Multi-level container interaction method and system based on large language model

Abstract

The embodiment of the invention provides a multi-level container interaction method and system based on a large language model, and belongs to the technical field of computers. The method comprises the steps of displaying a main dialogue on a main window interface; when obtaining that a user selects interested contents in the main dialogue, popping up a new secondary dialogue window menu at a selected text; obtaining operation information of a user on the secondary dialogue window menu; displaying a secondary window beside the main dialogue according to the operation information; wherein the secondary window obtains a context abstract in the main dialogue through a context automatic aggregation and injection mechanism so as to complete continuous question and answer of the interested content by the user in the secondary window. According to the method, through the multi-stage session type user interface, an interaction mode between a person and a large language model is redesigned. The design surpasses the limitation of linear chat, and provides a dynamic, intuitive and efficient environment for complex information exploration and knowledge synthesis.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular, to a multi-level container interaction method and system based on large language model. BACKGROUND

[0002] The current mainstream large language model (LLM) application generally adopts a single chat container for dialogue. This linear, round-based interaction method, although simple, causes a cumbersome and inefficient interaction mode when the user needs to clarify information, explore collateral ideas, or delve into a specific point.

[0003] The user is forced into a frustrating cycle of either constantly flipping back and forth in a growing single dialogue, leading to cumbersome operations and interactions for the user, or choosing to start a separate chat, which loses the connection context between related ideas and requires the user to manually repeat background information, thus artificially fragmenting their thought process. At the same time, it also brings cumbersome operations and hardware load of the user's use end. In some applications with strict context restrictions, information loss after a certain number of interactions is inevitable, leading to the large model's answers deviating from the topic when dealing with some complex learning and research.

[0004] This fundamental mismatch between the linear chat interface and human thinking methods is not only inconvenient, but it forces the user into an unnatural and inefficient cognitive workflow, which directly contradicts the actual process of complex problem-solving and research.

[0005] In addition to structural inefficiency, the massive amount of information generated by LLM in linear format also significantly exacerbates the user's information overload problem. When the user iteratively asks questions, especially for secondary topics, this often generates large amounts of collateral questions on the topic, giving the impression that the knowledge is very cumbersome and unstructured. This information overload is a real problem that makes users feel frustrated and completely lost. It leads to divided attention, increased stress and anxiety, and seriously affects the thought process for problem exploration. This not only affects immediate task efficiency, but also affects the user's long-term trust and willingness to participate in complex tasks using LLM.

[0006] Therefore, how to solve the above problems is a problem that needs to be solved at present. SUMMARY

[0007] The present application provides a multi-level container interaction method and system based on large language model, aiming to improve the above problems.

[0008] In a first aspect, the application provides a multi-level container interaction method based on a large language model, the method comprising:

[0009] Displaying a main dialogue in a main window interface;

[0010] When the user selects the content of interest in the main dialogue, a new secondary dialogue window menu is popped up at the selected text;

[0011] Obtaining operation information of the user on the secondary dialogue window menu;

[0012] Displaying a secondary window beside the main dialogue according to the operation information; wherein the secondary window obtains a context summary in the main dialogue through a context automatic aggregation and injection mechanism to complete the user's continued questioning and answering in the secondary window.

[0013] In a possible embodiment, the displaying of the secondary window beside the main dialogue according to the operation information comprises:

[0014] Highlighting the questioning text by capturing the operation information of the user on the popped-up menu, extracting the semantic content of the selected text passage, the globally unique identifier of the message where the text passage is located, and the precise character position information of the text passage in the main dialogue;

[0015] Packaging the structured data packet of the semantic content of the text passage, the globally unique identifier of the message where the text passage is located, and the precise character position information of the text passage in the main dialogue, and asynchronously sending a creation request to a backend server through an instance state transmission API call; wherein the backend server completes the logical creation of a new window according to the structured data packet after receiving the creation request;

[0016] After receiving the response of the successful creation of the new window and the complete metadata of the new window returned by the backend server, a secondary window is dynamically instantiated beside the main dialogue using a rendering engine.

[0017] In a possible embodiment, the dynamically instantiating of the secondary window beside the main dialogue using the rendering engine after receiving the response of the successful creation of the new window and the complete metadata of the new window returned by the backend server comprises:

[0018] After receiving the response of the successful creation of the new window and the complete metadata of the new window returned by the backend server, a secondary window connected to the highlighted text in the selected text passage through a connection line is dynamically instantiated beside the main dialogue using a rendering engine.

[0019] In a possible embodiment, the secondary window has a foldable and / or draggable function.

[0020] In a possible embodiment, the method further comprises:

[0021] acquiring new dialogue information input by the user in the secondary window;

[0022] restricting the new dialogue information and response information responding to the new dialogue information within the scope of the secondary window to avoid interference with the process of the main dialogue or other parallel branches.

[0023] In a possible embodiment, before the step of restricting the new dialogue information and response information responding to the new dialogue information within the scope of the secondary window, the method further comprises:

[0024] aggregating all contents in the main dialogue and complete dialogue records of all parent branches of the secondary window to construct a structured prompt;

[0025] performing deep semantic understanding and logical induction on the new dialogue information input by using a large-scale language model in the structured prompt to generate a highly condensed and clear response information.

[0026] In a possible embodiment, the method further comprises:

[0027] when the user selects the content of interest in the new dialogue information, a new secondary dialogue window menu is popped up at the selected text;

[0028] acquiring operation information of the user on the new secondary dialogue window menu;

[0029] displaying a new secondary window beside the new dialogue information according to the operation information.

[0030] In a possible embodiment, the method further comprises:

[0031] acquiring second operation information of the user on a summary operation button of the main window interface;

[0032] displaying a summary window according to the second operation information, and the summary window is used to display summary contents of the main dialogue and the new dialogue information;

[0033] acquiring display mode operation information of the user on the text summary contents of the summary window;

[0034] displaying the summary contents in the form of a mind map in the summary window according to the display mode operation information.

[0035] In a second aspect, the present application also provides a multi-level container interaction system based on a large language model, the system comprising:

[0036] a main window module configured to display a main dialogue in a main window interface;

[0037] a hierarchical question support module configured to, when it is obtained that a user selects content of interest in the main dialogue, pop up a new secondary dialogue window menu at the selected text;

[0038] an operation module configured to obtain operation information of the user on the secondary dialogue window menu;

[0039] a secondary window module configured to display a secondary window beside the main dialogue according to the operation information; wherein the secondary window obtains a context summary in the main dialogue through a context automatic aggregation and injection mechanism to complete the user's continued question and answer on the content of interest in the secondary window.

[0040] In a third aspect, the present application also provides a computer readable storage medium, the computer readable storage medium storing a computer program, the computer program being executed by a processing device to perform the steps of the multi-level container interaction method based on a large language model according to any one of the first aspect.

[0041] The multi-level container interaction method and system based on a large language model provided by the present application display a main dialogue in a main window interface, pop up a new secondary dialogue window menu at the selected text when it is obtained that a user selects content of interest in the main dialogue, obtain operation information of the user on the secondary dialogue window menu, and display a secondary window beside the main dialogue according to the operation information. Thus, the need for cumbersome scrolling, manual repeated context or opening multiple irrelevant chat windows is eliminated. In addition, the multi-level architecture and dynamic context summary in the secondary window significantly improve the context maintenance ability of the user and the LLM, directly alleviating the phenomenon of getting lost in the dialogue observed in multi-round dialogue. This structured interaction helps to guide the LLM more effectively, thereby producing more coherent and accurate responses, and thus improving user trust and satisfaction. The overall improvement of interaction quality and the reduction of frustrating experience help to improve user engagement and retention. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0043] Figure 1 A structural schematic diagram of an electronic device provided for the first embodiment of the present application;

[0044] Figure 2 A flowchart of a multi-level container interaction method based on a large language model provided for the second embodiment of the present application;

[0045] Figure 3 A schematic diagram of a main window interface in a multi-level container interaction method based on a large language model; Figure 2

[0046] Figure 4 A schematic diagram of a new window interface popped up after listening to interesting content in a multi-level container interaction method based on a large language model; Figure 2

[0047] A schematic diagram of a secondary window created in a multi-level container interaction method based on a large language model; Figure 5 Figure 2 A schematic diagram of a secondary secondary window created in a multi-level container interaction method based on a large language model;

[0048] Figure 6 Figure 2 A schematic diagram of a further main window interface created in a multi-level container interaction method based on a large language model;

[0049] Figure 7 A schematic diagram of a summary window created in a multi-level container interaction method based on a large language model; Figure 2

[0050] A schematic diagram of a summary window created in a multi-level container interaction method based on a large language model; Figure 8 Figure 2 A schematic diagram of a summary content presented in a mind map manner in a summary window;

[0051] Figure 9 Figure 8 A functional module schematic diagram of a multi-level container interaction system based on a large language model provided for the third embodiment of the present application. DETAILED DESCRIPTION

[0052] Figure 10 DETAILED DESCRIPTION

[0053] ​​​​​In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions of the present application will be described clearly and completely below in conjunction with the drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0054] First embodiment:

[0055] Figure 1 A structural schematic diagram of an electronic device provided by the embodiments of the present application can be described in the present application by the schematic diagram shown in Figure 1 The electronic device 100 shown in the schematic diagram can be used to implement an example of the multi-level container interaction method and system based on a large language model according to the embodiments of the present application.

[0056] As shown in a structural schematic diagram of an electronic device, Figure 1 The electronic device 100 includes one or more processors 102, one or more storage devices 104, and an input device 106, which are interconnected by a bus system and / or other forms of connection mechanism (not shown). It should be noted that Figure 1 The components and structure of the electronic device 100 shown are only exemplary and are not limiting, and the electronic device can have Figure 1 The components shown, and can also have Figure 1 Other components and structures not shown.

[0057] The processor 102 can be a central processing unit (CPU) or other forms of processing units with data processing and / or instruction execution capabilities, and can control other components in the electronic device 100 to perform desired functions.

[0058] It should be understood that the processor 102 in the embodiments of the present application can be a central processing unit (CPU), and the processor can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0059] The storage device 104 can include one or more computer program products, which can include various forms of computer-readable storage media.

[0060] It should be understood that the storage device 104 in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically EPROM (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM) used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced SDRAM (ESDRAM), synchl ink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0061] Among them, one or more computer program instructions can be stored on the computer readable storage medium, and the processor 102 can run the program instructions to realize the client functions (implemented by the processor) in the embodiments of the present application described below and / or other desired functions. Various application programs and various data, such as various data used and / or generated by the application programs, etc. can also be stored in the computer readable storage medium.

[0062] The input device 106 can be a device used by a user to input instructions, and can include one or more of a keyboard, a mouse, a microphone, a touch screen, etc.

[0063] Second embodiment:

[0064] Referring to Figure 2 A flow chart of a multi-level container interaction method based on a large language model is shown, and the method specifically includes the following steps:

[0065] Step S201: Display the main dialog box in the main window interface.

[0066] As one implementation, before step S201, the method further includes: receiving request information sent by the user; and generating a main dialogue based on the request information.

[0067] For example, if a user enters "recommend a men's facial cleanser" in the main window, the following will appear: Figure 3 The main dialog shown.

[0068] It should be understood that the above examples are merely examples and not limitations, and the content in the examples is only for demonstration purposes.

[0069] Understandably, the main window interface functions similarly to existing mainstream dialog interfaces, used to handle key questions and responses.

[0070] Step S202: When it is obtained that the user has selected content of interest in the main dialog, a menu for creating a new secondary dialog window pops up at the selected text.

[0071] For example, such as Figure 3 As shown, assuming the user's interest is detected as "Pond's Men's Rice Extract Moisturizing Facial Cleanser," after highlighting "Pond's Men's Rice Extract Moisturizing Facial Cleanser" in the main dialog, the semantic content of the selected text segment, the globally unique identifier of the message containing the text segment, and the precise character position information of the text segment in the main dialog will be extracted; then, a pop-up window will appear as follows: Figure 4 The secondary dialog window menu shown.

[0072] Step S203: Obtain the user's operation information on the secondary dialog window menu.

[0073] Optionally, the operation information includes "copy", "create on the left side of L2", and "create on the right side of L2".

[0074] Step S204: Display a secondary window next to the main dialog according to the operation information.

[0075] The secondary window obtains a context summary from the main dialogue through an automatic context aggregation and injection mechanism to allow the user to continue asking and answering questions about the content of interest in the secondary window.

[0076] As an implementation, step S204 comprises: displaying a secondary window beside the main conversation according to the operation information, comprising: capturing the operation information of the user selecting the pop-up menu, highlighting the text, extracting the semantic content of the selected text passage, the globally unique identifier of the message where the text passage is located, and the precise character position information of the text passage in the main conversation; sending a structured data packet encapsulating the semantic content of the text passage, the globally unique identifier of the message where the text passage is located, and the precise character position information of the text passage in the main conversation to the backend server through a RESTful API call; wherein, after receiving the creation request, the backend server completes the logical creation of a new window according to the structured data packet; after receiving the response of the backend server returning the successful creation of the new window and the complete metadata of the new window, a secondary window is dynamically instantiated beside the main conversation using a rendering engine.

[0077] The metadata includes a globally unique identifier, a title, a level, a parent-child association relationship, a position coordinate, etc.

[0078] Optionally, an example of the structured data packet is as follows:

[0079]

[0080]

[0081] Optionally, the secondary window is connected to the highlighted text in the selected text passage through a connection line.

[0082] Optionally, the highlighted text capture: the user selects an important text passage in the conversation; capture the text and use it as the title of the new window; store the association relationship between the highlighted text and the window in the database.

[0083] Continuing with the example of Figure 4 , if the user selects "create on the left side of L2", i.e. the operation information is to create on the left side of L2, a secondary window as shown in Figure 5 will be created on the left side of the main conversation, and the secondary window is connected to "Pianli Mr. Mi Ruzel Cleansing Milk" through a connection line in the figure to visually prompt the user; when the user drags the secondary window, the connection line will also move dynamically. The user can continue to input conversation information in the window as shown in Figure 5 , such as "introduce the development history of this product" input in Figure 5 , the secondary window will obtain the context summary in the main conversation through the context automatic aggregation and injection mechanism to complete the question "introduce the development history of this product" and give the answer.

[0084] Optionally, the secondary window has the function of being foldable and / or draggable. For example, as shown in the triangular symbol on the right side of the secondary window in Figure 5 , the user can fold the secondary window through the triangular symbol. As for dragging, the user needs to select the secondary window and drag it by mouse. No specific limitation is made here.

[0085] In a possible embodiment, the multi-level container interaction method based on a large language model further includes: obtaining new dialogue information input by the user in the secondary window; limiting the new dialogue information and response information responding to the new dialogue information within the scope of the secondary window to avoid interference with the process of the main dialogue or other parallel branches.

[0086] Optionally, before the new dialogue information and the response information responding to the new dialogue information are limited within the scope of the secondary window, the method further includes: aggregating all contents in the main dialogue and complete dialogue records of all parent branches of the secondary window to construct a structured prompt; and using a large-scale language model to perform deep semantic understanding and logical induction on the input new dialogue information in the structured prompt to generate a highly condensed and clear response information.

[0087] As an implementation, when the user initiates a dialogue in different levels of windows, the relevant context information is automatically aggregated according to the hierarchical relationship of the windows:

[0088] 1. Main dialogue window context processing:

[0089] Retrieving historical message records of the current session;

[0090] Building a dialogue history in chronological order;

[0091] Sending the historical messages together with the current user input to the language model;

[0092] 2. Secondary window context processing:

[0093] Retrieving all historical messages of the associated main dialogue window;

[0094] Taking these messages as background context;

[0095] Adding a special system prompt to guide the model to pay attention to the topic of the secondary window, which originates from which highlighted text in the main window;

[0096] Sending the aggregated context together with the current user input to the language model;

[0097] 3. Tertiary and deeper level window context processing:

[0098] retrieve all the history messages of its direct parent window;

[0099] use the parent window messages as background context;

[0100] add a system prompt to guide the model to understand the relationship between the current window and the parent window;

[0101] send the aggregated context together with the user's current input to the language model.

[0102] For example, continuing with the example of Figure 5 , if the user inputs "introduce the development history of this product" in the Figure 5 , the context summary in the main dialogue will be automatically aggregated and injected through the context automatic aggregation and injection mechanism to complete the question "introduce the development history of this product" in this secondary window, and the answer will be given, and the answer will only be used in this secondary window, and will not be displayed in other windows, so as to constitute a completely isolated interactive environment.

[0103] That is, any query or discussion initiated by the user in the window input box is strictly limited within the scope of the window, and will not interfere with the main dialogue or other parallel branch processes. When the user submits a message in the secondary window, the front end will send the message content together with the unique ID of the window to a backend interface that specifically handles secondary interactions. After completing the window validity check, the backend will store the user's message and the corresponding artificial intelligence (AI) reply in the database and forcibly associate them with the window ID. The AI reply will eventually be presented in real time in the message list of the window, forming a logically complete, content-focused, and independent dialogue record.

[0104] In a possible embodiment, the multi-level container interaction method based on a large language model further includes: popping up an operation window interface when the user listens to the content of interest in the new dialogue information; obtaining operation information input by the user based on the operation window interface; and displaying a secondary secondary window next to the new dialogue information according to the operation information.

[0105] For example, if the user selects "launch on the market" in the secondary window and creates a secondary secondary window as shown in Figure 6 , the user can continue to input related questions in the secondary secondary window as shown in Figure 6 to continue the dialogue.

[0106] Of course, the user can also create a new secondary window for the content in the secondary secondary window, or return to the main dialogue to create a secondary window for another new content, which will not be described here.

[0107] It should be noted that the present application supports infinite-level recursive exploration capability. If the user has new doubts or wants to further explore some part of the content in the conversation process of a sub-window, they can repeat the initial operation process: that is, select the text and create a new, deeper level of "sub-sub" window. It clearly records and maintains the parent-child relationship between the new window and the upper window in the database through fields such as parent_mini_chat_id, and automatically increments its level identifier, thereby constructing an infinitely expandable tree-shaped dialogue model in data structure. This mechanism enables users to conduct multi-level, structured deep exploration around a core issue, while ensuring that the logical relationship of all dialogue branches is accurately and completely preserved.

[0108] In a possible embodiment, the large language model-based multi-level container interaction method further includes: obtaining second operation information of a user on a summary operation button of the main window interface; displaying a summary window according to the second operation information, the summary window being used to display summary content of the main dialogue and the new dialogue information; obtaining display mode operation information of a user on a text summary content of the summary window; and displaying the summary content in the form of a mind map in the summary window according to the display mode operation information.

[0109] Optionally, the second operation information is operation information generated by the user clicking a "summary" button in an operation interface as shown in FIG. 8. Figure 7

[0110] Optionally, the display mode operation information includes a text summary and a mind map. As shown in FIG. 9, if the user clicks the mind map, the content summarized in FIG. 8 will be presented to the user in the format as shown in FIG. 9. Figure 8 Figure 8 Figure 9

[0111] In this embodiment, in order to ensure seamless navigation and clear information tracing, intuitive visual cues are integrated into the design.

[0112] (1) Connection line indication: there is a suggestive connection line between the sub-window and the highlighted text from which it comes.

[0113] (2) Real-time update: when the sub-window is dragged, this visual connection line will be updated in real time, clearly indicating the source of the information and maintaining the spatial perception of information interaction.

[0114] ​​​​Based on the above design system, the corresponding structured information storage is carried out by using the information structure of the multi-level container architecture. After the user completes the dialogue, a summary and summary of the global information in the form of a mind map in Markdown format is provided. This function automates and simplifies the summary process of complex multi-level information generated during the interaction. It significantly improves the information processing efficiency and overall user experience after interacting with the LLM.

[0115] This goes beyond real-time interaction, providing structured and easily digestible output that can be easily reviewed, shared, and integrated into other knowledge management systems, directly addressing the burden of information overload and the challenge of handling large amounts of conversational data.

[0116] It should be noted that, Figures 2-9 The dialogue content (text) in

[0117] Third embodiment:

[0118] Referring to Figure 10 A multi-level container interaction system based on a large language model, as shown in the figure, the system includes a main window module 510, a hierarchical questioning support module 520, an operation module 530, and a secondary window module 540, wherein the functions of each module are as follows:

[0119] The main window module 510 is used to display the main dialogue in the main window interface;

[0120] The hierarchical questioning support module 520 is used to pop up a new secondary dialogue window menu at the selected text when the user selects the content of interest in the main dialogue;

[0121] The operation module 530 is used to obtain the operation information of the user on the secondary dialogue window menu;

[0122] The secondary window module 540 is used to display a secondary window next to the main dialogue according to the operation information; wherein the secondary window obtains the context summary in the main dialogue through a context automatic aggregation and injection mechanism to complete the user's continued questioning and answering in the secondary window for the content of interest.

[0123] It should be noted that the specific function description of each module is described above with reference to the description of the method embodiment, and will not be repeated here.

[0124] In summary, the multi-level container interaction method and system based on a large language model provided by the embodiments of the present application have the following beneficial effects:

[0125] 1. Improve user interaction efficiency:

[0126] By providing a structured environment for deep exploration, this UI significantly improves user productivity. The ability to branch conversations and explore secondary topics in independent, context-aware windows eliminates the need for cumbersome scrolling, manual context repetition, or opening multiple unrelated chat windows. This streamlined workflow enables users to clarify details or explore collateral content while maintaining focus on the primary topic, thereby significantly improving overall interaction efficiency. For professionals engaged in complex learning and research, this means a direct reduction in the time required for information retrieval and organization, thereby accelerating the discovery and analysis process.

[0127] 2. Optimized context management:

[0128] The multi-level architecture and dynamic context summaries within secondary windows significantly improve the user's and LLM's ability to maintain context. This directly alleviates the conversation drift phenomenon observed in multi-turn conversations. By visually connecting secondary conversations to their origins and allowing recursive exploration, this UI transforms a potentially overwhelming information space into an intuitive, navigable knowledge graph. This structured approach reduces cognitive load, countering the information overload issues that currently lead to divided attention, stress, and anxiety.

[0129] 3. Simplified information processing workflow:

[0130] The structured input provided by the multi-level UI can indirectly improve the reliability of LLM responses in complex, multi-turn scenarios. By allowing users to precisely define the scope and context of secondary questions in dedicated windows, this UI reduces ambiguity, which is a major cause of premature assumptions and drift by LLMs. This structured interaction helps guide the LLM more effectively, resulting in more coherent and accurate responses, which in turn improves user trust and satisfaction. The overall improvement in interaction quality and reduction in frustrating experiences contribute to increased user engagement and retention.

[0131] Further, the embodiment also provides a computer readable storage medium, and the computer readable storage medium stores a computer program. The computer program is executed by a processing device to perform the steps of any one of the multi-level container interaction method based on a large language model provided in the above embodiment two.

[0132] The computer program product of the multi-level container interaction method based on a large language model provided in the embodiment of the application includes a computer readable storage medium storing program codes. The program codes include instructions for executing the methods described in the above method embodiments. The specific implementation can be referred to the method embodiments, and will not be described here.

[0133] It should be noted that the above-described embodiments can be implemented in whole or in part by software, hardware (such as a circuit), firmware, or any combination thereof. When implemented using software, the above-described embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another, for example, the computer instructions can be transferred from one website, computer, server or data center to another website, computer, server or data center through wired (such as infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid state disk.

[0134] It should be understood that the term "and / or" herein merely describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents that the associated objects before and after it are in an "or" relationship, but it can also represent an "and / or" relationship, which can be understood according to the context before and after it.

[0135] In this application, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or similar expressions means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b, or c can represent a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0136] It should be understood that in various embodiments of the present application, the size of the sequence number of the above-described processes does not mean the order of execution, and the execution order of the processes should be determined according to their functions and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0137] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0138] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0139] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the above-described device embodiments are only schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0140] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e. can be located in one place or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0141] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically, or two or more units can be integrated into one unit.

[0142] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application. It should be noted that similar reference numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

Claims

1. A method for multi-stage container interaction based on a large language model, characterized in that, The method comprises: displaying a main dialogue in a main window interface; when obtaining that a user selects content of interest in the main dialogue, a new secondary dialogue window menu is popped up at the selected text; obtaining operation information of the user on the secondary dialogue window menu; displaying a secondary window beside the main dialogue according to the operation information; wherein the secondary window obtains a context summary in the main dialogue through a context automatic aggregation and injection mechanism to complete the user's continued question and answer on the content of interest in the secondary window.

2. The method of claim 1, wherein, The operation information of the user on the secondary dialogue window menu comprises: by capturing the operation information of the user on the popped-up menu after selection, the question text is highlighted, the semantic content of the selected text paragraph, the globally unique identifier of the text paragraph and the accurate character position information of the text paragraph in the main dialogue are extracted; the structured data packet encapsulating the semantic content of the text paragraph, the globally unique identifier of the text paragraph and the accurate character position information of the text paragraph in the main dialogue is sent to the backend server asynchronously through an abstract state transmission API call; wherein the backend server completes the logical creation of a new window according to the structured data packet after receiving the creation request; after receiving the response of the backend server returning the successful creation of the new window and the complete metadata of the new window, a secondary window is dynamically instantiated beside the main dialogue by using a rendering engine.

3. The method of claim 2, wherein, The operation information of the user on the secondary dialogue window menu comprises: after receiving the response of the backend server returning the successful creation of the new window and the complete metadata of the new window, a secondary window connected to the highlighted text in the selected text paragraph through a connection line is dynamically instantiated beside the main dialogue by using a rendering engine.

4. The method according to claim 2 or 3, characterized in that, The secondary window has the functions of being foldable and / or being draggable.

5. The method of claim 4, wherein, The method further comprises: obtaining new dialogue information input by the user in the secondary window; restricting the new dialogue information and response information responding to the new dialogue information within the scope of the secondary window to avoid interference with the main dialogue or other parallel branch processes.

6. The method of claim 5, wherein, Before the new dialogue information and the response information responding to the new dialogue information are restricted within the scope of the secondary window, the method further comprises: aggregating all contents in the main dialogue and complete dialogue records of all parent branches of the secondary window to construct a structured prompt; using a large-scale language model to perform deep semantic understanding and logical induction on the input new dialogue information in the structured prompt to generate a highly condensed and clear response information.

7. The method of claim 5, wherein, The method further comprises: when obtaining that a user selects content of interest in the new dialogue information, a new secondary dialogue window menu is popped up at the selected text; obtaining operation information of the user on the secondary dialogue window menu; According to the operation information, a secondary sub-window is displayed beside the new dialogue information.

8. The method of claim 7, wherein, The method further comprises: obtaining second operation information of a user on a summary operation button of the main window interface; displaying a summary window according to the second operation information, and the summary window is used to display summary content of the main dialogue and the new dialogue information; obtaining display mode operation information of a user on a text summary content of the summary window; displaying the summary content in the form of a mind map in the summary window according to the display mode operation information.

9. A multi-stage container interaction system based on a large language model, characterized by, The system comprises: a main window module, configured to display a main dialogue in a main window interface; a hierarchical questioning support module, configured to pop up a new secondary dialogue window menu when obtaining that a user selects content of interest in the main dialogue; an operation module, configured to obtain operation information of a user on the secondary dialogue window menu; a secondary window module, configured to display a secondary window beside the main dialogue according to the operation information; wherein the secondary window obtains a context summary in the main dialogue through a context automatic aggregation and injection mechanism to complete a user's continued questioning and answering on the content of interest in the secondary window.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer readable storage medium, and the computer program is run by a processing device to execute the steps of the multi-level container interaction method based on a large language model according to any one of claims 1-7.

Images