the large language model deployed on a server.
[0157] Step 2110: In response to the smart reply trigger operation of the email, the terminal extracts the text to be replied to from the email, which includes the text to be replied to and the image to be replied to; Step 2120: The terminal extracts the text from the image to be replied to using OCR technology; Step 2130: The terminal merges the text to be replied to and the text in the image to be replied to to obtain the text to be processed; Step 2140: The server compresses the text to be processed to obtain the merged message; Step 2150: The server inputs the merged message into the large language model; Step 2160: The large language model sequentially obtains multiple first message units; Step 2170: The terminal refreshes the message interface every time it receives the first message unit sent by the server.
[0158] The following is a detailed explanation of steps 2110-2170 above.
[0159] First, in response to the smart reply trigger on the email, the terminal determines whether the email has finished loading. Specifically, the terminal can obtain the total amount of information in the email's full content and the amount of information in the downloaded content, and divide the amount of information in the downloaded content by the total amount of information in the email's full content to obtain the loading progress. The loading progress is a percentage within the range [0, 100%]. If the loading progress is 100%, the email is confirmed to be loaded successfully. If the loading progress is not 100%, it means the email has not yet finished loading. In this case, a retrieval request can be triggered based on the email's source information to retrieve the full content of the email from its source (such as a server).
[0160] Once the email has finished loading (i.e., the full content of the email has been obtained, specifically represented as a Hypertext Markup Language (HTML) file), for emails containing images to be replied to, the terminal can iterate through the corresponding HTML file to identify the `img` tags within the HTML file. These `img` tags contain the source URL of the image to be replied to. Since the large language model cannot directly read the image to be replied to based on its source URL, the source URL can be extracted from the `img` tag. The image can then be downloaded using an image download system based on this source URL and cached locally.
[0161] Here is an example of an img tag: <imgsrc="smiley.gif"alt="Smiley face"width="42"height="42"style="border:5px solid black"> The `src="smiley.gif"` attribute indicates the source address (source file path) of the image to be replied to. `alt="Smiley face"` indicates that the specified alternative text for the image to be replied to is "Smiley face", which will be displayed when the image to be replied to fails to load. `width="42"` indicates that the width of the image to be replied to is 42 pixels, `height="42"` indicates that the height of the image to be replied to is 42 pixels, and `style="border:5px solid black"` means that an inline Cascading Style Sheet is used to add a 5-pixel wide black border to the image to be replied to.
[0162] After the terminal caches the image to be replied to, it can call the optical character recognition (OCR) interface to perform text recognition processing on the image, converting it into text to obtain the text within the image, thus reducing resource consumption. Furthermore, to avoid interference from landscape images or blurry images on the large language model's processing, the text in the image to be replied to can be filtered based on the number of characters. That is, if the number of characters in the text of any image to be replied to is less than or equal to a threshold (e.g., 50), the text in that image to be replied to is filtered (ignored). The text in the image to be replied to, obtained through the above steps, is embedded into the corresponding Hypertext Markup Language (HTML) file of the email, replacing the `img` tags in the HTML file, and a marker is added at the embedding location. The marker can be "[Image Content] + Source URL of Image to be replied to + Text in Image to be replied to", where "[Image Content]" is only used to indicate the presence of text in the image to be replied to and has no actual meaning.
[0163] Subsequently, the terminal, following a depth-first approach, traverses multiple elements in the hypertext markup language file corresponding to the email from beginning to end. If a text element is encountered, the text within it is extracted (corresponding to the text to be replied to above). This extracted text is plain text, allowing for the removal of font, color, bold, or italic styles. If a markup element is encountered, the text within that markup element is extracted (i.e., the text in the image to be replied to). Then, based on the relative positions of different elements, the extracted text is merged into the text to be processed, thus ensuring the order and accuracy of the processed text.
[0164] Then, the terminal can send the text to be processed to the server, which will then execute the intelligent reply task using the intelligent reply model. Specifically, the server compresses the text to be processed. The compression process is explained below, taking the number of characters as an example. For the detailed process, please refer to [link to documentation / reference]. Figure 22 : Step 2210: Obtain the text to be processed; Step 2220: Determine if the number of characters in the text to be processed is less than 3000. If yes, end the process; otherwise, proceed to step 2230. Step 2230: Based on the number of characters to be split (corresponding to the amount of information to be split above), split the text to be processed to obtain multiple sub-messages; Step 2240: Compress multiple sub-messages separately using a large language model to obtain compressed sub-messages corresponding to each sub-message; Step 2250: Merge the compressed sub-messages corresponding to the multiple sub-messages to obtain the merged message.
[0165] After the server obtains the text to be processed, it can determine whether the number of characters in the text is less than 3000 (i.e., the information content threshold). If the number of characters in the text is less than or equal to 3000, no compression is performed, and the text is input into a large language model. The large language model processes the text to obtain multiple first message units. If the number of characters in the text is greater than 3000, the text is compressed. That is, based on the number of characters to be split (e.g., 2000), the text is split into multiple sub-messages. Subsequently, the large language model can compress each sub-message separately to obtain compressed sub-messages corresponding to each sub-message. Finally, the compressed sub-messages corresponding to each sub-message are merged to obtain a merged message.
[0166] The merged message is processed using a large language model. The large language model uses NextToken Prediction to process the merged message, sequentially generating multiple first message units (tokens). For each first message unit generated, the server sends it to the terminal in Markdown language, triggering a refresh on the terminal. Upon receiving each first message unit, the terminal invokes Markdown-to-Hypertext Markup Language (HML) format conversion logic to convert the Markdown-based first message unit to a HML-based first message unit, and then displays the HML-based first message unit in the message editing area using a WebView component.
[0167] Furthermore, during the processing of merged messages by the large language model, if the terminal receives the first edit operation on multiple first message units, the terminal sends a stop signal to the server to control the large language model to stop processing the merged messages.
[0168] After the large language model completes the processing of the merged message (the server can notify the terminal that the large language model has completed the processing), the terminal automatically displays the message insertion point (i.e., the cursor) after the multiple first message units displayed in the message editing area, and raises the message editing control so that the relevant objects can continue to edit the multiple first message units.
[0169] Compared to existing message processing methods, this application's embodiments incorporate a smart reply option into the traditional reply pop-up. Upon receiving a trigger action for the smart reply option, a smart reply to the email can be generated, eliminating the need for the recipient to spend significant time reading the email content and considering how to reply. Through this method, the recipient only needs to simply proofread or edit multiple first message units to obtain a reply message, significantly reducing the time spent processing emails and enabling the recipient to reply to a large number of emails in a shorter time. Furthermore, the recipient can view the smart reply process in real time and manually edit or rewrite the reply if dissatisfied with the generated first message units. In addition, by extracting text from both text-formatted and non-text-formatted content (such as images and charts) in the email, key details in the email can be avoided, improving the accuracy of the smart reply and preventing misunderstandings caused by missing crucial information.
[0170] To facilitate better implementation of the message processing method provided in the embodiments of this application, the embodiments of this application also provide an apparatus based on the above-described message processing method. The meanings of the terms used are the same as in the above-described message processing method, and specific implementation details can be found in the descriptions in the method embodiments.
[0171] Please see Figure 23 , Figure 23 This is a schematic diagram of the structure of a message processing device provided in an embodiment of this application. The message processing device is applied to a computer device and may include a first display module 2310, a second display module 2320, a third display module 2330, a control module 2340, and a first determination module 2350, wherein: The first display module 2310 is used to display messages to be replied to in the message interface; The second display module 2320 is used to display the message editing area in the message interface in response to the intelligent reply trigger operation of the message to be replied to; The third display module 2330 is used to sequentially display multiple first message units in the message editing area, wherein the multiple first message units are obtained sequentially by processing the message to be replied to through the intelligent reply model; Control module 2340 is used to control the intelligent reply model to stop processing the message to be replied to in response to the first editing operation on the multiple first message units during the process of sequentially displaying multiple first message units; The first determining module 2350 is used to determine the reply message for the message to be replied to based on the first editing operation.
[0172] In the above scheme, the first determining module 2350 is also used for: The message editing area displays the first candidate reply message, which is obtained by editing multiple first message units based on the first editing operation; Based on the first candidate reply message, generate a reply message for the message to be replied to.
[0173] In the above scheme, the first determining module 2350 is also used for: In response to the intelligent reply trigger operation of the message to be replied to and the first candidate reply message, multiple second message units are displayed sequentially after the first candidate reply message displayed in the message editing area. These multiple second message units are obtained sequentially by processing the message to be replied to and the first candidate reply message through the intelligent reply model. Based on the first candidate reply message and multiple second message units, a reply message is generated for the message to be replied to.
[0174] In the above scheme, the message processing device further includes a fourth display module (not shown), used for: In response to an edit trigger operation in the message editing area, display message editing controls in the message interface; The message editing control is used to receive the first editing operation on multiple first message units.
[0175] In the above scheme, the third display module 2330 is also used for: In the message editing area, multiple first message units are displayed sequentially until the intelligent reply model finishes processing the message to be replied to; The message processing apparatus also includes a second determining module (not shown), used for: In response to a second edit operation on multiple first message units, a reply message for the message to be replied to is determined based on the second edit operation.
[0176] In the above scheme, the second determining module is also used for: In the message editing area, a second candidate reply message is displayed, which is obtained by editing multiple first message units according to the second editing operation; Based on the second candidate reply message, generate a reply message for the message to be replied to.
[0177] In the above scheme, the message processing device also includes a fifth display module (not shown), used for: Display message editing controls in the message interface; The message editing control is used to receive second editing operations on multiple first message units.
[0178] In the above scheme, the message processing device further includes a third determining module (not shown), used for: In response to a manual reply to a message, a message editing area is displayed in the message interface; In response to the third editing operation, the third candidate reply message edited according to the third editing operation is displayed in the message editing area; In response to the intelligent reply trigger operation of the message to be replied to and the third candidate reply message, multiple third message units are displayed sequentially after the third candidate reply message displayed in the message editing area. These multiple third message units are obtained sequentially by processing the message to be replied to and the third candidate reply message through the intelligent reply model. Based on the third candidate reply message and multiple third message units, generate a reply message for the message to be replied to.
[0179] In the above scheme, the message processing device also includes a sliding display module (not shown), used for: In response to a swipe operation on the message interface, the message to be displayed in the message interface is displayed according to the direction indicated by the swipe operation. The message to be displayed includes multiple first message units and a message to be replied to.
[0180] In the above scheme, the message processing device also includes a style determination module (not shown), used for: The intelligent reply style for the message to be replied to is determined so that the intelligent reply model processes the message according to the intelligent reply style, thereby obtaining multiple first message units in sequence.
[0181] In the above scheme, the style determination module is also used for: The message interface displays multiple smart reply styles, each with its own selection option. In response to a selection of these options, the smart reply style corresponding to the selected option is chosen as the smart reply style for the message to be replied to; or... The system analyzes and processes multiple historical replies to determine the intelligent reply style for each message to be replied to.
[0182] In the above scheme, the message processing device further includes an extraction module (not shown), used for: The text extraction process is performed on the message to be replied to, resulting in the text to be processed in the message to be replied to. The unprocessed text in the message to be replied to is processed by the intelligent reply model to obtain multiple first message units in sequence.
[0183] In the above solution, the message to be replied to includes the text to be replied to and the image to be replied to. The extraction module is also used for: Perform text recognition processing on the image to be replied to to obtain the text in the image; The text to be replied to is merged with the text in the image to be replied to, resulting in the text to be processed.
[0184] In the above scheme, the extraction module is also used for: Determine the number of characters in the text within the image to be replied to; When the number of characters exceeds the threshold, the text to be replied to is merged with the text in the image to be replied to, resulting in the text to be processed. The extraction module is also used for: When the number of characters is less than or equal to the number threshold, the text to be replied to is determined as the text to be processed.
[0185] In the above scheme, the message processing device also includes a compression module (not shown), used for: Break the message to be replied to into multiple sub-messages; Multiple sub-messages are compressed separately to obtain compressed sub-messages corresponding to each sub-message. The compressed sub-messages corresponding to multiple sub-messages are merged to obtain a merged message; The merged message is processed by the intelligent response model to obtain multiple first message units in sequence.
[0186] In the above scheme, the compression module is also used for: Determine the amount of information in the message to be replied to; When the amount of information exceeds the information threshold, the message to be replied to will be split into multiple sub-messages.
[0187] In the above scheme, the message processing device further includes a first request module (not shown), used for: A first retrieval request is triggered based on the message source to retrieve a portion of the message to be replied to, wherein the portion of the message to be replied to is used to display in the message interface; The message processing apparatus also includes a second request module (not shown), for: A second retrieval request is triggered based on the source of the message, in order to obtain the full content of the message to be replied to through the second retrieval request, wherein the full content of the message to be replied to is used for processing by the intelligent reply model.
[0188] For details on the implementation of each module, please refer to the previous examples, which will not be repeated here.
[0189] Reference Figure 24 , Figure 24 To implement the partial structural block diagram of the terminal 140 in this application embodiment, the terminal 140 includes: a radio frequency (RF) circuit 2410, a memory 2415, an input unit 2430, a display unit 2440, a sensor 2450, an audio circuit 2460, a wireless fidelity (WiFi) module 2470, a processor 2480, and a power supply 2490, etc. Those skilled in the art will understand that... Figure 24 The terminal 140 structure shown does not constitute a limitation on a mobile phone or computer, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.
[0190] The RF circuit 2410 can be used to receive and transmit signals during information transmission or calls. In particular, it receives downlink information from the base station and processes it with the processor 2480; in addition, it transmits uplink data to the base station.
[0191] The memory 2415 can be used to store software programs and modules. The processor 2480 executes various terminal functions and message processing by running the software programs and modules stored in the memory 2415.
[0192] The input unit 2430 can be used to receive input numeric or character information, and to generate key signal inputs related to the terminal's settings and function control. Specifically, the input unit 2430 may include a touch panel 2431 and other input devices 2432.
[0193] Display unit 2440 can be used to display input or provided information, as well as various menus of the terminal. Display unit 2440 may include display panel 2441.
[0194] Audio circuitry 2460, speaker 2461, and microphone 2462 provide an audio interface.
[0195] In this embodiment, the processor 2480 included in the terminal 140 can execute the message processing method of the previous embodiment.
[0196] The terminal 140 in this application embodiment includes, but is not limited to, mobile phones, computers, intelligent voice interaction devices, smart home appliances, vehicle terminals, and aircraft. This application embodiment can be applied to various scenarios, including but not limited to cloud technology, artificial intelligence, smart transportation, and assisted driving.
[0197] Figure 25This is a partial structural block diagram of a server 110 implementing an embodiment of this application. The server 110 can vary significantly due to different configurations or performance characteristics, and may include one or more central processing units (CPUs) 2522 (e.g., one or more processors) and memory 2532, and one or more storage media 2530 (e.g., one or more mass storage devices) for storing application programs 2542 or data 2544. The memory 2532 and storage media 2530 may be temporary or persistent storage. The program stored in the storage media 2530 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the server 110. Furthermore, the CPU 2522 may be configured to communicate with the storage media 2530 and execute the series of instruction operations in the storage media 2530 on the server 110.
[0198] Server 110 may also include one or more power supplies 2526, one or more wired or wireless network interfaces 2550, one or more input / output interfaces 2558, and / or one or more operating systems 2541, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.
[0199] The central processing unit 2522 in server 110 can be used to execute the message processing method of the embodiments of this application.
[0200] This application also provides a computer-readable storage medium for storing program code for executing the message processing methods of the foregoing embodiments.
[0201] This application also provides a computer program product, which includes a computer program. A processor of a computer device reads and executes the computer program, causing the computer device to perform the message processing method described above.
[0202] Furthermore, the terms “comprising” and “including”, and any variations thereof, are intended to cover non-exclusive inclusion, such that a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or apparatus.
[0203] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.
[0204] It should be understood that in the description of the embodiments of this application, "multiple" means two or more, "greater than", "less than", "exceeding" etc. are understood to exclude the number itself, and "above", "below", "within" etc. are understood to include the number itself.
[0205] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.
[0206] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.
[0207] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0208] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0209] It should also be understood that the various implementation methods provided in this application can be combined arbitrarily to achieve different technical effects.
[0210] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.
[0211] The above is a detailed description of the embodiments of this application. However, this application is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.
Claims
1. A message processing method, characterized in that, include: Display pending messages in the message interface; In response to the intelligent reply trigger operation of the message to be replied to, a message editing area is displayed in the message interface; In the message editing area, multiple first message units are displayed sequentially, wherein the multiple first message units are obtained sequentially by processing the message to be replied to through an intelligent reply model; During the sequential display of the plurality of first message units, in response to the first editing operation on the plurality of first message units, the intelligent reply model is controlled to stop processing the message to be replied to; The reply message for the message to be replied to is determined based on the first editing operation.
2. The message processing method according to claim 1, characterized in that, The step of determining the reply message for the message to be replied to based on the first editing operation includes: In the message editing area, a first candidate reply message is displayed, wherein the first candidate reply message is obtained by editing the plurality of first message units according to the first editing operation; Based on the first candidate reply message, a reply message is generated for the message to be replied to.
3. The message processing method according to claim 2, characterized in that, The step of generating a reply message for the message to be replied to based on the first candidate reply message includes: In response to the intelligent reply triggering operation on the message to be replied to and the first candidate reply message, a plurality of second message units are displayed sequentially after the first candidate reply message displayed in the message editing area. The plurality of second message units are obtained sequentially by processing the message to be replied to and the first candidate reply message through the intelligent reply model. Based on the first candidate reply message and the plurality of second message units, a reply message is generated for the message to be replied to.
4. The message processing method according to claim 1, characterized in that, Before controlling the intelligent reply model to stop processing the message to be replied to in response to the first edit operation on the plurality of first message units, the message processing method further includes: In response to an edit trigger operation on the message editing area, a message editing control is displayed on the message interface; The message editing control is used to receive a first editing operation on the plurality of first message units.
5. The message processing method according to claim 1, characterized in that, In the message editing area, a plurality of first message units are displayed sequentially, including: In the message editing area, multiple first message units are displayed sequentially until the intelligent reply model finishes processing the message to be replied to; The message processing method further includes: In response to a second edit operation on the plurality of first message units, a reply message for the message to be replied to is determined based on the second edit operation.
6. The message processing method according to claim 5, characterized in that, The step of determining the reply message for the message to be replied to based on the second editing operation includes: In the message editing area, a second candidate reply message is displayed, wherein the second candidate reply message is obtained by editing the plurality of first message units according to the second editing operation; Based on the second candidate reply message, a reply message is generated for the message to be replied to.
7. The message processing method according to claim 5, characterized in that, Before determining a reply message for the message to be replied to based on the second editing operation in response to the plurality of first message units, the message processing method further includes: The message editing control is displayed in the message interface; The message editing control is used to receive a second editing operation on the plurality of first message units.
8. The message processing method according to claim 1, characterized in that, After displaying the message to be replied to in the message interface, the message processing method further includes: In response to a manual reply to the message to be replied to, a message editing area is displayed in the message interface; In response to the third editing operation, a third candidate reply message edited according to the third editing operation is displayed in the message editing area; In response to the intelligent reply triggering operation on the message to be replied to and the third candidate reply message, a plurality of third message units are displayed sequentially after the third candidate reply message displayed in the message editing area. The plurality of third message units are obtained sequentially by processing the message to be replied to and the third candidate reply message through the intelligent reply model. A reply message is generated for the message to be replied to based on the third candidate reply message and the plurality of third message units.
9. The message processing method according to claim 1, characterized in that, During the sequential display of the plurality of first message units, the message processing method further includes: In response to a swipe operation on the message interface, the message to be displayed in the message interface is swiped to show the message in the direction indicated by the swipe operation. The message to be displayed includes the plurality of first message units and the message to be replied to.
10. The message processing method according to claim 1, characterized in that, The message processing method further includes: A smart reply style is determined for the message to be replied to, so that the smart reply model processes the message to be replied to according to the smart reply style, thereby obtaining the plurality of first message units in sequence.
11. The message processing method according to claim 10, characterized in that, The determination of the intelligent reply style for the message to be replied to includes: The message interface displays multiple smart reply styles, each with its own options. In response to a selection operation of these options, the smart reply style corresponding to the selected option is determined as the smart reply style for the message to be replied to; or... Analyze the response styles of multiple historical response messages to obtain an intelligent response style for the message to be responded to.
12. The message processing method according to claim 1, characterized in that, The message processing method further includes: The message to be replied to is processed by extracting text to obtain the text to be processed in the message to be replied to; The text to be processed in the message to be replied to is processed by the intelligent reply model to obtain the plurality of first message units in sequence.
13. The message processing method according to claim 12, characterized in that, The message to be replied to includes a text to be replied to and an image to be replied to. The step of extracting text from the message to be replied to, to obtain the text to be processed in the message to be replied to, includes: The image to be replied to is processed by text recognition to obtain the text in the image to be replied to; The text to be replied to is merged with the text in the image to be replied to, resulting in the text to be processed.
14. The message processing method according to claim 13, characterized in that, The step of merging the text to be replied to with the text in the image to be replied to, to obtain the text to be processed, includes: Determine the number of characters in the text of the image to be replied to; When the number of characters exceeds the threshold, the text to be replied to is merged with the text in the image to be replied to to obtain the text to be processed. The message processing method further includes: When the number of characters is less than or equal to the number threshold, the text to be replied to is determined as text to be processed.
15. The message processing method according to claim 1, characterized in that, The message processing method further includes: The message to be replied to is split into multiple sub-messages; The multiple sub-messages are compressed separately to obtain compressed sub-messages corresponding to the multiple sub-messages; The compressed sub-messages corresponding to the multiple sub-messages are merged to obtain a merged message; The merged message is processed by the intelligent reply model to obtain the plurality of first message units in sequence.
16. The message processing method according to claim 15, characterized in that, The step of splitting the message to be replied to into multiple sub-messages includes: Determine the amount of information in the message to be replied to; When the amount of information exceeds the information threshold, the message to be replied to is split into multiple sub-messages.
17. A message processing device, characterized in that, The message processing device includes: The first display module is used to display messages awaiting reply in the message interface; The second display module is used to display a message editing area in the message interface in response to the intelligent reply trigger operation of the message to be replied to; The third display module is used to sequentially display multiple first message units in the message editing area, wherein the multiple first message units are obtained sequentially by processing the message to be replied to through an intelligent reply model; The control module is used to control the intelligent reply model to stop processing the message to be replied to in response to a first editing operation on the plurality of first message units during the sequential display of the plurality of first message units. The first determining module is used to determine a reply message for the message to be replied to based on the first editing operation.
18. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted for loading by a processor to perform the message processing method according to any one of claims 1 to 16.
19. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the message processing method according to any one of claims 1 to 16.
20. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the message processing method according to any one of claims 1 to 16.
