Information processing method and apparatus, device, computer-readable storage medium and computer program product

By displaying the reference annotations of key information and their search result locations in smart sessions, the problem of low information acquisition efficiency in human-computer interaction is solved, and the effect of quickly obtaining and verifying information is achieved.

WO2024234614A1PCT designated stage expired Publication Date: 2024-11-21TENCENT TECHNOLOGY (SHENZHEN) CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
PCT/CN2023/135279
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-16
Filing Date
2023-11-30
Publication Date
2024-11-21

AI Technical Summary

Technical Problem

In the prior art, users cannot effectively improve information acquisition efficiency and human-computer interaction efficiency in human-computer interaction, especially in smart conversation scenarios, users need to interact multiple times to obtain more information or verify the credibility of information.

Method used

By displaying a reply message in the session interface, it contains key information and its reference annotations, and displays the location of the key information in the target search results after the reference annotation is triggered, reducing the user's search time in the search results.

Benefits of technology

It improves information acquisition efficiency and human-computer interaction efficiency, enables users to quickly locate and verify information sources, and reduces the number of interactions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2023135279_21112024_PF_FP_ABST
    Figure CN2023135279_21112024_PF_FP_ABST
Patent Text Reader

Abstract

Provided in the present application are an information processing method and apparatus, an electronic device, a computer-readable storage medium, and a computer program product. The method comprises: in response to a message sending operation, displaying a first message in a session interface, and displaying in the session interface a second message used for replying to the first message, the second message comprising key information used for replying to the first message and a reference mark of each piece of key information, and the key information corresponding to a search result associated with the reference mark; and, in response to a trigger operation for a target reference mark, displaying, in a reference interface, information at a first position in a target search result, the information at the first position comprising target key information, and the target key information being key information corresponding to the target reference mark.
Need to check novelty before this filing date? Find Prior Art

Description

Information processing method, device, equipment, computer-readable storage medium, and computer program product

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application is based on the Chinese patent application with application number 202310552381.3 and application date May 16, 2023, and claims the priority of the Chinese patent application. The entire content of the Chinese patent application is hereby introduced into this application as a reference. Technical Field

[0003] The present application relates to computer technology, and in particular to an information processing method, device, electronic device, computer-readable storage medium, and computer program product. Background Art

[0004] With the rapid development of Internet technology, search technology is no longer completely bound to search scenarios. Search technology can also be used in intelligent conversation scenarios. For the first message input by the user, the intent of the first message can be identified, and the identified first message can be used as the basis for search. Based on the search results, the second message can be used to reply to the first message.

[0005] However, the second message provided in the related art can only provide a limited amount of information, which cannot meet the needs of users who want to learn more information, or the user may doubt the credibility of the second message, which requires the user to continue to initiate multiple subsequent dialogue interactions.

[0006] There is no effective solution in related technologies for improving the efficiency of human-computer interaction in human-computer conversations.

[0007] Summary of the Invention

[0008] The embodiments of the present application provide an information processing method, device, electronic device, computer-readable storage medium, and computer program product, which can improve the efficiency of human-computer interaction during a conversation.

[0009] The technical solution of the embodiment of the present application is implemented as follows:

[0010] The present invention provides an information processing method, including:

[0011] Display the conversation interface;

[0012] In response to the message sending operation, displaying a first message in a conversation interface and displaying a second message in the conversation interface for replying to the first message; wherein the second message includes key information for replying to the first message and a reference mark for each key information, and the key information corresponds to the search result associated with the reference mark;

[0013] In response to a trigger operation for a target reference annotation, information of the first position in the target search result is displayed in a reference interface; wherein the information of the first position includes the target key information, and the target key information is the key information corresponding to the target reference annotation.

[0014] An embodiment of the present application provides an information processing device, including:

[0015] a display module configured to display a conversation interface;

[0016] a conversation module configured to, in response to a message sending operation, display a first message in a conversation interface and display a second message in the conversation interface for replying to the first message;

[0017] The second message includes key information for replying to the first message and a reference mark of each key information, and the key information corresponds to the search result associated with the reference mark;

[0018] The jump module is configured to display the information of the first position in the target search result in the reference interface in response to a trigger operation for the target reference annotation; wherein the information of the first position includes the target key information, and the target key information is the key information corresponding to the target reference annotation.

[0019] An embodiment of the present application provides an electronic device, including:

[0020] a memory for storing computer-executable instructions;

[0021] The processor is used to implement the information processing method provided in the embodiment of the present application when executing the computer-executable instructions stored in the memory.

[0022] An embodiment of the present application provides a computer-readable storage medium storing computer-executable instructions for causing a processor to execute the information processing method provided in the embodiment of the present application.

[0023] An embodiment of the present application provides a computer program product, including computer-executable instructions. When the computer-executable instructions are executed by a processor, the information processing method provided in the embodiment of the present application is implemented.

[0024] The embodiments of the present application have the following beneficial effects:

[0025] Through the embodiment of the present application, when automatically replying to the first message, a reference mark of the key information can be displayed in the second message, so that the user can view the search results of the source of the key information through the reference mark, and can locate the position of the key information in the search results for display, so that the user does not need to search up and down in the search results for the key information, thereby improving the efficiency of information acquisition and human-computer interaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] FIG1 is a schematic diagram of the architecture of an information processing system provided in an embodiment of the present application;

[0027] FIG2 is a schematic structural diagram of an electronic device provided in an embodiment of the present application;

[0028] 3A-3D are flowcharts of an information processing method according to an embodiment of the present application;

[0029] 4A-4D are schematic diagrams of interfaces of the information processing method provided in an embodiment of the present application;

[0030] 5A-5C are schematic diagrams of interfaces of the information processing method provided in an embodiment of the present application;

[0031] FIG6 is a flow chart of an information processing method according to an embodiment of the present application;

[0032] FIG7 is a schematic diagram of the architecture of the information processing method provided in an embodiment of the present application. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0034] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0035] In the following description, the terms "first\second\third" involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0037] Before further describing the embodiments of the present application in detail, the nouns and terms involved in the embodiments of the present application are explained. The nouns and terms involved in the embodiments of the present application are subject to the following interpretations.

[0038] 1) In response, it is used to indicate the conditions or states on which the executed operations depend. When the dependent conditions or states are met, one or more operations executed can be real-time or have a set delay. Unless otherwise specified, there is no restriction on the order in which the multiple operations executed are executed.

[0039] 2) The first message includes all information generated during a conversation (such as chatting, consulting, etc.), including text, images, videos, voice files, etc. The first message can be a certain first message, such as "How to farm crayfish". The first message can also be a message with the intention of asking a question, such as "Learn to play Frisbee", which can be identified as a question about how to play Frisbee.

[0040] 3) Reference mark: a mark used to indicate the source of information. For example, if the reference mark after a piece of information is "1", it means that the reference source of the information is the reference source numbered 1.

[0041] 4) Source link. The source link is a trigger entry. By triggering the source link, you can jump to a web page or document (reference source). Each reference mark has its own source link. For example, the reference mark after a certain information is "1". Reference mark 1 corresponds to the source link. By triggering the source link, you can jump to the web page and display the information on the web page.

[0042] In the related technology, the terminal receives a first message input by the user. For example, the first message is a specific first message A. The server generates query information (query) according to the first message A and searches based on the query information, obtains search results and generates answers based on the search results, and returns the answers as a second message to the terminal used by the user.

[0043] The embodiments of the present application provide an information processing method, device, electronic device, computer-readable storage medium, and computer program product, which can improve the information acquisition efficiency and human-computer interaction efficiency during the conversation process.

[0044] The following describes exemplary applications of the electronic devices provided in the embodiments of the present application. The electronic devices provided in the embodiments of the present application can be implemented as various types of user terminals such as laptop computers, tablet computers, desktop computers, set-top boxes, mobile devices (for example, mobile phones, portable music players, personal digital assistants, dedicated information devices, portable gaming devices), intelligent voice interaction devices, smart home appliances, vehicle-mounted terminals, aircraft, etc.

[0045] Refer to Figure 1, which is a structural diagram of the information processing system provided in an embodiment of the present application. In order to support a social application, the terminal 400 is connected to the server 200 through the network 300. The network 300 can be a wide area network or a local area network, or a combination of the two.

[0046] In some embodiments, the terminal 400 of the target object A displays a conversation interface and receives a message sending operation, displays a first message in the conversation interface of the terminal 400, sends the first message to the server 200, and the server 200 obtains a second message for replying to the first message, the second message including key information for replying to the first message and a reference markup for each key information, and the key information corresponds to the search result associated with the reference markup; the server 200 returns the second message to the terminal 400 and displays the second message in the conversation interface of the terminal 400; the terminal 400 receives a trigger operation for the target reference markup, generates a source view request based on the trigger operation, and sends the source view request to the server 200, the server 200 obtains the target search result corresponding to the target reference markup, and obtains the first position of the key information corresponding to the target reference markup in the target search result, the server 200 returns the first position and the target search result to the terminal 400, and displays the information in the first position of the target search result in the reference interface of the terminal 400; wherein the information in the first position includes the target key information, which is the key information corresponding to the target reference markup.

[0047] In some embodiments, a terminal or server can implement the information processing method provided in the embodiments of the present application by running a computer program. For example, the computer program can be a native program or software module in the operating system; it can be a native application (APP, Application), that is, a program that needs to be installed in the operating system to run, for example, a social network client; it can also be a small program, that is, a program that can be run only by downloading it into a browser environment; it can also be a small program that can be embedded in any APP. In short, the above-mentioned computer program can be any form of application, module or plug-in.

[0048] The embodiments of the present application can be implemented with the help of cloud technology. Cloud technology refers to a hosting technology that unifies a series of resources such as hardware, software, and network within a wide area network or a local area network to realize data calculation, storage, processing, and sharing.

[0049] Cloud technology is a general term for network, information, integration, management platform, and application technologies used in the cloud computing business model. It can form a resource pool that can be used flexibly and conveniently on demand. Cloud computing technology will become a key support. The backend services of technical network systems require a large amount of computing and storage resources.

[0050] As an example, the server 200 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal 400 can be a smart phone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, smart voice interaction device, smart home appliance, vehicle terminal, aircraft, etc., but is not limited to this. The terminal 400 and the server 200 can be directly or indirectly connected via wired or wireless communication, which is not limited in the embodiments of the present application.

[0051] Referring to Figure 2, Figure 2 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The terminal shown in Figure 2 includes: at least one processor 410, a memory 450, at least one network interface 420, and a user interface 430. The various components in the terminal 400 are coupled together via a bus system 440. It will be understood that the bus system 440 is used to achieve connection and communication between these components. In addition to including a data bus, the bus system 440 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, in Figure 2, various buses are labeled as bus system 440.

[0052] The processor 410 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., where the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0053] The user interface 430 includes one or more output devices 431 that enable presentation of media content, including one or more speakers and / or one or more visual display screens. The user interface 430 also includes one or more input devices 432, including user interface components that facilitate user input, such as a keyboard, mouse, microphone, touch screen display, camera, other input buttons and controls.

[0054] The memory 450 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state memory, hard drives, optical drives, etc. The memory 450 may optionally include one or more storage devices that are physically remote from the processor 410.

[0055] The memory 450 includes volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be a read-only memory (ROM), and the volatile memory may be a random access memory (RAM). The memory 450 described in the embodiments of the present application is intended to include any suitable type of memory.

[0056] In some embodiments, the memory 450 can store data to support various operations, examples of which include programs, modules, and data structures, or a subset or superset thereof, as exemplified below.

[0057] Operating system 451, including system programs for processing various basic system services and performing hardware-related tasks, such as the framework layer, core library layer, and driver layer, which are used to implement various basic services and process hardware-based tasks;

[0058] A network communication module 452 is used to reach other electronic devices via one or more (wired or wireless) network interfaces 420. Exemplary network interfaces 420 include Bluetooth, Wi-Fi, and Universal Serial Bus (USB).

[0059] a presentation module 453 for enabling presentation of information via one or more output devices 431 (e.g., a display screen, a speaker, etc.) associated with the user interface 430 (e.g., a user interface for operating peripheral devices and displaying content and information);

[0060] The input processing module 454 is configured to detect one or more user inputs or interactions from one of the one or more input devices 432 and to translate the detected inputs or interactions.

[0061] In some embodiments, the information processing device provided in the embodiments of the present application can be implemented in software. FIG2 shows an information processing device 455 stored in memory 450. The information processing device 455 can be software in the form of a program or plug-in, and includes the following software modules: a display module 4551, a conversation module 4552, and a jump module 4553. These modules are logical and can be arbitrarily combined or further separated according to the functions implemented. The functions of each module will be described below.

[0062] The information processing method provided in the embodiment of the present application will be described in conjunction with the exemplary application and implementation of the terminal provided in the embodiment of the present application.

[0063] Referring to FIG. 3A , FIG. 3A is a flow chart of the information processing method provided in an embodiment of the present application, which will be described in conjunction with steps 101 to 103 shown in FIG. 3A .

[0064] In step 101, a conversation interface is displayed.

[0065] As an example, the conversation interface here can be a conversation interface between a user account and an AI account.

[0066] In step 102, in response to a message sending operation, a first message is displayed in the conversation interface, and a second message for replying to the first message is displayed in the conversation interface.

[0067] As an example, the second message includes key information for replying to the first message and a reference annotation for each key information, and the key information corresponds to the search results associated with the reference annotation. The key information corresponds to the search results associated with the reference annotation. The search results here are equivalent to the reference source represented by the reference annotation. The key information is directly derived from the search results associated with the reference annotation, or the key information is obtained by semantic generation processing based on the key content in the search results associated with the reference annotation. Referring to Figure 4A, the human-computer interaction interface displays a question 401A and an answer 402A. The answer 402A includes answer content 403A and a reference annotation 404A. The question 401A here is the conversation information carried in the first message in the conversation interface, and the answer 402A here is the conversation information carried in the second message. The answer content 403A is the key information in the conversation information. The key information is the information in the second message used to reply to the first message. The reference annotation 404A corresponding to the answer content 403A can mark the reference source of the answer content 403A.

[0068] As an example, the conversation information in the first message can also be any conversation sentence. For example, the conversation information in the first message can be "I am a little tired", and the conversation information in the second message is used to reply to this sentence in the first message. Natural language processing technology can be used to match the potential needs of this sentence and search based on the potential needs. For example, multiple methods of overcoming fatigue can be searched and obtained, so that the returned second message includes key information about how to overcome fatigue and corresponding reference annotations to represent the reference source of the key information.

[0069] As an example, the conversation information included in the first message may be text information, image information, audio information, video information, etc. Similarly, the conversation information included in the second message may also be text information, image information, audio information, video information, etc.

[0070] In step 103, in response to a trigger operation on the target reference annotation, information at the first position in the target search result is displayed in the reference interface. The information at the first position includes target key information, which is key information corresponding to the target reference annotation.

[0071] As an example, the target search result is a search result that is displayed by jumping after triggering the target reference annotation. For example, the second message includes the target reference annotation A. By triggering the target reference annotation A, you can jump to the web page and display the target key information identified by the target reference annotation on the web page. The web page here is the target search result.

[0072] As an example, when the display size of the reference interface is smaller than a size threshold (for example, a size that can just fully display the information of the first position, such as 3 lines in height and 20 characters in width), the information of the first position in the target search results is displayed in the reference interface; when the display size of the reference interface is larger than the size threshold, the information of the first position in the target search results and information of other positions except the first position, such as information of adjacent positions (the previous row and the next row) of the first position, are displayed in the reference interface.

[0073] In some embodiments, displaying the information of the first position in the target search results in the reference interface in step 103 is achieved by the following technical solution: locating the first position in the target search results in the reference interface and displaying the information of the first position (the information here includes the target key information). For example, the reference interface can be used to display a single page of content in the search results. Usually, when jumping to display the search results, the first page of the search results is displayed first. In the embodiment of the present application, when the target key information is on the fourth line of the second page, the first position here is the fourth line of the second page. In response to the triggering operation for the target reference annotation, the fourth line of the second page is determined by anchor positioning, and then the information of the fourth line of the second page of the target search results is displayed in the reference interface.

[0074] As an example, the target reference annotation comes from at least one reference annotation, the target key information is the key information corresponding to the target reference annotation, the target key information corresponds to the target search result, the target key information comes from the target search result, or the target key information is obtained by semantic generation processing based on the key content in the target search result. Referring to Figure 4A, the human-computer interaction interface displays a question 401A (session information included in the first message) and an answer 402A (session information included in the second message). Answer 402A includes answer content 403A (key information) and a citation annotation 404A. In response to a triggering operation on citation annotation 404A, a reference source window (i.e., a citation interface) pops up in the human-computer interaction interface, and answer content 403A is displayed in the reference source window. The reference source window is used to display the reference source of answer content 403A. The reference source can be an article or a web page. However, after the citation annotation is triggered, the reference source is directly located to the first position related to answer content 403A in the reference source, and the target key information located in the first position in the reference source is displayed. In the case where the display area of ​​the reference interface is insufficient to display all the information in the search results, by locating to the first position, the target key information in the target search result can be displayed in the display area of ​​the reference interface. In the case where the display area of ​​the reference interface is sufficient to display all the information in the search results, by locating to the first position, the target key information in the target search result can be displayed in the central display area of ​​the reference interface.

[0075] As an example, the target key information here is related to the original content of the first position in the target search result. For example, the target key information can be directly copied from the original content of the first position in the target search result, or the target key information can be obtained by summarizing and organizing the original content of the first position in the target search result.

[0076] In some embodiments, the target key information also has another reference annotation. When the information at the first position in the target search result is displayed in the reference interface, a annotation control corresponding to the other reference annotation is displayed; in response to a triggering operation on the annotation control for the other reference annotation, the information at the second position in the other target search result is displayed in the reference interface, wherein the information at the second position includes the target key information, and the other target search result is the search result associated with the source link of the other reference annotation. Through the embodiments of the present application, a quick viewing effect can be achieved, eliminating the need to return to the conversation interface and click to jump, which can improve the efficiency of human-computer interaction.

[0077] As an example, referring to FIG5A , a question 501A (session information included in the first message) and an answer 502A (session information included in the second message) are displayed in the human-computer interaction interface. The answer 502A includes the answer content, a reference annotation 503A corresponding to the answer, and a reference annotation 508A. In response to the triggering operation on the target reference annotation 503A, since the reference source corresponding to the triggered reference annotation 503A is "Flower Growing Encyclopedia", the target key information (key information corresponding to the reference annotation 503A) can be located in the summary card at the first position in the target search result ("Flower Growing Encyclopedia" corresponding to the reference annotation 503A). It is displayed that since the key information corresponding to the reference annotation 503A actually hits two reference sources, namely "Flower Growing Encyclopedia" and "Lao Zhang's Flower Growing", the annotation control 505A corresponding to another reference annotation 508A can be displayed simultaneously in the summary card. In response to the trigger operation of the annotation control 505A for another reference annotation, the target key information (the key information corresponding to the target reference annotation 503A) is located at the second position in another target search result ("Lao Zhang's Flower Growing") corresponding to the reference annotation 508A) for display. The second position here is the position of the reference reference content of the target key information in the other target search result.

[0078] In some embodiments, the second message includes an associated reference annotation with the same annotation sequence number as the target reference annotation. When the information at the first position in the target search result is displayed in the reference interface, a jump control corresponding to the associated reference annotation is displayed; in response to a triggering operation on the jump control, the information at the third position in the target search result is displayed in the reference interface, wherein the information at the third position includes key information corresponding to the associated reference annotation. This embodiment of the present application can achieve a quick viewing effect, eliminating the need to return to the conversation interface to click the jump, thereby improving the efficiency of human-computer interaction.

[0079] As an example, referring to Figure 5A, a question (conversation information included in the first message) and an answer (conversation information included in the second message) are displayed in the human-computer interaction interface. The answer includes key information 510A, key information 511A and reference annotation 509A. In response to the trigger operation on the target reference annotation 509A, the target key information (key information 510A) can be located in the summary card and displayed at the first position in the target search result ("Hua Mu Shuo" corresponding to the target reference annotation 509A). Since the target reference annotation 509A corresponds to both key information 510A and key information 511A, a jump control corresponding to another key information 511A can be displayed at the same time in the summary card. In response to the trigger operation on the jump control on the other key information 511A, the other key information (key information 511A) is located and displayed at the third position in the target search result ("Hua Mu Shuo").

[0080] In some embodiments, the second message further includes a source link for each reference annotation, see FIG. 3B , and step 104 shown in FIG. 3B may also be performed.

[0081] In step 104 , in response to a triggering operation on a source link of any reference annotation, search results associated with the source link of any reference annotation are displayed.

[0082] As an example, the key information corresponding to any reference annotation corresponds to the search result associated with the source link of any reference annotation, indicating that the key information corresponding to any reference annotation is directly derived from the search result associated with the source link of any reference annotation, or the key information corresponding to any reference annotation is obtained by semantic generation processing based on the key content in the search result associated with the source link of any reference annotation. Referring to Figure 4C, the human-computer interaction interface displays a question 401C (session information included in the first message) and an answer 402C (session information included in the second message), and the answer 402C also includes a link 403C of the reference source (the source link of the reference annotation 405C). In response to the trigger operation for the source link 403C, a reference source window pops up in the human-computer interaction interface, and the search result 404C associated with the source link is displayed in the reference source window, that is, the full text of the reference source can be displayed in the reference source window. The source viewing efficiency can be improved through the embodiments of the present application.

[0083] In some embodiments, if the key information corresponding to multiple reference annotations comes from the same search result, the multiple reference annotations have the same annotation serial number; the annotation serial number of any reference annotation is the same as the source link serial number of the source link corresponding to any reference annotation.

[0084] As an example, referring to Figure 4A, the answer in Figure 4A includes two reference annotations with the serial number 2. These two reference annotations have the same citation source, "Lao Zhang Yang Hua," and therefore correspond to the same source link. The answer displays reference annotations and source links. Corresponding reference annotations and source links will have the same serial number. For example, the source link serial number of the source link for "Lao Zhang Yang Hua" is the same as the serial number of the reference annotation corresponding to the key information citing "Lao Zhang Yang Hua." This allows users to directly obtain the correspondence between the key information and the citation source through the same serial number.

[0085] In some embodiments, displaying the second message in reply to the first message in the conversation interface in step 102 can be achieved by the following technical solution: displaying key information in the first message in a third display mode, and displaying other information in the second message other than the key information in a fourth display mode different from the third display mode. This embodiment of the present application allows users to quickly focus on key information, thereby improving the efficiency of human-computer interaction.

[0086] As an example, for the display of the second message, key information and non-key information can be displayed in a differentiated display manner. Key information is information with a reference source used to reply to the first message, while non-key information is information that exists to make the second message have a complete and coherent expression of meaning. Here, both key information and non-key information can be displayed in the fourth display manner. In response to a mouse hover operation on any reference mark, the key information corresponding to any reference mark is displayed based on the third display manner. In response to the end of the mouse hover operation, the fourth display manner is restored to display the key information corresponding to any reference mark.

[0087] In some embodiments, referring to FIG. 3C , step 102 of displaying the first message in the conversation interface in response to the message sending operation may be implemented through steps 1021 and 1022 shown in FIG. 3C .

[0088] In step 1021 , at least one associative information is displayed in the conversation interface, wherein the associative information is related to the conversation information included in the most recent historical messages in the conversation interface.

[0089] As an example, the most recent historical message here refers to the last message displayed in the conversation interface. Referring to FIG4D , the human-computer interaction interface displays a question 401D and an answer 402D. The answer 402D here is the most recent historical message.

[0090] In step 1022 , in response to a trigger operation on any associated information, the trigger operation is identified as a message sending operation, and any associated information is carried in the first message for display.

[0091] As an example, referring to FIG4D , the human-computer interaction interface also displays an associative question 403D (associative information). Associative question 403D is obtained by associating question 401D and answer 402D. Here, the first historical message and the second historical message are recent historical messages. In response to a trigger operation for associative question 403D, the trigger operation is identified as a message sending operation, and any associative information is carried in the first message for display, that is, associative question 403D is displayed in the human-computer interaction interface. Through the embodiments of the present application, conversation guidance can be performed to improve the efficiency of human-computer interaction.

[0092] In some embodiments, displaying the information at the first position of the target search result in the reference interface in step 103 can be achieved by the following technical solution: displaying the target key information in a first display mode; and displaying other information in the first position, except for the target key information, in a second display mode different from the first display mode. Through the embodiments of the present application, users can quickly focus on key information, thereby improving the efficiency of human-computer interaction.

[0093] As an example, when displaying the information in the first position of the target search results in the reference interface, the target key information and other information can be displayed in a differentiated manner. The target key information is the key information corresponding to the triggered reference annotation, and the other information is the information in the target search results that serves as the reference source other than the target key information.

[0094] In some embodiments, displaying the first position information of the target search result in the reference interface in step 103 can be achieved by the following technical solutions: displaying a portion of the target search result; and in response to an expansion operation on the target search result, displaying the first position information of the target search result in the reference interface. This embodiment of the present application can save display resources and improve display resource utilization.

[0095] As an example, referring to Figure 5B, the human-computer interaction interface displays question 501B (session information included in the first message) and answer 502B (session information included in the second message). Answer 502B includes answer content (key information) and reference annotation 503B. In response to the trigger operation on reference annotation 503B, the reference source "Huamu Shuo" corresponding to reference annotation 503B is displayed in the summary card. The reference source here is the target search result. Only part of the target search result information is displayed here. An expand full-text control 504B is also displayed in the summary card. In response to the trigger operation (expand operation) on the expand full-text control 504B, the full text is expanded and then anchor alignment is performed to locate the fragment position (first position) of the corresponding answer content (target key information) in the full text of the reference source "Huamu Shuo". The summary card in Figure 5B can be displayed on the right side of the conversation interface as shown in Figure 5A.

[0096] In some embodiments, when displaying part of the target search result, when the number of reference annotations of the target key information is multiple, the target annotation control and other annotation controls in the selected state are displayed in the reference interface displaying the target search result, wherein the control serial number of the target annotation control is the same as the annotation serial number of the triggered target reference annotation; when the other annotation controls are in the selected state, part of the content of another target search result is displayed, wherein the control serial number of the other annotation controls is the same as the annotation serial number of another reference annotation of the target key information, and the target key information corresponds to the other target search result. Through the embodiments of the present application, fast switching can be performed to improve query efficiency.

[0097] As an example, referring to FIG5A , a question 501A and an answer 502A are displayed in the human-computer interaction interface. The answer 502A includes the answer content and a reference mark 503A. In response to a trigger operation on the reference mark 503A (or in response to the mouse hovering over the reference mark 503A, or in response to the mouse hovering over the answer content corresponding to the reference mark 503A), a summary card 507A pops up on the right side of the conversation interface and highlights the answer content corresponding to the reference mark 503A in the answer 502A. The summary card 507A displays the reference source of the answer content corresponding to the reference mark 503A. The reference source is not directly located here. At the position related to the answer content in the reference source, answer 502 actually hits two reference sources, namely "Flower Growing Encyclopedia" and "Lao Zhang Growing Flowers". Since the reference source corresponding to the triggered reference annotation 503A is "Flower Growing Encyclopedia", the reference source serial number 504A corresponding to "Flower Growing Encyclopedia" in the summary card 507A is in a selected state, that is, the reference source currently displayed in the summary card 507A is "Flower Growing Encyclopedia". In response to the triggering operation on the page turning control 506A, the adjacent reference source serial number 505A is in a selected state, and the application source "Lao Zhang Growing Flowers" corresponding to the reference source serial number 505A is displayed in the summary card 507A.

[0098] In some embodiments, referring to FIG. 3D , steps 105 to 107 may be further performed before displaying the second message for replying to the first message in the conversation interface.

[0099] In step 105, semantic decomposition processing is performed on the session information included in the first message to obtain query information matching the first message.

[0100] As an example, semantic decomposition processing can be performed here by a semantic generation model, or by natural language processing technology. When performing semantic decomposition processing by a semantic generation model, the following task template can be used to assign tasks to the generative language model: "You are a query understanding assistant. According to my task, list queries suitable for retrieval. Each query is output in a pair of {}. Task: Identify the most important requirement when the user searches for "%s", and only output one retrieval query that can meet this requirement." Where "%s" is replaced with the question input by the user, and the query statement corresponding to the question is generated by the generative language model. By inputting the above task template into the above semantic generation model, the semantic generation model can output query information that matches the first message. For example, the first message may be "I really want to learn frisbee", and the query information that matches the first message may be "How to learn frisbee".

[0101] In step 106, a search process is performed based on the query information to obtain a plurality of candidate search results.

[0102] In some embodiments, in step 106, search processing is performed based on the query information to obtain at least one candidate search result, which can be achieved through the following technical solution: calling a search engine to obtain multiple original search results that match the query information; based on the relevance of each original search result to the first message, the multiple original search results are sorted in descending order, and at least one search result with a higher ranking is used as a candidate search result.

[0103] In some embodiments, based on the relevance of each original search result to the first message, before sorting the multiple original search results in descending order, the following processing is performed on each original search result: the conversation information and query information included in the first message are segmented to obtain multiple first segmentation results, and the multiple first segmentation results are query understood to obtain first understanding information; each fragment of the original search result is segmented to obtain multiple second segmentation results, and the multiple second segmentation results are query understood to obtain second understanding information for each fragment; the understanding query relevance between the first understanding information and each second understanding information is determined; and the highest understanding query relevance is used as the relevance of the original search result to the first message.

[0104] As an example, during the search result relevance filtering phase, natural language processing can be used to understand the question entered by the user and the query extracted by the generative language model. The understanding capabilities of natural language processing technology are then invoked to perform word segmentation, word weighting, and closeness calculations on the question (session information) and query to obtain first understanding information. Word segmentation, word weighting, and closeness calculations are also performed on the search result title and each full-text segment (the original search result) to obtain second understanding information. The word weighting and closeness calculations described here are technical means of the query understanding branch within the search scenario. Based on this, text relevance or vector semantic relevance calculations can be used to determine the relevance between the question and each search result segment (understanding query relevance). The relevance of the entire search result is determined by the maximum relevance among the included segments, and the entire search result with high relevance is retained. Generative language models can also be used to determine the relevance of each search result segment, such as using the following prompt template: "Do you think "%answer" is the correct answer to the question "%quest"? If so, return "yes"; if not, return "no"."

[0105] In step 107, at least one key content related to the first message is extracted from the plurality of candidate search results, and text generation processing is performed on the at least one key content to obtain a second message for replying to the first message.

[0106] In some embodiments, each key information corresponds to at least one key content. Extracting at least one key content related to the first message from multiple candidate search results in step 107 can be achieved through the following technical solution: performing the following processing on each candidate search result: segmenting the candidate search result to obtain multiple segments; obtaining the semantic relevance of each segment to the first message; sorting the multiple segments in descending order based on the semantic relevance of the multiple segments, obtaining at least one segment with the highest ranking as a candidate segment; extending each candidate segment before and after a set window length in the candidate search results corresponding to the candidate segment, obtaining an extended segment of each candidate segment, and using the extended segment as the key content of the corresponding candidate search result. Through the embodiments of the present application, the relevance of the key content to the question can be improved, thereby improving the accuracy of the answer.

[0107] For example, during the search result fragment extraction phase, when calculating the relevance between the search results and the input question, we can obtain fragments with a high relevance to the input question within the full text of the search results. Each fragment is then expanded within the original text to a desired length window, with the preceding and following text extended to obtain the expanded fragments. The relevance score between the pre-expanded fragment and the input question is used as the relevance score for the expanded fragment. After obtaining the relevant expanded fragments for multiple search results, they are sorted by relevance to obtain the search result fragments with the highest relevance.

[0108] In some embodiments, the above-mentioned text generation processing of at least one key content to obtain a second message for replying to the first message can be achieved through the following technical solution: combining at least one key content to obtain a combined result; performing natural language text generation processing on the combined result to obtain a second message for replying to the first message.

[0109] As an example, in the search result fragment splicing stage: set the number of fragments extracted from each search result, retain the fragments of each search result in descending order of relevance, and number the search results corresponding to the fragments, such as the fragment of the first search result is numbered 1, and the fragments are numbered in sequence. Splice the text fragments together, and add the numbers of the corresponding fragment texts to the spliced ​​content. For example, fragment 1 (content is xxxxxxx) and fragment 2 (content is yyyyyy) are extracted from article 1, and fragment 3 (content is zzzzzz) is extracted from article 2. The spliced ​​content is xxxxxxx[1]yyyyyy[1]zzzzzz[2]. The combined content here is actually the combined result of the key content. Through natural language text generation processing, the key content can be converted into key information, making the answer personified and in line with normal language expression habits.

[0110] As an example, natural language text generation processing can be implemented through a generative language model: the combined result is input into the generative language model for answer generation. The task assignment template that can be used is as follows: "Generate an answer of no more than 400 words based on the user search term "%s" and the following paragraphs, and use the form of \"[Arabic numerals]\" to indicate which paragraph each sentence is from. Only give citations and do not provide annotations. Paragraphs with poor relevance should not be cited, and "%s" is replaced with user-input questions"; or "Extract information based on the content below, with the goal of answering this question "%s" as perfectly as possible. Discard irrelevant information. The semantics are smooth and the format is clear. Use the form of \"[number]\" to indicate which paragraph each sentence in your reply is from. Only give citations and do not provide annotations. Paragraphs with poor relevance should not be cited."

[0111] In some embodiments, the above-mentioned combination processing of at least one key content to obtain a combined result can be achieved through the following technical solutions: numbering the candidate search results corresponding to each key content to obtain a search result number of each candidate search result; splicing at least one key content to obtain a splicing result, and adding a corresponding reference annotation for each key content in the splicing result; obtaining the position mark information corresponding to the key content in the candidate search results of the source, adding the position mark information to the reference annotation of the corresponding key content to obtain a combined result; wherein the annotation serial number of the reference annotation is the same as the search result number of the candidate search result of the source of the corresponding key content.

[0112] As an example, click on the reference mark to jump and align the anchor point to the corresponding fragment position (the location of the reference content of the key information). According to the mapping relationship between the annotation serial number of the recorded reference mark and the corresponding article (search result), combined with the position of the reference fragment (key information) in the article, the reference mark is actually the article URL link. The corresponding position mark information can be spliced ​​on the article URL link as the HTML tag of the reference mark. When you click on the reference mark, you can jump to the corresponding fragment position.

[0113] The following describes an exemplary application of the embodiments of the present application in a practical application scenario.

[0114] In some embodiments, the terminal of the target object displays a conversation interface and receives a message sending operation, displays a first message in the conversation interface of the terminal, the terminal sends the first message to the server, the server obtains a second message for replying to the first message, the second message includes key information for replying to the first message and a reference markup for each key information, the server returns the second message to the terminal, and displays the second message in the conversation interface of the terminal; the terminal receives a trigger operation for the target reference markup, generates a source viewing request based on the trigger operation, and sends the source viewing request to the server, the server obtains the target search result corresponding to the target reference markup, and obtains the first position of the key information corresponding to the target reference markup in the target search result, the server returns the first position and the target search result to the terminal, and displays the information of the first position in the target search result in the reference interface of the terminal.

[0115] ​In some embodiments, a question is input through a dialogue entry, and a generated answer is returned after background processing. The reference source of the corresponding answer content is marked on the answer text, and the reference source is attached after the answer. At the same time, the question and answer are actively extended below the answer to provide a guidance prompt for the next question. In response to the trigger operation for the reference source, the corresponding article link can be opened, and the jump anchor point can be aligned to the position of the relevant referenced fragment for highlighting. In response to the trigger for the reference source, the full text of the reference source article is directly opened. In response to the trigger operation for the next question that is actively extended, a new round of dialogue process is carried out.

[0116] In some embodiments, referring to FIG4A , a question 401A and an answer 402A are displayed in the human-computer interaction interface. The answer 402A includes answer content 403A and a reference mark 404A. In response to a trigger operation on the reference mark 404A, a reference source window pops up in the human-computer interaction interface, and the answer content 403A is displayed in a highlighted manner (e.g., highlighted) in the reference source window. Here, the reference source window can actually be used to display the source article of the answer content 403A, but after triggering the reference mark, it directly locates the location of the answer content 403A in the source article for display.

[0117] In some embodiments, referring to FIG4B , a question 401B and an answer 402B are displayed in the human-computer interaction interface, and the answer 402B includes answer content 403B and a reference mark 404B. In response to a trigger operation on the reference mark 404B, a reference source window pops up in the human-computer interaction interface, and the answer content 403B is displayed in a highlighted manner (e.g., highlighted) in the reference source window. Here, the reference source window can actually be used to display the source article of the answer content 403B, but after triggering the reference mark, it directly locates the location of the answer content 403B in the source article for display.

[0118] In some embodiments, referring to FIG4C , the human-computer interaction interface displays a question 401C and an answer 402C, and the answer 402C also includes a source link 403C of the reference source. In response to a trigger operation on the link 403C, a reference source window pops up in the human-computer interaction interface, and the search result 404C associated with the source link 403C is displayed in the reference source window, that is, the full text of the reference source can be displayed in the reference source window.

[0119] In some embodiments, referring to FIG4D , a question 401D and an answer 402D are displayed in the human-computer interaction interface. An associative question 403D is also displayed in the human-computer interaction interface. The associative question 403D is obtained by associating the question 401D and the answer 402D. In response to a triggering operation on the associative question 403D, the associative question 403D is displayed in the human-computer interaction interface, and an associative answer 404D corresponding to the associative question 403D is displayed.

[0120] In some embodiments, referring to FIG5A , a question 501A and an answer 502A are displayed in the human-computer interaction interface. The answer 502A includes the answer content and a reference mark 503A. In response to a trigger operation on the reference mark 503A (or in response to the mouse hovering over the reference mark 503A, or in response to the mouse hovering over the answer content corresponding to the reference mark 503A), a summary card 507A pops up on the right side of the conversation interface and highlights the answer content corresponding to the reference mark 503A in the answer 502A. The summary card 507A displays the reference source of the answer content corresponding to the reference mark 503A. There is no direct definition here. The answer 502 actually hits two reference sources, namely "Flower Growing Encyclopedia" and "Lao Zhang Growing Flowers". Since the reference source corresponding to the triggered reference annotation 503A is "Flower Growing Encyclopedia", the reference source serial number 504A corresponding to "Flower Growing Encyclopedia" in the summary card 507A is in a selected state, that is, the reference source currently displayed in the summary card 507A is "Flower Growing Encyclopedia". In response to the triggering operation on the page turning control 506A, the adjacent reference source serial number 505A is in a selected state, and the application source "Lao Zhang Growing Flowers" corresponding to the reference source serial number 505A is displayed in the summary card 507A.

[0121] In some embodiments, referring to FIG5B , a question 501B and an answer 502B are displayed in the human-computer interaction interface, and the answer 502B includes the answer content and a reference mark 503B. In response to a trigger operation on the reference mark 503B, the reference source "Hua Mu Shuo" corresponding to the reference mark 503B is displayed in the summary card, and an expand full-text control 504B is displayed in the summary card. In response to a trigger operation on the expand full-text control 504B, the full text is expanded and then anchor alignment is performed to locate the fragment position of the corresponding answer content in the full text of the reference source "Hua Mu Shuo". The summary card in FIG5B can be displayed on the right side of the conversation interface as shown in FIG5A .

[0122] In some embodiments, referring to FIG5C , a question 501C and an answer 502C are displayed in the human-computer interaction interface. The answer 502C includes a reference source 503C. In response to a trigger operation on the reference source 503C, the full text of the reference source is displayed.

[0123] In some embodiments, referring to FIG6 , first, the terminal sends the question input by the user to the conversation server chatsvr; chatsvr calls the generative language model to extract the query statement that matches the question from the question; calls the search engine system to query based on the extracted query statement to obtain the search results, and selects the head search results from all the search results; calculates the relevance between the full text of the title of the head search result and the question, and retains the most relevant search results; extracts the full text content of the retained head search results, and each head search result can extract multiple fragments; splices the multiple fragments extracted from multiple articles according to the template for generating summaries, and marks the fragments with the annotation serial number of the original text reference annotation to form a summary. To input text; when marking the annotation serial number of the original text reference annotation, record the original text segment and original text link corresponding to each number; use the summary input text as the input of the generative language model, and the generative language model generates the corresponding answer summary content, and returns the corresponding annotation serial number of the reference annotation in the answer summary content; extract the text content corresponding to the annotation serial number of the reference annotation in the answer output by the generative language model, and perform text semantic matching based on the original text segment corresponding to the annotation serial number of the recorded reference annotation. If there is a match, retain the reference annotation, and add the corresponding <a href=”https: / / xxx”>html tag to the corresponding reference annotation plain text to add the reference source link; use the retained header search result as the reference source, and add the corresponding html tag in the answer Tags form hyperlink text; based on the question and answer, the generative language model is called to obtain the next step of question and answer extension.

[0124] In some embodiments, the specific method for extracting a query statement that matches the question is as follows. The following prompt template can be used to assign a task to the generative language model: "You are a query understanding assistant. According to my task, list queries suitable for retrieval. Each query is output in a pair of {}. Task: Identify the most important requirement when the user searches for "%s" and output only one retrieval query that can meet this requirement." Where "%s" is replaced with the question entered by the user, and the query statement corresponding to the question is generated by the generative language model.

[0125] In some embodiments, during the search result relevance filtering stage, a natural language processing understanding method can be used to call upon the understanding capabilities of natural language processing technology for the questions input by the user and the query statements extracted by the generative language model, perform word segmentation, word weight and closeness calculations on the questions and query statements, and also perform word segmentation, word weight and closeness calculations on the search result title and each fragment of the full text to obtain understanding information at the word segmentation level of the query and the original text. On this basis, text relevance or vector semantic relevance calculations can be used to obtain the relevance between the question and each fragment of the search result. The relevance of the entire search result is determined by the maximum value of the relevance among the included fragments, and the entire search result with high relevance is retained. A generative language model can also be used to judge the relevance of each fragment of the search result, such as using the following prompt template: "Do you think "%answer" is the correct answer to the question "%quest"? If so, please return "yes", if not, please return "no".

[0126] In some embodiments, during the search result fragment extraction phase, when calculating the relevance between the search results and the input question, fragments with a high relevance between the full text of the search results and the input question can be obtained. Each fragment is then expanded within the original text by a pre-set length window to obtain the expanded fragments. The relevance score between the pre-expanded fragment and the input question is used as the relevance of the expanded fragment to the input question. After obtaining the relevance expanded fragments of multiple search results, they are sorted by relevance value to obtain the search result fragments with the highest relevance.

[0127] In some embodiments, during the search result fragment splicing stage: the number of fragments extracted from each search result is set, and the fragments of each search result are retained in descending order of relevance, and the search results corresponding to the fragments are numbered accordingly, such as the fragment of the first search result is numbered 1, and the fragments are numbered in sequence thereafter. The fragment texts are spliced ​​together, and the numbers of the corresponding fragment texts are added to the spliced ​​content. For example, fragment 1 (content is xxxxxxx) and fragment 2 (content is yyyyyy) are extracted from article 1, and fragment 3 (content is zzzzzz) is extracted from article 2, then the spliced ​​content is xxxxxxx[1]yyyyyy[1]zzzzzz[2].

[0128] In some embodiments, during the answer generation phase of the generative language model, the concatenated content is input to the generative language model for answer generation. The following task arrangement template may be used:

[0129] Generate an answer of no more than 400 words based on the user search term "%s" and the following paragraphs. Use "[Arabic numerals]" to indicate the source of each sentence. Only provide citations, not annotations. Paragraphs with poor relevance should not be cited. "%s" is replaced with the question entered by the user.

[0130] "Based on the following content, extract information with the goal of answering the question "%s" as perfectly as possible. Discard irrelevant information. Make the text coherent and the format clear. Use \"[number]\" to indicate the source of each sentence in your response. Only provide citations, not annotations. Paragraphs with poor relevance should not be cited."

[0131] In some embodiments, during the question extension stage, the following task assignment template may be used: "You are a query understanding assistant. List query statements suitable for retrieval according to the task. Each query statement is output in a pair of {}. Task: Identify the description subject and related popular needs when the user searches for "%s", ask what questions can meet this need, and list 3 query statements."

[0132] In some embodiments, clicking on the reference mark jumps and aligns the anchor point to the corresponding fragment position. Based on the mapping relationship between the annotation serial number of the recorded reference mark and the corresponding article, and combined with the position of the reference fragment in the article, the corresponding position mark information can be added to the article URL link as the HTML tag of the reference mark. When clicking on the reference mark, you can jump to the corresponding fragment position.

[0133] ​In some embodiments, see Figure 7, during the interaction process, since the overall process takes a long time, the background triggers the entire answer generation process after the user enters the dialogue, and at the same time, the client and the background use an asynchronous polling process to check whether the background has generated the answer. In implementation, after a user enters a question, Chatsvr calls UserWeb to send the first message to Chatsvr, which then returns the key data corresponding to the first message to the conversation page. After sending the first message to Chatsvr, Chatsvr performs the following processing: 1. Chatsvr calls the generative language model proxy to rewrite the conversation information in the first message into query information; 2. Chatsvr calls a search engine to perform a search based on the query information and obtain search results; 3. Chatsvr extracts TopK article fragments based on the full text of the search results; 4. Chatsvr inputs the extracted fragments into the generative language model proxy to obtain a summary for replying; 5. Chatsvr obtains the fragment location information required for rich text based on the full text segment location information; 6. Chatsvr generates rich text information based on the determined location information and stores the rich text information as a key-value pair with the key data of the first message. The conversation page then queries the value data corresponding to the key data through WebCgl through asynchronous polling and returns the result to the conversation page.

[0134] After generating the answer,<user,questionid> The answer is written into the KV storage as the key and the key is returned to the client. The client calls WebCGI with the key at regular intervals to poll for the corresponding answer, and the WebCGI backend calls KV to obtain the generated answer.

[0135] The embodiment of the present application generates a final answer based on the fragments extracted from the search results with high relevance, thereby improving the accuracy of the generated answer and providing corresponding citation annotations and reference sources to further confirm the reliability of the answer. By extracting and splicing the fragments of the full text of the search results and then calling the generative language model, the number of input tokens of each interaction of the generative language model can be reduced, and as many result fragments as possible can be input to the generative language model to ensure the input richness of the generative language model. The answer is generated based on the extracted content fragments, and the extracted content fragments are controllable, so as to ensure that the content of the generated answer is not highly summarized by the generative language model as much as possible.

[0136] In an embodiment of the present application, a generative language model is used to generate a query prompt for an input question to obtain a core query statement. Alternatively, natural language understanding technology may be used to semantically understand, extract, or rewrite the input question to obtain a query statement. In an embodiment of the present application, a generative language model is used to generate a query prompt for an input question to obtain a query statement. Each query statement is searched and then aggregated to obtain highly relevant search results. In an embodiment of the present application, direct text splicing is used to obtain the input for generating the answer. Alternatively, a more structured splicing method may be used.

[0137] The embodiment of the present application proposes an interactive dialogue product that combines search with a generative language model. The user submits a question, and the client obtains the final answer through background processing and returns it to the user, and at the same time gives the reference source in the answer description; the user can click on the reference source to pop up the reference source article page in the dialog box and anchor it to the specific location of the article; at the same time, the question and answer are combined to further extend related questions, and the user clicks to trigger a new question-and-answer dialogue process. During the background processing, the user's question is rewritten into a search query statement, and the top few results that are most relevant to the question are extracted after retrieval. After extracting fragments, they are spliced ​​into complete texts for the generative language model to generate the final answer and make reference annotations. The embodiment of the present application upgrades the plain text dialogue form of the generative language model to a rich text form, and at the same time, combines it with the search results and gives corresponding reference annotations to improve the standardization and credibility of the generated answers.

[0138] It is understandable that in the embodiments of the present application, when user information and other related data are involved, when the embodiments of the present application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0139] The following continues to describe an exemplary structure of the information processing device 455 provided in an embodiment of the present application implemented as a software module. In some embodiments, as shown in Figure 2, the software modules stored in the information processing device 455 of the memory 450 may include: a display module 4551, configured to display a conversation interface; a conversation module 4552, configured to display a first message in the conversation interface in response to a message sending operation, and display a second message for replying to the first message in the conversation interface; wherein the second message includes key information for replying to the first message and a reference annotation for each key information, and the key information corresponds to the search result associated with the reference annotation; a jump module 4553, configured to display a reference interface in response to a trigger operation for a target reference annotation and display the information of the first position in the target search result in the reference interface, wherein the target key information is the key information corresponding to the target reference annotation, and the target key information corresponds to the target search result corresponding to the target reference annotation.

[0140] In some embodiments, the target key information also has another reference annotation. When the information at the first position in the target search result is displayed in the reference interface, the jump module 4553 is also configured to: display an annotation control corresponding to another reference annotation; in response to a trigger operation on the annotation control of another reference annotation, display the information at the second position in another target search result in the reference interface, wherein the information at the second position includes the target key information, and the other target search result is a search result associated with the source link of another reference annotation.

[0141] In some embodiments, when the information at the first position in the target search result is displayed in the reference interface, the jump module 4553 is also configured to: display a jump control corresponding to the associated reference annotation; in response to a trigger operation on the jump control, display the information at the third position in the target search result in the reference interface, wherein the information at the third position includes key information corresponding to the associated reference annotation.

[0142] In some embodiments, the second message also includes a source link for each reference annotation, and the jump module 4553 is further configured to: in response to a trigger operation on the source link of any reference annotation, display the search results associated with the source link of any reference annotation; wherein the key information corresponding to any reference annotation corresponds to the search results associated with the source link of any reference annotation.

[0143] In some embodiments, if the key information corresponding to multiple reference annotations comes from the same search result, the multiple reference annotations have the same annotation serial number; the annotation serial number of any reference annotation is the same as the source link serial number of the source link corresponding to any reference annotation.

[0144] In some embodiments, the jump module 4553 is further configured to: display the target key information in a first display mode; and display other information in the information at the first position except the target key information in a second display mode different from the first display mode.

[0145] In some embodiments, the conversation module 4552 is further configured to display key information in the first message in a third display mode, and to display other information in the second message except the key information in a fourth display mode different from the third display mode.

[0146] In some embodiments, the conversation module 4552 is further configured to display at least one associative information in the conversation interface, wherein the associative information is related to the conversation information included in the most recent historical messages in the conversation interface; in response to a trigger operation for any associative information, the trigger operation is identified as a message sending operation, and any associative information is carried in the first message for display.

[0147] In some embodiments, the jump module 4553 is further configured to: display part of the target search result; and in response to an expansion operation on the target search result, display the information of the first position in the target search result in the reference interface.

[0148] In some embodiments, when partial content of the target search result is displayed, the jump module 4553 is further configured to: when the number of reference annotations of the target key information is multiple, in the reference interface displaying the target search result, the target annotation control and other annotation controls in the selected state are displayed, wherein the control serial number of the target annotation control is the same as the annotation serial number of the triggered target reference annotation; when other annotation controls are in the selected state, partial content of another target search result is displayed, wherein the control serial number of the other annotation control is the same as the annotation serial number of another reference annotation of the target key information, and the target key information corresponds to another target search result.

[0149] In some embodiments, before displaying the second message used to reply to the first message in the conversation interface, the conversation module 4552 is further configured to perform semantic decomposition processing on the conversation information included in the first message to obtain query information matching the first message; perform search processing based on the query information to obtain multiple candidate search results; extract at least one key content related to the first message from the multiple candidate search results, and perform text generation processing on the at least one key content to obtain the second message used to reply to the first message.

[0150] In some embodiments, the conversation module 4552 is further configured to call a search engine to obtain multiple original search results that match the query information; based on the relevance of each original search result to the first message, the multiple original search results are sorted in descending order, and at least one search result with a higher ranking is used as a candidate search result.

[0151] In some embodiments, before sorting the multiple original search results in descending order based on the relevance of each original search result to the first message, the conversation module 4552 is also used to perform the following processing on each original search result: performing word segmentation processing on the conversation information and query information included in the first message to obtain multiple first word segmentation results, and performing query understanding processing on the multiple first word segmentation results to obtain first understanding information; performing word segmentation processing on each fragment of the original search result to obtain multiple second word segmentation results, and performing query understanding processing on the multiple second word segmentation results to obtain second understanding information for each fragment; determining the understanding query relevance between the first understanding information and each second understanding information; and taking the highest understanding query relevance as the relevance of the original search result to the first message.

[0152] In some embodiments, the conversation module 4552 is further configured to perform the following processing for each candidate search result: segment the candidate search result to obtain multiple segments; obtain the semantic relevance of each segment to the first message; sort the multiple segments in descending order based on the semantic relevance of the multiple segments to obtain at least one segment with the highest ranking as a candidate segment; expand each candidate segment in the candidate search result of the corresponding candidate segment before and after a set window length to obtain an extended segment of each candidate segment, and use the extended segment as the key content of the corresponding candidate search result.

[0153] In some embodiments, the conversation module 4552 is further configured to perform combination processing on at least one key content to obtain a combination result; and perform natural language text generation processing on the combination result to obtain a second message for replying to the first message.

[0154] In some embodiments, the session module 4552 is further configured to number the candidate search results corresponding to each key content to obtain a search result number for each candidate search result; perform splicing processing on at least one key content to obtain a splicing result, and add a corresponding reference annotation for each key content in the splicing result; obtain the position mark information corresponding to the key content in the candidate search results of the source, add the position mark information to the reference annotation of the corresponding key content, and obtain a combined result; wherein the annotation serial number of the reference annotation is the same as the search result number of the candidate search result of the source of the corresponding key content.

[0155] The present invention provides a computer program product or computer program, which includes computer-executable instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer-executable instructions from the computer-readable storage medium and executes the computer-executable instructions, causing the computer device to perform the information processing method described in the present invention.

[0156] An embodiment of the present application provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, the processor will execute the information processing method provided by the embodiment of the present application, for example, the information processing method shown in Figures 3A-3D.

[0157] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface storage, optical disk, or CD-ROM; or various devices including one or any combination of the above memories.

[0158] In some embodiments, computer-executable instructions may be in the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0159] As an example, computer-executable instructions may, but need not, correspond to a file in a file system, may be stored as part of a file that stores other programs or data, such as in one or more scripts in a HyperText Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple coordinating files (e.g., files storing one or more modules, subroutines, or code portions).

[0160] By way of example, computer-executable instructions may be deployed to be executed on one electronic device, or on multiple electronic devices located at one site, or on multiple electronic devices distributed across multiple sites and interconnected by a communication network.

[0161] To sum up, through the embodiment of the present application, when automatically replying to the first message, the reference mark of the key information can be displayed in the second message, so that the user can view the search results of the source of the key information through the reference mark, and can locate the position of the key information in the search results for display, so that the user does not need to search up and down in the search results for key information, thereby improving the efficiency of information acquisition and human-computer interaction.

[0162] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. Any modifications, equivalent replacements, and improvements made within the spirit and scope of the present application are included in the scope of protection of the present application. ​

Claims

1. An information processing method, the method being performed by an electronic device, the method comprising: In response to the message sending operation, displaying a first message in the conversation interface, and displaying a second message for replying to the first message in the conversation interface; The second message includes key information used to reply to the first message and a reference mark of each key information, and the key information corresponds to the search result associated with the reference mark; In response to a trigger operation for a target reference annotation, displaying information at the first position in the target search result in a reference interface; The information of the first position includes the target key information, and the target key information is the key information corresponding to the target reference annotation.

2. The method according to claim 1, wherein: The second message also includes a source link of each of the reference annotations, and the method further includes: In response to a trigger operation on a source link of any reference annotation, displaying search results associated with the source link of any reference annotation; The key information corresponding to any of the reference annotations corresponds to the search results associated with the source link of any of the reference annotations.

3. The method according to claim 2, wherein: If the key information corresponding to the multiple reference annotations is derived from the same search result, the multiple reference annotations have the same annotation serial number; The mark number of any reference mark is the same as the source link number of the source link of any reference mark.

4. The method according to claim 1, wherein: The second message includes an associated reference annotation having the same annotation sequence number as the target reference annotation, and when information of the first position in the target search result is displayed in the reference interface, the method further includes: Display a jump control corresponding to the associated reference annotation; In response to a trigger operation on the jump control, information at a third position in the target search result is displayed in the reference interface, wherein the information at the third position includes the key information corresponding to the associated reference annotation.

5. The method according to claim 1, wherein: The target key information also has another reference mark. When the information at the first position in the target search result is displayed in the reference interface, the method further includes: Display a callout control corresponding to another reference callout; In response to a triggering operation on an annotation control of the other reference annotation, information of a second position in another target search result is displayed in the reference interface, wherein the information of the second position includes the target key information, and the other target search result is a search result associated with a source link of the other reference annotation.

6. The method according to claim 1, wherein: The displaying of the first position information in the target search result in the reference interface includes: Displaying the target key information in a first display mode; The other information in the information of the first position except the target key information is displayed in a second display manner different from the first display manner.

7. The method according to claim 1, wherein: Displaying a second message for replying to the first message in the conversation interface includes: The key information in the first message is displayed in a third display mode, and other information in the second message except the key information is displayed in a fourth display mode different from the third display mode.

8. The method according to claim 1, wherein: The step of displaying the first message in the conversation interface in response to the message sending operation includes: Displaying at least one associative information in the conversation interface, wherein the associative information is related to the conversation information included in the most recent historical message in the conversation interface; In response to a trigger operation for any associative information, the trigger operation is identified as the message sending operation, and the any associative information is carried in the first message for display.

9. The method according to claim 1, wherein: The displaying of the first position information in the target search result in the reference interface includes: Displaying part of the target search results; In response to an expansion operation on the target search result, information of the first position in the target search result is displayed in a reference interface.

10. The method according to claim 9, wherein: When displaying part of the target search result, the method further includes: When there are multiple reference annotations of the target key information, in the reference interface displaying the target search results, a target annotation control in a selected state and other annotation controls are displayed, wherein the control sequence number of the target annotation control is the same as the annotation sequence number of the triggered target reference annotation; When the other annotation control is in a selected state, part of the content of another target search result is displayed, wherein the control serial number of the other annotation control is the same as the annotation serial number of another reference annotation of the target key information, and the target key information corresponds to the other target search result.

11. The method according to any one of claims 1 to 10, wherein: Before displaying a second message for replying to the first message in the conversation interface, the method further includes: Performing semantic decomposition processing on the session information included in the first message to obtain query information matching the first message; Perform search processing based on the query information to obtain multiple candidate search results; At least one key content related to the first message is extracted from the multiple candidate search results, and text generation processing is performed on the at least one key content to obtain a second message for replying to the first message, wherein each key information corresponds to at least one key content.

12. The method according to claim 11, wherein: The performing search processing based on the query information to obtain at least one candidate search result includes: Invoke a search engine to obtain a plurality of original search results matching the query information; Based on the relevance of each of the original search results to the first message, the multiple original search results are sorted in descending order, and at least one search result with a higher order is used as a candidate search result.

13. The method according to claim 12, wherein: Before sorting the plurality of original search results in descending order based on the relevance of each original search result to the first message, the method further includes: The following processing is performed for each of the original search results: Performing word segmentation processing on the session information and the query information included in the first message to obtain a plurality of first word segmentation results, and performing query understanding processing on the plurality of first word segmentation results to obtain first understanding information; Performing word segmentation processing on each fragment of the original search result to obtain a plurality of second word segmentation results, and performing query understanding processing on the plurality of second word segmentation results to obtain second understanding information of each fragment; determining a comprehension query relevance between the first comprehension information and each of the second comprehension information; The highest understanding query relevance is used as the relevance of the original search result to the first message.

14. The method according to claim 11, wherein: The step of extracting at least one key content related to the first message from the plurality of candidate search results includes: The following processing is performed for each candidate search result: Performing segmentation processing on the candidate search results to obtain the multiple segments; Obtaining the semantic relevance between each of the segments and the first message; Sorting the multiple segments in descending order based on the semantic relevance of the multiple segments, and obtaining at least one segment with a top ranking as a candidate segment; Each candidate segment is subjected to a front-back extension process of a set window length in a candidate search result corresponding to the candidate segment to obtain an extended segment of each candidate segment, and the extended segment is used as the key content corresponding to the candidate search result.

15. The method according to claim 11, wherein: The performing text generation processing on the at least one key content to obtain a second message for replying to the first message includes: Performing combination processing on the at least one key content to obtain a combination result; The combination result is processed by natural language text generation to obtain a second message for replying to the first message.

16. The method according to claim 15, wherein: The combining process of the at least one key content to obtain a combined result includes: Numbering each candidate search result corresponding to the key content to obtain a search result number for each candidate search result; Performing splicing processing on the at least one key content to obtain a splicing result, and adding a corresponding reference mark to each key content in the splicing result; Obtaining position mark information corresponding to the key content in the candidate search results of the source, and adding the position mark information to the reference annotation corresponding to the key content to obtain the combined result; The annotation serial number of the reference annotation is the same as the search result serial number of the candidate search result of the source of the corresponding key content.

17. An information processing device, the device comprising: A display module configured to display a session interface; A conversation module, configured to display a first message in a conversation interface in response to a message sending operation, and to display a second message in the conversation interface for replying to the first message; The second message includes key information used to reply to the first message and a reference mark of each key information, and the key information corresponds to the search result associated with the reference mark; A jump module is configured to display the information of the first position in the target search result in the reference interface in response to a trigger operation on the target reference annotation; wherein the information of the first position includes the target key information, and the target key information is the key information corresponding to the target reference annotation.

18. An electronic device, comprising: A memory for storing computer executable instructions; A processor, configured to implement the information processing method according to any one of claims 1 to 16 when executing the computer executable instructions stored in the memory.

19. A computer-readable storage medium storing computer-executable instructions, wherein the executable instructions, when executed by a processor, implement the information processing method according to any one of claims 1 to 16.

20. A computer program product, comprising computer executable instructions, wherein when the computer executable instructions are executed by a processor, the information processing method according to any one of claims 1 to 16 is implemented.

Citation Information

Patent Citations

  • Page redirection method and page redirection device

    CN107402953A

  • Information processing method and device, equipment, storage medium and program product

    CN116975437A

  • System and method for search result visualisation

    US20200218439A1