Method and device for determining output content

Abstract

The invention discloses a method and device for determining output content. The method comprises the steps that feature information of a target application window operated by a user is determined; based on the feature information, at least one piece of target operation scene information is determined from a scene memory library, the scene memory library comprises at least one piece of operation scene information, and the operation scene information is used for representing operation behaviors or operation habits of a user in a historical operation scene; reasoning a to-be-input target result in the target application window based on the feature information and the at least one piece of target operation scene information; and outputting the target result to the target application window.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and in particular to a method and device for determining output content. BACKGROUND

[0002] In the process of using an electronic device, the user often needs to switch between different application windows.

[0003] In many cases, the input operation required by the user in a certain application window may be related to some historical operation behaviors of the user. On this basis, when the user performs an operation in the current application window, the user may need to frequently review the information in the application window that has been operated on, or rely on memory to complete the required input operation, which not only has high input complexity, but also is difficult to guarantee the accuracy of information input. SUMMARY

[0004] In one aspect, the present application provides a method for determining output content, comprising:

[0005] determining feature information of a target application window of a user operation;

[0006] determining at least one target operation scenario information from a scenario memory bank based on the feature information, the scenario memory bank comprising at least one operation scenario information, the operation scenario information being used to represent operation behaviors or operation habits of the user in a historical operation scenario;

[0007] reasoning a target result to be input in the target application window based on the feature information and the at least one target operation scenario information;

[0008] outputting the target result to the target application window.

[0009] In one possible implementation, the determining at least one target operation scenario information from a scenario memory bank based on the feature information comprises:

[0010] obtaining a target operation event input in the target application window, the target operation event being used to trigger reasoning tasks in combination with historical operation scenarios;

[0011] in response to the target operation event, determining at least one target operation scenario information from a scenario memory bank based on the feature information.

[0012] In another possible implementation, the scenario memory bank comprises at least one of a short-term memory set and a long-term memory set.

[0013] The short-term memory set includes: short-term operation scenario information of the user at at least one point in time within the first historical time period, and the short-term operation scenario information includes: the historical window interface operated by the user and the interface content in the historical window interface.

[0014] The long-term memory set includes: at least one long-term operation scenario information, which is used to characterize at least one of the user's operation habits and the sequence of operation behaviors performed in a historical operation scenario during a second historical time period;

[0015] The second historical period is earlier than the first historical period.

[0016] In another possible implementation, at least one piece of long-term operational scenario information includes at least one of the following:

[0017] At least one first-type contextual information, which is used to record the interface content of the user's browsing history window interface;

[0018] At least one piece of second-type context information, which is used to record the interface information of at least one historical window interface of the user's continuous operation;

[0019] At least one piece of operation habit information, which is used to record at least one operation performed by the user in a historical operation scenario.

[0020] In another possible implementation, the feature information of the target application window for determining the user operation includes:

[0021] Determine the window function type corresponding to the target application window that the user switches to;

[0022] Alternatively, determine the window function type corresponding to the target application window where the user's input operation is located, as well as the input information entered in the input box of the target application window.

[0023] In another possible implementation, where the feature information of the target application window includes the window function type corresponding to the target application window and the input information entered in the input box of the target application window, the step of determining at least one target operation scenario information from the scenario memory based on the feature information includes:

[0024] If the window function type corresponding to the target application window is an interactive window of an intelligent model, and the input information in the target application window is a first prompt statement, determine the interface content of at least one historical window interface that the user has previously browsed and that is associated with the first prompt statement from the context memory library, wherein the first prompt statement is at least used to prompt the content features corresponding to the interface content displayed in the window interface that the user has previously browsed.

[0025] Alternatively, if the window function type corresponding to the target application window is an interactive window of an intelligent model, and the input information of the target application window is used to instruct a comparative analysis of the target article most recently read by the user and the reference article read before the target article, the target article and at least one of the reference articles are determined from the context memory bank, wherein the reference article and the target article have the same keywords;

[0026] Alternatively, if the window function type corresponding to the target application window is the operating system's search box, and the input information is an operation execution command, the operation habit information related to the operation execution command is determined from the context memory.

[0027] The target results include:

[0028] The first prompt statement after expansion;

[0029] Alternatively, a second prompt statement can be obtained by optimizing the input information, wherein the second prompt statement includes at least a portion of the content of the target article and at least a portion of the content of at least one of the reference articles;

[0030] Alternatively, an operation command sequence, the operation command sequence including at least one operation command required to complete the target operation indicated by the operation execution instruction.

[0031] In another possible implementation, where the feature information of the target application window includes the window function type corresponding to the target application window, determining at least one target operation context based on the feature information from the context memory includes:

[0032] If the window function type of the target application window indicates that the target application window is an information recording window for information recording, determine the interface content of at least one historical window interface browsed by the user within the first historical time period from the context memory.

[0033] Alternatively, if the window function type of the target application window is an interactive window of a first intelligent model, the historical interaction information of the user in at least one interactive window of a second intelligent model within a first historical time period is determined from the context memory; the target result includes:

[0034] The content to be recorded in the target application window;

[0035] Alternatively, a third prompt statement associated with the historical interaction information.

[0036] In another possible implementation, determining at least one target operation context based on the feature information from the context memory includes:

[0037] A target time range is determined, which is a time range determined based on the user's adjustment operation of the time adjustment control. In response to the adjustment operation, the electronic device can display an operation background label corresponding to the target time range. The operation background label is used to characterize the content summary of the operation scenario information in the scenario memory that belongs to the target time range.

[0038] Determine at least one target operation scenario from the scenario memory bank that corresponds to the target time range and matches the feature information.

[0039] In another possible implementation, the step of inferring the target result to be input in the target application window based on the feature information and the at least one target operation context information includes:

[0040] Based on the feature information and at least one target operation scenario information, at least one candidate operation intention and a summary of the operation scenario referenced by the candidate operation intention are determined and displayed. The operation scenario summary is used to indicate the target operation scenario information on which the candidate operation intention is determined.

[0041] Determine the target operation intent selected by the user from the at least one candidate operation intent;

[0042] Based on the target operation intention, feature information, and at least one target operation scenario information, the target result to be input in the target application window is inferred.

[0043] In another aspect, this application also provides an electronic device, including: a display unit, a memory, and a processor;

[0044] The display unit is used to display the target application window operated by the user;

[0045] The memory is used to store a context memory bank, which includes at least one operation context information, which is used to characterize the user's operation behavior or operation habits in a historical operation context.

[0046] The processor is configured to determine the feature information of the target application window, determine at least one target operation scenario from the scenario memory, and infer the target result to be input in the target application window based on the feature information and the at least one target operation scenario.

[0047] The display unit is also used to output the target result to the target application window. Attached Figure Description

[0048] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0049] Figure 1 A flowchart illustrating the method for determining output content provided in this application;

[0050] Figure 2 Another flowchart illustrating the method for determining output content provided in this application;

[0051] Figure 3 Another flowchart illustrating the method for determining output content provided in this application;

[0052] Figure 4 Another flowchart illustrating the method for determining output content provided in this application;

[0053] Figure 5 Another schematic diagram of the apparatus for determining output content provided in this application;

[0054] Figure 6 A schematic diagram of the composition structure of an electronic device provided in this application. Detailed Implementation

[0055] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is only for explaining specific embodiments and is not intended to limit the application. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0056] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0057] like Figure 1 The diagram illustrates a flowchart of a method for determining output content provided in this application. This method can be applied to electronic devices such as mobile phones, tablets, laptops, or wearable electronic devices. These electronic devices may have intelligent programs (also called intelligent agents) deployed within them, which then perform the operations described in this embodiment.

[0058] The method in this embodiment may include:

[0059] S101, Determine the characteristic information of the target application window of the user operation.

[0060] The target application window for user interaction can be the application window the user has currently switched to, such as the application window the user currently opens, or the application window the user navigates to from another application window. The target application window can also be the application window where the user is currently performing the input operation, such as the application window where the user is performing the input operation or the application window where the input cursor is located.

[0061] In this application, the characteristic information of the target application window refers to the characteristic information possessed by the target application window that reflects the input content that the user expects to input into the target application window. For example, the characteristic information of the target application window may include at least one of the window function characteristics of the target application window and the input information entered in the input box of the target application window.

[0062] Among them, window functional characteristics are used to characterize the functions that the target application window can perform. For example, for a presentation document window such as PPT, the window functional characteristics are content editing and presentation; for a note-taking application window such as Notebook, the window functional characteristics are information recording; and for the interactive interface of an application that can call intelligent models such as large language models, the window functional characteristics are information questioning and answering, or large language model interaction.

[0063] Of course, the window functionality of the target application window may vary depending on the application type to which the target application window belongs and the specific window interface of the target application window. These will not be elaborated here.

[0064] In practical applications, the characteristic information of the target application window will vary depending on the user's operation scenario (such as the type of the target application window and the type of operation the user performs on the target application window).

[0065] For example, in one possible scenario, the target application window is the target application window that the user has currently switched to. At this time, there is no input information entered by the user in the target application window. In this case, the window function type corresponding to the target application window that the user has switched to can be determined.

[0066] In another possible scenario, in order to more comprehensively infer the content that the user expects to input, it is also possible to determine the window function type of the target application window where the user's input operation is located, as well as the input information entered in the input box of the target application window.

[0067] For example, if the target application window is not a newly switched-to window at the current moment—for instance, if the target application window is a window where the user has already entered information or the input cursor has been hovering within the target application window for a period of time—then both the window's function type and the input information already entered within the target application window can be considered. Of course, in this case, the input information can also be empty, such as if the user entered some content and then deleted it, or if the user entered meaningless content, such as spaces.

[0068] S102, Based on feature information, determine at least one target operation scenario from the scenario memory bank.

[0069] The contextual memory bank includes at least one operational context. This operational context information is used to characterize a user's operational behavior or habits in a historical operational context.

[0070] In this application, an operation scenario refers to a specific operational situation in which a user uses an electronic device. Under different operation scenarios, the user's purpose in using the electronic device, the applications accessed, and one or more of the window interfaces within those applications may differ. For ease of distinction, the operation scenarios of the user's operation of the electronic device that have been collected and recorded by the electronic device are referred to as historical operation scenarios.

[0071] The contextual information in the contextual memory bank is generated based on the user's actions in different historical contexts. For example, a single instance of contextual information in the contextual memory bank might record specific content accessed by the user in a historical context, such as the user's historical window interfaces and their content. Contextual information can also be related to the user's actions within a specific context, such as the historical window interfaces accessed and the interactive behaviors within those interfaces. Finally, contextual information can also represent the user's operational habits within a specific context, which may include a series of habitual actions performed by the user in that context.

[0072] Understandably, with user permission, the intelligent agent of an electronic device can continuously acquire and record user operation behavior information, generate operation context information based on this behavior information, and store it in a context memory. On this basis, the operation context information recorded in the context memory can be used to obtain the user's previous operation context information and operating habits as target operation context information, in order to subsequently assist in determining the content the user actually needs to input within the target application window.

[0073] In this application, at least one target operation scenario information is operation scenario information in the scenario memory that is associated with the feature information of the target application window operated by the user (i.e., the operation scenario the user is currently in). For example, the target operation scenario information can be operation scenario information in the scenario memory corresponding to the operation scenario represented by the feature information of the target application window. Alternatively, the target operation scenario may also be operation scenario information that occurred before the user operated the target application window and is associated with the window type or input information of the target application window, such as user behavior information within an application window of the same type as the target application window before the user operated the target application window.

[0074] S103, based on feature information and at least one target operation scenario information, infer the target result to be input in the target application window.

[0075] For example, an intelligent program can use this feature information and at least one target operation context to invoke an inference model to infer the target result to be input in the target application window. For instance, the intelligent program may include an inference engine that can invoke an inference model to infer the target result. This inference model can be a large language model or other intelligent models, without specific limitations.

[0076] It is understandable that at least one piece of target operation context information is historical operation context information that matches the feature information of the target application window. This at least one piece of target operation context information reflects the user's actions prior to the current operation of the target application window, or the user's past actions or behaviors within the current operation context. Therefore, this at least one piece of target operation context information is equivalent to the operation background information or operation context information corresponding to the feature information of the target application window. Based on this, by combining this at least one piece of target operation context information, it is possible to more accurately infer the content the user wishes to input into the target application window, given the feature information of the target application window—that is, the inferred target result.

[0077] In one possible implementation, to more accurately deduce the target result, this application can further determine the target operation intention based on feature information and at least one target operation scenario information. Based on this, the target result is deduced by combining the target operation intention, the feature information, and the at least one target operation scenario information.

[0078] S104, output the target result to the target application window.

[0079] For example, the target result can be output to an input area such as an input field in the target application window. Alternatively, a floating window can be displayed within the target application window, and the target result can be output within the floating window, allowing the user to select the target result displayed in the floating window. Of course, there are other ways to implement outputting the target result to the target application window, and there are no restrictions on this.

[0080] As can be seen from the above, this application can record at least one operation scenario information in the scenario memory bank. Based on this, after determining the feature information of the target application window operated by the user, this application can determine at least one target operation scenario information from the scenario memory bank based on the feature information. Since the target operation scenario information is associated with the feature information of the target application window currently operated by the user, and can characterize the user's operation behavior or operation habits in historical operation scenarios, combining the at least one target operation scenario information and the feature information of the target application window can more accurately infer the target result that the user wants to input in the target application window and output it to the target application window. This reduces the situation where the user frequently switches between different application windows to view historical input information or relies on memory to input information and constantly corrects the input information, which naturally improves the editability of the input operation and the accuracy of inputting information into the target application window.

[0081] It is understood that, in this application, the application window currently being operated by the user is referred to as the target application window.

[0082] In this application, the task reasoning operation, which combines contextual operation information in the context memory, can be automatically triggered during the user's switching to or operation of the target application window. For example, when the user switches to the target application window, the characteristic information of the target application window is determined, and at least one target operation contextual information and subsequent operation are automatically determined from the context memory based on the characteristic information such as the window function type of the target application window. Similarly, during the user's input operation in the target application window, at least one target operation contextual information is determined from the context memory based on the characteristic information such as the window function type of the target application window and the currently entered input information within the target application window, and subsequent related reasoning is performed.

[0083] In one possible implementation, to reduce resource waste caused by blindly performing reasoning operations when the user does not wish to use the context memory bank for reasoning, this application allows the user to trigger the execution of the reasoning task through a specific operation. The following is combined with... Figure 2 Please provide an explanation.

[0084] like Figure 2 This illustration shows another flowchart of determining output content provided by this application. The method of this embodiment can be applied to the aforementioned electronic device, which runs an intelligent program. The method of this embodiment may include:

[0085] S201, Determine the characteristic information of the target application window of the user operation.

[0086] This step can be referred to in the relevant description of the previous embodiments, and will not be repeated here.

[0087] S202, obtain the target operation event input within the target application window.

[0088] Specifically, this target operation event is used to trigger a reasoning task based on historical operation scenarios. Therefore, if a user inputs this target operation event, it indicates that the user wishes to reason about the input content to be entered in the target application window by combining operation scenario information in the scenario memory bank.

[0089] The target operation event can be predefined without any restrictions.

[0090] For example, an intelligent program can display an operation button on the target application window via a floating window; or a right-click or specified operation can trigger the intelligent program or operating system to display an operation menu, which includes a target option. Both the operation button and the target option can be used to trigger a reasoning task based on historical operation scenarios. If the user clicks the button or selects the target option, the target operation event is confirmed. Alternatively, a shortcut key can be predefined for inputting the target operation event; when the user clicks or presses this shortcut key, the electronic device receives the target operation event.

[0091] Of course, there are other possibilities for the predefined target operation event, which will not be elaborated here.

[0092] It should be noted that in this embodiment, the example is that the feature information of the target application window is obtained before the operation event is received. In practical applications, in order to reduce the resource consumption caused by frequently or continuously obtaining the feature information of the target application window, when the electronic device is currently switched to or in the target application window, this application can also execute step S201 only after the target operation event is received. In this case, the order of the target steps S201 and S202 can be interchanged, such as determining the feature information of the target application window in response to the target operation event. Of course, steps S201 and S202 can also be executed simultaneously, which will not be elaborated further.

[0093] S203, in response to the target operation event, determine at least one target operation context information from the context memory based on feature information.

[0094] The context memory database includes at least one operation context information, which is used to characterize the user's operation behavior or operation habits in a historical operation context.

[0095] S204, based on feature information and at least one target operation scenario information, infer the target result to be input in the target application window.

[0096] S205, output the target result to the target application window.

[0097] The above steps can be referred to in the relevant descriptions of the previous embodiments, and will not be repeated here.

[0098] In this embodiment, only when a target operation event is obtained from the target application window will the target operation context information be determined from the context memory bank based on the feature information of the target application window, and the target result to be input in the target application window be inferred. This can more reasonably trigger the inference of the content to be input in the target application window by combining historical operation context information. This can avoid blindly executing task inference when the user does not want to combine the operation context information in the context memory bank for input operation, and naturally reduce the waste of resources caused by blindly combining the context memory bank to execute task inference.

[0099] In this application, the specific form of the operational context information recorded in the context memory bank can be varied, and can be set according to actual needs without restriction.

[0100] In one possible implementation, considering that human memory is divided into short-term memory and long-term memory, and that the characteristics of long-term memory and short-term memory are different, in order to make the operational context information recorded in the context memory bank more consistent with the characteristics of human memory, so as to better understand the user's input intent in the target application window by combining the context memory bank, in this application, the context memory bank may include at least one of the short-term memory set and the long-term memory set.

[0101] The short-term memory set includes: short-term operational context information of the user at at least one point in time within a first historical time period. This short-term operational context information includes: the user's historical window interfaces and the content within those interfaces.

[0102] The user's historical window interface can be any window interface that the user has visited. For example, windows that the user opened, performed input operations, or other actions on.

[0103] The interface content in the history window refers to the content displayed in that history window. For example, the interface content may include content displayed in windows such as web pages or PowerPoint documents, or content included in windows such as the clipboard. The interface content may also include information entered by the user into the history window, such as input information entered into a Word document or note-taking application window, as well as information entered into input boxes such as the search bar in the history window.

[0104] For example, the first historical time period can be the 24 hours preceding the current moment, or the last two days. The specific time period can be set according to actual needs without any restrictions.

[0105] The short-term memory set can be obtained by capturing and recording the window interface accessed by the user in real time, and by taking screenshots or extracting text from the interface content of the window interface, without any specific restrictions.

[0106] The long-term memory set includes at least one long-term operational scenario information. This long-term operational scenario information is used to characterize at least one of the user's operational habits and the sequence of operational behaviors performed in a historical operational scenario within a second historical time period.

[0107] The operation sequence includes at least one operation with a sequential order. The operation is represented by the interface information of the window interface operated by the user, as well as input operations or input content within the window interface. Therefore, the operation can include the interface information of the window interface being operated on and the input operations within the window interface. The interface information can be information used to characterize the window interface, such as the window interface identifier (e.g., window handle) and the interface content contained within the window interface.

[0108] In this context, a user's operational habits in a historical operational scenario can be derived by analyzing a sequence of actions performed by the user in one or more instances within that historical operational scenario. Based on this, operational habits primarily emphasize the window interface and input operations that the user habitually uses in that historical operational scenario, without needing to emphasize the content of the user-operated window interface.

[0109] The second historical time period is earlier than the first historical time period. Moreover, the duration of the second historical time period is usually longer than that of the first historical time period. For example, the first historical time period can be the 24 hours before the current time, while the second historical time period is the most recent month (which may or may not include the 24 hours before the current time).

[0110] Furthermore, considering the large amount of operational context information covered by long-term memory records, in order to reduce the complexity of subsequent matching of operational context information, this application can further classify long-term operational context information into different categories. For example, in one possible implementation, at least one piece of long-term operational context information in the long-term memory set includes at least one of the following:

[0111] At least one piece of first-class contextual information, which is used to record the interface content of the user's historical browsing window interface;

[0112] At least one second-type context information, which is used to record the interface information of at least one historical window interface of the user's continuous operation;

[0113] At least one operation habit information, which is used to record at least one operation performed by the user in a historical operation scenario.

[0114] The first type of contextual information focuses on the content displayed in various pages and other window interfaces that the user has visited. Therefore, the first type of contextual information can reflect the semantic information of the interface content of each window interface that the user has visited.

[0115] For example, the first type of contextual information can record text information and semantic information expressed by images in Word or PPT documents accessed by the user within the first historical time period, as well as interface content such as text content displayed in web pages or emails accessed.

[0116] Among them, the interface content in the first type of contextual information is used to characterize the semantic content of the displayed content of the historical window interface.

[0117] For example, the interface content of the window can be obtained by extracting a summary and concluding the content displayed on the window interface.

[0118] For example, screen recording technology can be used to acquire or capture images of the window interface, and content can be extracted from the recorded or captured images to obtain the interface content of the window. In this case, considering that the content of the window interface may change during user interaction, in order to reduce data processing and storage, this application can also use Differential UI Element Analysis (DUEA) to acquire and record only the enhanced interface information of the current window interface and the previous frame's window interface.

[0119] In addition, considering that some window interfaces may involve sensitive elements such as input boxes for entering passwords or bank card numbers, this application will also encrypt the text information in sensitive elements in the window interface to improve the security of the text information in sensitive elements in the window interface, so that the text information of sensitive elements in the window interface content in the context memory bank is information in non-plaintext form.

[0120] This second type of contextual information primarily reflects the recording of specific contextual information about the user's operation, such as what application tools were used and what operations were performed. Based on this, the second type of contextual information focuses on recording the user's interactive behavior through electronic devices. Therefore, it is necessary to record the interface information of at least one window interface that the user continuously operates on. Interface information may include only the interface identifier of the window interface, or it may include the interface content of the window interface.

[0121] For example, if a user visits a webpage, the content of that webpage will be recorded in the first type of context information, while the webpage identifier, clicked links, and search bar entries will be recorded in the second type of context information. When the user's window is an interactive window such as an instant messaging window, the instant messaging messages within the application window to which that window belongs can be recorded in the second type of context information to reconstruct the instant messaging interaction context.

[0122] The operations recorded in the operation habit information can be the actions performed, such as opening a report or generating a chart.

[0123] It is understandable that there are correlations between the above-mentioned categories of long-term contextual operation information in long-term memory records. For example, the second category of contextual information records the interface identifier of a window, and the interface content of that window can be retrieved from the first category of contextual information based on that interface identifier. Similarly, if the user opens a report, the report content can be retrieved from the first category of contextual information if the user's operation habit information records that the report was opened. Based on this, this application can utilize graph databases such as knowledge graphs to store the first category of contextual information, the second category of contextual information, and the operation habit information, so as to accurately and conveniently retrieve all long-term operation contextual information associated with the target application window.

[0124] In this application, the intelligent program of the electronic device may include a scene capture engine, which captures user operation scene information in different operation scenarios in real time. The intelligent program may also include a memory module, which can store the real-time captured operation scene information as a short-term memory set, and after the time point corresponding to the short-term operation scene information in the short-term memory set exceeds a first historical time period, convert the short-term operation scene information exceeding the first historical time period into the corresponding operation scene information in the long-term memory set.

[0125] It is understandable that in this application, when the characteristic information of the target application window operated by the user is different, that is, when the current operation scenario is different, the specific method of matching the target operation scenario information from the scenario memory and the matched target operation scenario information will also be different.

[0126] The following explains several possible scenarios based on the characteristic information of the target application window operated by the user:

[0127] In the first possible scenario, when the user switches to the target application window, if the window's function type indicates it's for information recording (such as an information editing page in a document application or a note-taking page in a note-taking application), the user will generally record information based on recently viewed web pages or documents. Therefore, at least one target operation scenario information matched from the context memory bank can include: the interface content of at least one historical window interface viewed by the user within a first historical time period. For example, in response to a target operation event, based on feature information, the interface content of at least one historical window interface viewed by the user is determined from the short-term memory set in the context memory bank. Based on this, and using the feature information of the target application window and the at least one target operation scenario information, the recording content to be recorded in the target application window can be inferred.

[0128] For example:

[0129] After a user browses multiple web pages introducing "solid electrolyte technology for lithium-ion batteries" on an electronic device, if the user switches to a note-taking application, the target window for the user's current operation will be the note-taking interface of that application. The multiple web pages related to "solid electrolyte technology for lithium-ion batteries" that the user visited, along with their content, will also be captured and stored as multiple short-term operation scenario information in the short-term memory set.

[0130] Building upon this foundation, users can trigger a pop-up menu by clicking the left mouse button in a blank area of ​​the note-taking application. Selecting the "Context Enhancement" option from the pop-up menu confirms that the electronic device has detected the target operation event. The device then queries its short-term memory to retrieve the identifiers of web pages visited within the last 15 minutes and obtains the content of those pages, thus acquiring multiple web page contents related to "lithium-ion battery solid electrolyte technology." The device's intelligent program then deduplicates, organizes, and refines this content, generating a structured note containing core viewpoints, key data, and data sources. This note is then output to the note-taking area of ​​the application. In this way, users do not need to frequently review previously visited web pages during note-taking. The proposed solution efficiently and accurately generates notes based on the content of the web pages visited, improving note-taking efficiency and accuracy.

[0131] In the second possible scenario, if the characteristic information of the target application window operated by the user can include the window function type and input information corresponding to the target application window, the window function type of the target application window can be an interactive window of the intelligent model, and the input information within the target application window can be a first prompt statement.

[0132] The interactive window of the intelligent model serves as the user interface for interacting with the model. For example, this window may include a question input area where the user enters a first prompt, and a question answer display area showing the first prompt and the intelligent model's reasoning results in response.

[0133] The first prompt statement serves at least to indicate the content characteristics corresponding to the interface content displayed in the user's previously viewed windows. These content characteristics are used to provide background or contextual information needed to process the user's question. Of course, the first prompt statement may also include prompt words indicating the user's question requirements. For example, if the first prompt statement is "The future development prospects of artificial intelligence technology," then the keywords "artificial intelligence" and "development prospects" can be extracted. Therefore, "artificial intelligence technology" is the content characteristic that the user's previously viewed windows related to this first prompt statement should possess.

[0134] In this case, at least one target operation context information matched from the context memory based on feature information may include: the interface content of at least one historical window interface that the user has browsed in the past and that is associated with the first prompt statement.

[0135] For example, the first prompt statement in the interactive window of the intelligent model and the information that the target application window is the interactive window of the intelligent model are input into the context matching model. The context matching model is then used to match the interface content of at least one historical window interface from the context memory bank. The interface content of at least one historical window interface can be extracted from at least one of the short-term memory set and the long-term memory set.

[0136] For example, it can first detect whether there is a historical window interface in the short-term memory set that matches the first prompt statement in the interactive window of the intelligent model. If so, the interface content of at least one window interface is obtained from the short-term memory set. If there is no interface content of at least one window interface that matches the first prompt statement in the interactive window of the intelligent model in the short-term memory set, the interface content of at least one window interface is matched from the long-term memory set.

[0137] Accordingly, based on the first prompt statement input into the intelligent model and the interface content of at least one historical window interface matched from the context memory bank, an expanded first prompt statement can be inferred. The expanded first prompt statement includes a description of the interface content of the historical window interface associated with it, so that the first prompt statement not only contains the question but also the background information of the question, making the first prompt statement more accurately describe the user's problem.

[0138] For example:

[0139] Suppose a user opens and reads a PowerPoint presentation on the "Commercialization Challenges of Sodium-ion Batteries" on an electronic device and simultaneously browses two related news web pages. The electronic device will then record the content of the PowerPoint presentation and the two news web pages in its short-term memory set. Based on this, if the user switches to a chat interface using a large language model and enters a question such as, "Can battery life be improved quickly in the near future?", and simultaneously clicks a shortcut key representing "contextual enhancement," or selects "contextual enhancement" from a pop-up menu, the electronic device's intelligent program will confirm that the target action event has been detected.

[0140] The intelligent program responds to the target operation event and calls the inference engine to input the question, "Can the battery life be improved quickly in the near future?", based on the large language model. It can analyze whether there are any documents or web pages related to "battery" and "foreground" that the user has viewed in the context memory bank. Thus, it can extract the content of the PPT and news web pages about "commercial challenges of sodium-ion batteries" that the user has read from the context memory bank.

[0141] Building upon this, the intelligent model adds background information from PPT presentations and news articles on "Commercialization Challenges of Sodium-ion Batteries" to the question, enriching and completing it. For example, the question could be expanded and rewritten as: "I am researching battery technology, particularly focusing on the commercial prospects of next-generation batteries, such as sodium-ion batteries. Considering the challenges I've recently seen regarding their energy density, cycle life, and cost (as mentioned in XX company's PPT and XX news), I would like to understand from a more professional perspective: Compared to the current mainstream lithium-ion batteries, is it possible for technologies like sodium-ion batteries to achieve a breakthrough and become commercially competitive in the key indicator of 'duration of use' (i.e., energy density) within the next 2-3 years? What are its main bottlenecks?"

[0142] By comparing the initial question with the expanded question, it can be seen that outputting the expanded question to the chat interface of the large language model and then inputting the expanded question into the large language model enables the large language model to understand the user's intent more accurately and provide more comprehensive and accurate answers to the questions raised by the user.

[0143] In the third possible scenario, the specific feature information for determining the target application window of the user's operation can be: the window function type corresponding to the target application window that the user switches to, and the window function type corresponding to the target application window is an interactive window of the first intelligent model. In this case, based on the fact that the window function type of the target application window is an interactive window of the first intelligent model, at least one target operation scenario information matched from the context memory bank can include: the user's historical interaction information within at least one interactive window of the second intelligent model within a first historical time period. That is, the historical interaction information (i.e., interface content) within at least one interactive window of the second intelligent model that the user has historically browsed can be matched from the short-term memory set.

[0144] Based on this, the third prompt statement that the user is expected to input into the interaction window of the first intelligent model can be inferred from the historical interaction information within the interaction window of at least one second intelligent model. This third prompt statement is a prompt statement associated with the historical interaction information within the interaction window of the at least one second intelligent model.

[0145] Understandably, when a user switches from the interaction window of another intelligent model to the interaction window of the current intelligent model, the user is likely to want to compare the answers of other intelligent models to a certain question with the answers of the current intelligent model to the same question. Therefore, the interaction information of other intelligent models that the user has visited in the past can be used to determine the question that the user wants to ask in the current intelligent model.

[0146] For example:

[0147] Suppose a user engages in multiple rounds of dialogue within a large language model A's chat window regarding the question, "How to use AI for code review?" If the user wants to see what answer large language model B can provide, they switch from chat model A to chat model B. If the user then selects the "context enhancement" option (one of the methods mentioned earlier) to trigger the input target action event, the electronic device's intelligent program can determine, based on its context memory, that the user recently visited chat model A's chat window. This allows the program to determine the user's intention to further analyze the problem using large language model B. Based on this, the intelligent program can generate a prompt statement based on the dialogue content within chat model A's chat window. For example, the prompt statement could be, "I've discussed using AI for code review with another AI. We talked about static analysis and patterns based on historical commits. Now I'd like to understand from your perspective what other cutting-edge or different methods exist? In particular, how can AI be effectively used to discover deep logical flaws, not just style issues? Please provide specific technical paths or tool suggestions."

[0148] In the fourth possible scenario, the specific feature information of the target application window operated by the user can be: the window function type corresponding to the target application window operated by the user and the input information within the target application window, wherein the window function type corresponding to the target application window is an interactive window of the intelligent model, and the input information within the target application window is used to instruct a comparative analysis of the target article most recently read by the user and the reference articles read before the target article.

[0149] Among them, the reference article and the target article share the same keywords. For example, both the reference article and the target article are related to the keyword "battery".

[0150] In this application, the referenced articles and target articles can be articles that users read when accessing web pages, documents, or other application windows through electronic devices, without any specific restrictions.

[0151] Based on this, the electronic device, using the input information entered within the interactive window of the intelligent model, can obtain at least one target operation scenario from the scenario memory bank, which may include: the target article and at least one reference article. Both the target article and the reference article belong to the window content displayed within the user's visited history window interface.

[0152] Of course, the at least one target operation scenario information may also include the identifier of the historical window interface for displaying the target article and the interface identifier of the historical window interface for displaying the reference articles, etc., without any specific restrictions.

[0153] Based on this, considering that the target application window is the interaction window of the intelligent model, the input information within the interaction window of the target intelligent model, the target article, and the reference articles, the target result to be input into the intelligent model can be deduced as: the second prompt statement obtained by optimizing the input information within the interaction window of the intelligent model.

[0154] The second prompt statement includes at least a portion of the target article and at least a portion of the content from at least one of the reference articles. It also includes prompts indicating whether to compare and contrast the content of the target article with at least a portion of the content from each reference article.

[0155] For example:

[0156] Suppose a user has just read an article titled "Silicon-Carbon Batteries" The user had previously read several articles about battery technology. Based on this, the user entered the question, "What's new about the battery article I just read compared to others I've read before?" into the chat window of the large language model (or other intelligent models). Furthermore, the user selected the "contextual enhancement" option within the large language model's chat window, i.e., entered the target action event. The electronic device can then determine, based on this question, the user's most recently read battery-related article, titled "Silicon-Carbon Batteries," from the short-term memory set of the contextual memory bank. The article "Applications" was consulted, and reference articles related to the "battery" were obtained from the short-term memory set and the long-term memory set.

[0157] Based on this, the intelligent program in the electronic device can extract the core viewpoints of the target article and reference articles, and use these viewpoints as context to construct a prompt statement along with the question. For example, the newly constructed prompt statement could be: "The viewpoints about batteries in the article A we just saw are:" Furthermore, I previously saw the following viewpoints regarding batteries in articles B and C: "So, what new insights does article A offer compared to articles B and C?" Correspondingly, the newly generated prompt can be output to the input bar of the large language model's chat window or displayed via a floating window. If the user chooses to input the prompt, it can be sent to the large language model.

[0158] In this example, while the intelligent program searches its contextual memory for target and reference articles related to "battery," it can also leverage this memory to obtain information about the user's operational habits when using the intelligent model to analyze the differences between various articles. For instance, the user's operational habits might indicate that the user needs to combine the core themes of the searched target and reference articles and optimize the question accordingly. Therefore, the intelligent program will sequentially extract the core themes from the target and reference articles based on the user's operational habits and use these extracted themes as context to optimize the question.

[0159] In the fifth possible scenario, the specific characteristic information for determining the target application window operated by the user can be: the window function type corresponding to the target application window and the input information within the target application window, wherein the window function type corresponding to the target application window is the operating system's search box, and the input information within the target application window is an operation execution command. In this case, the electronic device's intelligent program can determine the operation habit information related to the operation execution command from the context memory based on the operation execution command input into the operating system's search box.

[0160] Accordingly, based on the operating habit information and the operation execution instruction entered in the operating system's search box, the sequence of operation commands to be entered in the search box can be deduced. This sequence of operation commands includes at least one operation command required to complete the target operation indicated by the operation execution instruction.

[0161] For example:

[0162] Suppose a user enters the command "Download a panda icon for me" in the operating system's search box. After detecting that the user triggers a pop-up menu by right-clicking and selects the "Enhanced Context" option, the intelligent program confirms the target operation event. Based on this command, it searches the context memory for the user's past experience downloading panda icons. For example, the retrieved information might indicate that the user needs to first open a browser, visit the target icon website, and then search for and download the red panda icon from that website. Based on this, the electronic device's intelligent program can generate a sequence of commands, including: a first command to open a browser; a second command to access the target icon website; a third command to search for the red panda icon on the website; and a fourth command to download the found red panda icon.

[0163] Correspondingly, after the user enters the command sequence into the operating system's search box, if the user triggers the execution of the command sequence, the operating system can execute the command sequence and ultimately download the red panda icon that the user needs.

[0164] Of course, in practical applications, there may be other possibilities for the window function characteristics of the target application window operated by the user and the input information within the target application window. Correspondingly, there may also be other possibilities for matching the target operation scenario from the context memory and inferring the target result, which will not be elaborated here.

[0165] In addition, when the electronic device is a wearable electronic device such as smart glasses or smart headphones, the smart program in the wearable electronic device can also interact with the smart program of electronic devices such as mobile phones. In this case, the wearable device can perform target result reasoning based on at least one of its own maintained context memory bank and the context memory bank maintained in the electronic device that has established a communication connection with the wearable electronic device, and in combination with the feature information of the target application window currently operated by the user.

[0166] For example, after a wearable electronic device receives a context projection command from a second electronic device (i.e., another electronic device besides the wearable device), the wearable electronic device can establish a communication connection with the second electronic device in response to the context projection command, and obtain the display content captured by the second electronic device's current display interface. Based on this, the wearable electronic device can display the display content, allowing the user to continue viewing the content of the second electronic device's display interface even after leaving the second electronic device. Furthermore, after the second electronic device projects its display content onto the wearable electronic device, the second electronic device can control its display interface to lock, thereby reducing information leakage.

[0167] For example, after the second electronic device opens the calendar application and displays the calendar, if the second electronic device sends a scenario projection command to the wearable electronic device, and the wearable electronic device returns response information to the scenario projection command, the second electronic device can send the calendar as scenario operation information to the wearable electronic device. Based on this, the wearable electronic device uses the calendar as recorded scenario operation information, and combined with the current operation scenario of the wearable electronic device, infers the target result that needs to be output, and outputs the target result in the form of voice or text.

[0168] For example, after a wearable electronic device detects that a user has entered the question "What should I do first tonight?" on the main interface, it can determine the user's schedule based on their calendar, which includes picking up clothes and an online meeting at 8:30 PM. Furthermore, by combining auxiliary reference information such as the dry cleaner's opening hours obtained online, the wearable electronic device can deduce a response to be entered on the main interface, such as: "Since the dry cleaner closes at 8 PM, we suggest you pick up your clothes first. Would you like me to plan your walking route?" This response can then be displayed or played on the main interface.

[0169] It is understood that in any of the above embodiments of this application, since the context memory may record operational context information over a relatively long period of time, but the user is actually only concerned with operational context information within a certain period of time, in order to more accurately match the target operational context information from the context memory and reduce the amount of data search, this application may also determine the target time range before matching the target operational context information from the context memory. Accordingly, at least one target operational context information corresponding to the target time range and matching the feature information can be determined from the context memory.

[0170] The target time range is determined based on the user's adjustment of the time adjustment control. The time adjustment control is an operation control displayed on top of the target application window, used to select the time range. Based on this, the user can set the reference time range required to search for operation scenario information from the scenario memory library by adjusting the time adjustment control.

[0171] Furthermore, to enable users to more reasonably and accurately determine the required time range based on the context information recorded in the context memory bank, in this application, the electronic device, in response to an adjustment operation on the time adjustment control, can display an operation background label corresponding to the target time range adjusted by the adjustment operation. This operation background label is used to represent a summary of the content of the operation context information in the context memory bank that falls within the target time range.

[0172] To facilitate understanding of the specific implementation of determining the target time range, the following explanation uses one implementation method as an example. Figure 3 This illustration shows another implementation flow diagram of the method for determining output content provided in this application. The method in this embodiment may include:

[0173] S301, Determine the characteristic information of the target application window of the user operation.

[0174] S302, obtain the target operation event input within the target application window.

[0175] Among them, the target operation event is used to trigger the execution of reasoning tasks in combination with historical operation scenarios.

[0176] The above two steps can be referred to in the relevant descriptions of the previous embodiments, and will not be repeated here.

[0177] It is understood that this application uses the example of reasoning based on the operation context information in the context memory of an electronic device triggered by a target operation event. However, the implementation of reasoning based on the operation context information in the context memory by default is also applicable to this embodiment when there is no target operation event, and will not be described in detail here.

[0178] S303, responding to the target operation event, displays the time adjustment control.

[0179] It should be noted that this embodiment uses the example of an electronic device displaying a time adjustment control via a smart program in response to a target operation event. However, it is understood that this embodiment also applies even if the target operation event does not exist.

[0180] For example, this application can display the time adjustment control in the target application window through a smart program when the electronic device displays the target application window; or, the time adjustment control can be displayed when a user inputs a specified operation, such as when the user clicks the right mouse button and selects the "time adjustment" option in the displayed menu, or when the user triggers the display of the time adjustment control by operating a specific shortcut key.

[0181] S304, in response to the user's adjustment operation on the time adjustment control, determine the target time range that matches the adjustment operation, and display the operation background label corresponding to the target time range.

[0182] The operation background label is used to represent the content summary of the operation scenario information in the scenario memory that belongs to the target time range. For example, if the operation scenario information in the target time range includes articles displayed on the page, then the content summary can be the article title and article summary; if the operation scenario information in the target time range includes interactive content in a chat window, then the content summary can be a brief overview of the interactive content.

[0183] Understandably, as users continuously adjust or change the time adjustment controls, the target time range matched by the adjustment will also change dynamically. Correspondingly, the displayed operation background label will also change continuously with the change of the target time range, so that users can reasonably adjust the target time range based on the displayed operation background label.

[0184] S305, in response to the cessation of the adjustment operation, based on the target time range set by the adjustment operation, at least one target operation scenario information corresponding to the target time range and matching the feature information is determined from the scenario memory bank.

[0185] It is understandable that when the adjustment operation stops, the target time range set through the adjustment operation is also the target time range that the user ultimately wants to select. Based on this, target operation scenario information belonging to the target time range and matching the feature information can be matched from the scenario memory bank.

[0186] For example, if the first historical time period is within 24 hours, and the target time range can be 10 minutes, then at least one target operation scenario information that matches the feature can be determined from the operation scenario information generated in the most recent 10 minutes.

[0187] S306, Output the target result to the target application window.

[0188] This step can be referred to in the relevant description of the previous embodiments, and will not be repeated here.

[0189] Understandably, in order to accurately infer the target result to be input within the target application window, this application can also pre-analyze the user's target operation intent based on at least one determined target operation scenario information. However, in practical applications, multiple operation intents may be analyzed based on at least one target operation scenario information. Therefore, in order to more accurately determine the user's operation intent, this application can also allow the user to select the most suitable operation intent from the determined operation intents. The following will combine... Figure 4 Please provide an explanation.

[0190] like Figure 4 This illustrates another flowchart of the method for determining output content provided in this application. The method of this embodiment can be applied to the aforementioned electronic devices. The method of this embodiment may include:

[0191] S401, determine the characteristic information of the target application window of the user operation.

[0192] S402, obtain the target operation event input within the target application window.

[0193] Among them, the target operation event is used to trigger the execution of reasoning tasks in combination with historical operation scenarios.

[0194] It should be noted that step S402 is an optional step. In practical applications, if step S402 is not executed and the operation of determining at least one target operation scenario information from the scenario memory bank based on feature information is directly executed in step S403, it is also applicable to this embodiment.

[0195] S403, in response to the target operation event, determine at least one target operation context information from the context memory based on the feature information.

[0196] The context memory database includes at least one operation context information, which is used to characterize the user's operation behavior or operation habits in a historical operation context.

[0197] S404, Based on feature information and at least one target operation scenario information, determine and display at least one candidate operation intention and a summary of the operation scenario referenced by the candidate operation intention.

[0198] For example, based on this feature information and at least one target operation context information, an intent determination model such as a large language model can be invoked to determine at least one candidate operation intent.

[0199] The summary of the operational scenario referenced by the candidate operational intent is used to indicate the target operational scenario information on which the candidate operational intent is determined. For example, the operational scenario summary may be at least part of the target operational scenario information on which the candidate operational intent is determined, or it may be an information summary of the target operational scenario information.

[0200] Understandably, by using the summary of the operational context referenced by the candidate operational intent, users can determine whether the operational context information on which the intelligent program infers the reference intent is accurate. This allows for a more reasonable selection of candidate operational intents that are closest to the user's true intent and whose operational context information is reasonable.

[0201] S405, determine the target operation intent selected by the user from at least one candidate operation intent.

[0202] For ease of distinction, the candidate operation intent selected by the user is referred to as the target operation intent.

[0203] S406, based on the target operation intention, feature information and at least one target operation scenario information, infer the target result to be input in the target application window.

[0204] For example, if the feature information includes the input information within the target application window and the window function type of the target application window, then the input information can be optimized based on the window function type of the target application window and at least one target operation context information, combined with the target operation intention, to obtain the target result.

[0205] For example, if the feature information includes the window function type of the target application window, then the target result to be input in the target application window can be inferred based on the window function type of the target application window and at least one target operation context information, combined with the target operation intention.

[0206] S407, output the target result to the target application window.

[0207] Corresponding to the method for determining output content provided in this application, this application also provides an apparatus for determining output content. For example... Figure 5 This illustration shows a schematic diagram of the structure of an apparatus for determining output content provided in this application. The apparatus of this embodiment can be applied to electronic devices, and the method of this embodiment may include:

[0208] The feature determination unit 501 is used to determine the feature information of the target application window operated by the user.

[0209] Information determination unit 502 is used to determine at least one target operation scenario information from the scenario memory bank based on the feature information. The scenario memory bank includes at least one operation scenario information, which is used to characterize the user's operation behavior or operation habits in a historical operation scenario.

[0210] The result reasoning unit 503 is used to reason out the target result to be input in the target application window based on the feature information and the at least one target operation scenario information.

[0211] The result output unit 504 is used to output the target result to the target application window.

[0212] In one possible scenario, the information determining unit includes:

[0213] The event acquisition subunit is used to acquire the target operation event input within the target application window. The target operation event is used to trigger the execution of a reasoning task in conjunction with historical operation scenarios.

[0214] The information determination subunit is used to determine at least one target operation scenario information from the scenario memory bank based on the feature information in response to the target operation event.

[0215] In another possible scenario, the information determining unit determines that the contextual memory bank queried for the target operation context information includes at least one of the short-term memory set and the long-term memory set;

[0216] The short-term memory set includes: short-term operation scenario information of the user at at least one point in time within the first historical time period, and the short-term operation scenario information includes: the historical window interface operated by the user and the interface content in the historical window interface.

[0217] The long-term memory set includes: at least one long-term operation scenario information, which is used to characterize at least one of the user's operation habits and the sequence of operation behaviors performed in a historical operation scenario during a second historical time period;

[0218] The second historical period is earlier than the first historical period.

[0219] In another possible implementation, at least one piece of long-term operational scenario information includes at least one of the following:

[0220] At least one first-type contextual information, which is used to record the interface content of the user's browsing history window interface;

[0221] At least one piece of second-type context information, which is used to record the interface information of at least one historical window interface of the user's continuous operation;

[0222] At least one piece of operation habit information, which is used to record at least one operation performed by the user in a historical operation scenario.

[0223] In yet another possible implementation, the feature determination unit includes:

[0224] The first feature determination subunit is used to determine the window function type corresponding to the target application window that the user switches to.

[0225] Alternatively, the second feature determination subunit is used to determine the window function type corresponding to the target application window where the user's input operation is located, as well as the input information entered in the input box of the target application window.

[0226] In one possible implementation, if the window function type of the target application window determined by the first feature determining subunit indicates that the target application window is an information recording window for information recording, the at least one target operation scenario information determined by the information determining unit includes: the interface content of at least one historical window interface browsed by the user within the first historical time period.

[0227] Accordingly, the target result inferred by the result inference unit includes: the recording content to be recorded in the target application window.

[0228] In another possible implementation, the window function type corresponding to the target application window determined by the second feature determining subunit is an interactive window of the intelligent model, and the input information in the target application window is a first prompt statement, which is at least used to prompt the content features corresponding to the interface content displayed in the window interface that the user has browsed in the past.

[0229] The information determining unit determines at least one target operation scenario information, including: the interface content of at least one historical window interface that the user has browsed in the past and that is associated with the first prompt statement.

[0230] Accordingly, the target result determined by the result reasoning unit includes: the expanded first prompt statement.

[0231] In another possible implementation, the window function type corresponding to the target application window determined by the first feature determining subunit is the interactive window of the first intelligent model;

[0232] The information determination unit determines at least one target operation scenario information, including: the user's historical interaction information within the interaction window of at least one second intelligent model during the first historical time period;

[0233] The target result determined by the result reasoning unit includes: a third prompt statement associated with the historical interaction information.

[0234] In another possible implementation, the window function type corresponding to the target application window determined by the second feature subunit is the interactive window of the intelligent model, and the input information of the target application window is used to instruct the user to compare and analyze the target article most recently read by the user with the reference article read before the target article, wherein the reference article and the target article have the same keywords.

[0235] The information determination unit identifies at least one target operation scenario, including: the target article and at least one reference article;

[0236] The target result determined by the result reasoning unit includes: a second prompt statement obtained by optimizing the input information, wherein the second prompt statement includes at least part of the content of the target article and at least part of the content of at least one of the reference articles.

[0237] In another possible implementation, the window function type corresponding to the target application window determined by the first feature subunit is the operating system's search box, and the input information is an operation execution command;

[0238] The information determining unit identifies at least one target operation scenario information, including: operation habit information related to the operation execution instruction;

[0239] The target result determined by the result reasoning unit includes: an operation command sequence, which includes at least one operation command required to complete the target operation indicated by the operation execution instruction.

[0240] In another possible implementation, the information determining unit includes:

[0241] A time determination subunit is used to determine a target time range, which is a time range determined based on the user's adjustment operation of the time adjustment control. In response to the adjustment operation, the electronic device can display an operation background label corresponding to the target time range. The operation background label is used to characterize the content summary of the operation scenario information in the scenario memory that belongs to the target time range.

[0242] The information matching subunit is used to determine at least one target operation scenario information that corresponds to the target time range and matches the feature information from the scenario memory bank.

[0243] In yet another possible implementation, the result reasoning unit includes:

[0244] An intent recognition subunit is configured to determine and display at least one candidate operation intent and an operation scenario summary referenced by the candidate operation intent based on the feature information and at least one target operation scenario information. The operation scenario summary is used to indicate the target operation scenario information on which the candidate operation intent is determined.

[0245] An intent determination subunit is used to determine the target operation intent selected by the user from the at least one candidate operation intents;

[0246] The result reasoning subunit is used to infer the target result to be input in the target application window based on the target operation intention, feature information, and at least one target operation scenario information.

[0247] This application also provides an electronic device in its embodiments. For example... Figure 6 As shown, it illustrates a schematic diagram of the composition structure of the electronic device, which includes at least a processor 601, a memory 602, and a display unit 603;

[0248] The display unit 603 is used to display the target application window of the user operation;

[0249] The memory 602 is used to store a scenario memory bank, which includes at least one operation scenario information, which is used to characterize the user's operation behavior or operation habits in a historical operation scenario.

[0250] The processor 601 is configured to determine the feature information of the target application window, determine at least one target operation scenario information from the scenario memory, and infer the target result to be input in the target application window based on the feature information and the at least one target operation scenario information.

[0251] The display unit 603 is also used to output the target result to the target application window.

[0252] For details on the operation of the processor, memory, and display unit, please refer to the relevant descriptions in the preceding embodiments, which will not be repeated here. It is understood that the electronic device may also include an input unit 604.

[0253] Of course, the electronic device can also have more than Figure 6 There are no restrictions on the number of components, whether more or fewer.

[0254] This application also provides a computer program product, including computer-readable instructions, which, when executed on an electronic device, cause the electronic device to implement any of the methods for determining output content provided in this application.

[0255] This application also provides a computer-readable storage medium that carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the methods for determining output content provided in this application.

[0256] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0257] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0258] In the above embodiments, the implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, in the form of a computer program product.

[0259] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A method for determining output content, comprising: determining feature information of a target application window operated by a user; determining at least one target operation scenario information from a scenario memory bank based on the feature information, the scenario memory bank comprising at least one operation scenario information used to represent an operation behavior or an operation habit of the user in a historical operation scenario; reasoning a target result to be input in the target application window based on the feature information and the at least one target operation scenario information; and outputting the target result to the target application window. 2.The method of claim 1, wherein the determining at least one target operation scenario information from the scenario memory bank based on the feature information comprises: obtaining a target operation event input in the target application window, the target operation event used to trigger a reasoning task in combination with a historical operation scenario; and determining at least one target operation scenario information from the scenario memory bank based on the feature information in response to the target operation event.

3. The method of determining output content of claim 1, the episodic memory bank comprising: at least one of a short-term memory set and a long-term memory set, wherein the short-term memory set comprises short-term operation scenario information of the user at at least one time point in a first historical time period, the short-term operation scenario information comprising a historical window interface operated by the user and interface content in the historical window interface; the long-term memory set comprises at least one long-term operation scenario information used to represent at least one of an operation habit and an operation behavior sequence of the user in a historical operation scenario in a second historical time period; wherein the second historical time period is earlier than the first historical time period. 4.The method of claim 3, wherein the at least one long-term operation scenario information comprises at least one of: at least one first type of scenario information used to record interface content of a historical window interface browsed by the user; at least one second type of scenario information used to record interface information of at least one historical window interface continuously operated by the user; at least one operation habit information used to record at least one operation performed by the user in a historical operation scenario. 5.The method of any one of claims 1 to 4, wherein the determining feature information of a target application window operated by a user comprises: determining a window function type corresponding to the target application window switched to by the user; or determining a window function type corresponding to the target application window in which an input operation of the user is performed and input information input in an input box of the target application window. 6.The method of claim 5, wherein, when the feature information of the target application window comprises the window function type corresponding to the target application window and the input information input in the input box of the target application window, the determining at least one target operation scenario information from the scenario memory bank based on the feature information comprises: if the window function type corresponding to the target application window is an interactive window of a smart model, and the input information in the target application window is a first prompt statement, determining, from the scenario memory library, interface content of at least one historical window interface of the user historical browsing and associated with the first prompt statement, wherein the first prompt statement is used at least for prompting content features of interface content displayed in the window interface of the user historical browsing; or, if the window function type corresponding to the target application window is an interactive window of a smart model, and the input information of the target application window is used for indicating that the target article recently read by the user is compared and analyzed with reference articles read before the target article, determining, from the scenario memory library, the target article and at least one of the reference articles, wherein the reference articles have the same keyword as the target article; or, if the window function type corresponding to the target application window is a search box of an operating system, and the input information is an operation execution instruction, determining, from the scenario memory library, operation habit information related to the operation execution instruction. the target result includes: the first prompt statement after expansion; or a second prompt statement obtained by optimizing the input information, the second prompt statement including at least part of content of the target article and at least part of content of at least one of the reference articles; or an operation command sequence including at least one operation command required to complete a target operation indicated by the operation execution instruction.

7. The method of determining output content according to claim 5, in the case where the feature information of the target application window includes a window function type corresponding to the target application window, the determining at least one target operation scenario information from the scenario memory library based on the feature information includes: if the window function type of the target application window indicates that the target application window is an information recording window used for information recording, determining, from the scenario memory library, interface content of at least one historical window interface browsed by the user in a first historical time period; or, if the window function type of the target application window is an interactive window of a first smart model, determining, from the scenario memory library, historical interaction information of the user in at least one interactive window of a second smart model in a first historical time period; the target result includes: recorded content to be recorded in the target application window; or a third prompt statement associated with the historical interaction information.

8. The method of determining output content according to claim 1, the determining at least one target operation scenario information from the scenario memory library based on the feature information includes: determine a target time range based on a user's adjustment operation on a time adjustment control, wherein the electronic device can present an operation context label corresponding to the target time range in response to the adjustment operation, the operation context label being used to represent a content digest corresponding to operation context information in the episodic memory bank that belongs to the target time range; determine at least one target operation context information from the episodic memory bank that corresponds to the target time range and matches the feature information.

9. The method of determining output content according to claim 1, wherein the reasoning of the target result to be input in the target application window based on the feature information and the at least one target operation context information comprises: determining and presenting at least one candidate operation intent and an operation context digest referenced by the candidate operation intent based on the feature information and the at least one target operation context information, the operation context digest being used to indicate target operation context information on which the candidate operation intent is determined; determining a target operation intent selected by the user from the at least one candidate operation intent; reasoning the target result to be input in the target application window based on the target operation intent, the feature information and the at least one target operation context information.

10. An electronic device comprising: a display unit, a memory and a processor; the display unit is configured to display a target application window operated by a user; the memory is configured to store an episodic memory bank, the episodic memory bank comprising at least one operation context information, the operation context information being used to represent operation behavior or operation habit of the user in a historical operation context; the processor is configured to determine feature information of the target application window, determine at least one target operation context information from the episodic memory bank, and reason a target result to be input in the target application window based on the feature information and the at least one target operation context information; the display unit is further configured to output the target result to the target application window.

Images