Intelligent adsorption assistance recommendation model

Abstract

The technology disclosed herein enables a system to provide intelligent fast assistant recommendations using a diverse set of factors and factor weights. To generate recommendations, the system receives user input that places a first item in a region of a sticky configuration in a display environment. In response, the system configures confidence scores for a plurality of items, including items that are open in the display environment as well as items that are not open. The system then ranks the items based on the confidence scores and selects a list of recommended items from the ranked list. The recommended items are then presented in a second region of the sticky configuration for selection. The system is also configured to receive and analyze user selections of the sticky items to learn over time and adjust the confidence scoring.

Description

Background Technology

[0001] With the increasing prevalence of remote work, personal computing devices (such as laptops, tablets, and desktop computers) have become the primary workspace for a growing number of users. Providing an efficient and intuitive user experience is a crucial aspect of the software platforms and operating systems that power these personal computing devices. In many scenarios, a software platform can include a desktop environment provided by the operating system that displays multiple items, such as applications, files, media content, etc.

[0002] To streamline the use of personal computing devices, software platforms can offer multitasking features to allow users to easily organize items in a desktop environment. For example, one multitasking feature provides a side-by-side grid where items are positioned and resized. Users can initiate a "snap" operation by placing an item in a position on one side of the display and then selecting an additional item to snap to another position on the other side of the display. This can be achieved using a snap-access menu that responds to user input initiating the snap operation, presenting the user with available items for snapping.

[0003] Unfortunately, typical snapping assistance solutions only present items already open in the desktop environment, thus limiting the snapping options available to the user. Furthermore, existing snapping assistance solutions only display available items without any organization or underlying logic to dictate how or which items are displayed. For example, items presented in the snapping assistance menu might be irrelevant to the user's current task, causing them to abandon the snapping assistance feature, resulting in a degraded user experience and reduced productivity. In another example, a user might want to snap to an application, file, or website that is not currently open in the desktop environment. Because existing snapping assistance solutions are limited to currently open items, users must spend significant time and resources manually finding and opening the desired item, further degrading the user experience.

[0004] It is precisely because of these and other considerations that this article is presented publicly. Summary of the Invention

[0005] The disclosed technology improves the efficiency and functionality of software platforms that provide snapping-assisted multitasking features. Generally, the system utilizes machine learning models to analyze multiple factors to provide recommendations in the snapping-assisted user interface (UI) in response to user input initiating the snapping operation. Snapped items enable users to organize projects and streamline productivity within a desktop environment.

[0006] In various examples, the desktop environment can be provided by personal computing devices such as laptops, tablets, foldable computing devices, desktop computers, etc. Additionally, the disclosed techniques can be used in environments other than desktop environments, such as mobile devices, large-format displays in conference rooms, or other display environments. Items displayed in one of these types of environments can include applications, files, websites, events, etc. Snapping items can involve a user initiating a snapping operation by placing the displayed item in a first position on an organized grid. In response, the system can present the user with a snapping assistance menu that displays available items that can be snapped to additional positions on the organized grid.

[0007] As discussed above, existing snap-in assistance solutions only present items currently open and displayed in the display environment. However, the items currently displayed may be irrelevant to the user's current task, thus defeating the purpose of quick assistance. In another example, a user might want to snap to items that are not currently open. Therefore, the user must manually locate and open the item they want to snap to in association with the first snapped item. Therefore, there is a need for software platforms that provide snap-in assistance with multitasking features to offer more meaningful recommendations, streamline the user experience, and enhance productivity.

[0008] In various examples, machine learning models can be trained using diverse sets of system signals, applied signals, and / or historical user data to provide meaningful adsorption-assisted recommendations. This set of signals and data may also be referred to as factors in this paper. By providing adsorption-assisted recommendations that are useful to the user, the system can streamline the user experience and improve productivity. Furthermore, the system can be adjusted over time to adapt to the habits, tasks, and past actions of specific users. Therefore, quick assistance suggestions can be improved over time and customized for individual users. Additionally, recommended items displayed in the adsorption-assisted UI can be presented in a visual hierarchy by highlighting the most likely useful items. By guiding the user's attention to the most useful items, the user experience can be further streamlined and enhanced.

[0009] In contrast to existing solutions, training machine learning models using various factors enables intelligent snapping recommendations. Previously, snapping recommendations only included currently open items and were presented without any organization or underlying intelligence. Therefore, existing snapping assistance solutions made it difficult for users to find the desired items, thus increasing the time required for users to set up their workspaces. As discussed in more detail below, the provided intelligent snapping recommendations can include items that are not currently open or displayed in the display environment but are considered relevant to the user's current task. By streamlining the snapping process with useful and meaningful snapping recommendations, users can reach peak productivity faster and more efficiently.

[0010] To provide adsorption-assisted recommendations, the system can receive user input initiating an adsorption operation to adjust the size of a first item in a first position within an organized grid and / or to position the first item within that first position. In response, the system can determine a confidence score for the set of items based on a set of factors discussed above. The system can rank the items in the set according to the confidence scores and select items with confidence scores that meet a threshold. In one example, the system can determine items with confidence scores exceeding a threshold to recommend to the user. In another example, an item can be selected based on its position within one of the top N positions in a ranked list, where N is the threshold number of the highest-ranked items. In a specific example, the ranked list could contain ten items, and only the top five (e.g., N=5) items could be selected to recommend to the user. In various examples, the threshold N can be a predefined number of the highest-ranked items and configured by a system engineer or other administrative entity. The threshold N can also be adjusted over time to suit different users or contexts.

[0011] Furthermore, the system can adjust various thresholds to provide an intuitive user experience. For example, the system can initially select recommended items with confidence scores exceeding a certain threshold. However, in some cases, the number of recommended items with sufficiently high confidence scores may fall below the threshold number of items (e.g., only one item qualifies for recommendation). In response, the system can lower the threshold confidence score and re-evaluate the candidate items for recommendation. In this way, the system can generate a candidate set of recommended items to ensure that many options are available to the user.

[0012] These factors can include those concerning a specific item, such as frequency of use, recentity of use, pre-existing location in the display environment, adsorption frequency, etc. These factors can also include the relationship between the item and the first item it adsorbs, such as the frequency with which the item is grouped with the first item, the level of the relationship between the item and the first item (e.g., similar titles, categories or types of similar items, etc.), and pre-existing location in the display environment associated with the first item. Additionally, weights can be applied to factors to emphasize or de-emphasize individual factors. The values ​​of the weights can be determined and / or adjusted (e.g., learned) over time based on historical user data (such as past user adsorption choices). This determination and adjustment can be based on adsorption choices indicative of trends and / or patterns in a general population (e.g., a large number of users of adsorption aids), or it can be based on adsorption choices indicative of trends and / or patterns in individual users.

[0013] Additionally, the display engine can receive a ranked list of recommended items from a machine learning model and generate intuitive and / or customized recommendation layouts to present the items to the user. For example, the display engine can generate recommendation layouts to display snapped recommendations within the visual hierarchy discussed above, drawing the user's attention to highly ranked items. In a specific example, the recommendation layout could include a first area displaying the first item in a magnified format near the top of the display device and a second area displaying supplementary items in a smaller format below the first item. The display engine can also generate recommendation layouts based on the size of the ranked list to improve user access to and viewing of recommended items. For example, the recommendation layout could include a scrollable menu to allow the user to view supplementary items in the ranked list.

[0014] Furthermore, the display engine can receive user input regarding item selection from the snap-in auxiliary UI and adjust future layout generation based on this input. For example, if the user selects a high-ranking item that is highlighted in the layout, the display engine can detect the user input and continue generating the layout accordingly. Conversely, if the user selects a lower-ranking item that is not highlighted, the display engine can adjust future layouts and customized recommendation presentations over time based on user activity. In specific examples, the display engine can adjust various weights to refine future confidence scores and ranking lists.

[0015] As described above, and further as further illustrated herein, by employing machine learning models to provide intelligent snapping recommendations, the disclosed system enables users to organize various types of display environments (e.g., desktop environments) more efficiently and effectively. This is achieved by calculating confidence scores using multiple factors and providing a ranked list of recommended items. These factors help provide a clear picture of the user's intent and allow the machine learning model to provide recommendations that are more likely to be useful to the individual user. In existing solutions without intelligent snapping recommendations, users may need to manually locate and open desired items, degrading the user experience and unnecessarily consuming computational and network resources. By providing intelligent snapping recommendations, users can organize their display environments more quickly and efficiently to achieve maximum productivity, thereby saving computational resources and improving performance.

[0016] Features and technical advantages beyond those explicitly described above will become apparent upon reading the following detailed description and viewing the associated drawings. This summary section is provided to present a selection of concepts in a simplified form, which will be further described in the detailed description below. This summary section is not intended to identify key or essential features of the claimed subject matter, nor is it intended to serve as an aid in determining the scope of the claimed subject matter. For example, as permitted by the foregoing context and throughout this document, the term "technology" may refer to (multiple) systems, (multiple) methods, computer-readable instructions, (multiple) modules, algorithms, hardware logic, and / or (multiple) operations permissible in the described context. Attached Figure Description

[0017] Detailed embodiments are described with reference to the accompanying drawings. In the drawings, the leftmost numeral of the reference numeral(s) indicates the drawing in which the reference numeral(s) first appear. Identical reference numerals in different drawings indicate similar or identical items. References to individual items among multiple items can be made using reference numerals with letters from a alphabetical sequence to refer to each individual item. General references to items can be made using specific reference numerals without an alphabetical sequence.

[0018] Figure 1A An environment with adsorption-assisted multitasking features is shown, which utilizes intelligent adsorption recommendation to present multiple recommended items.

[0019] Figure 1B The environment is shown after the user selects a recommended item to snap to next to the first item, displaying the snapped-to-adhere configuration.

[0020] Figure 2 This is a block diagram illustrating various aspects of machine learning methods that use various factors and factor weights to calculate confidence scores.

[0021] Figure 3AThis is an example dataset that shows the system based on Figure 2 The operation shown generates confidence scores for ranking and recommending projects.

[0022] Figure 3B This is another example dataset, which shows the system based on Figure 2 The operation shown generates confidence scores for ranking and recommending projects.

[0023] Figure 4 Additional aspects of an environment with adsorption-assisted multitasking features utilizing intelligent adsorption recommendations are shown.

[0024] Figure 5 This is a flowchart illustrating various aspects of routines used to implement the techniques disclosed herein.

[0025] Figure 6 This is a block diagram illustrating the functionality of a layout engine that generates a recommended layout for displaying recommended items.

[0026] Figure 7 An environment is shown that features adsorption-assisted multitasking with intelligent adsorption recommendations and a visual hierarchy that highlights high-ranking items.

[0027] Figure 8 Additional aspects of the environment are shown, including adsorption-assisted multitasking features that utilize intelligent adsorption recommendations and a visual hierarchy that prominently displays high-ranking items.

[0028] Figure 9 Additional aspects of the environment are also shown, including adsorption-assisted multitasking features that utilize intelligent adsorption recommendations and a visual hierarchy that prominently displays high-ranking items.

[0029] Figure 10 This shows the implementation Figure 7 The flowcharts show various aspects of the routines for the visual hierarchy shown.

[0030] Figure 11 This is a computer architecture diagram illustrating the computer hardware and software architecture used in computing systems capable of implementing various aspects of the technologies and processes presented herein. Detailed Implementation

[0031] The techniques described herein provide a system for improving the operation of adsorption-assisted multitasking characteristics in a display environment by introducing a machine learning model that intelligently recommends items for adsorption using various factors. These factors may include application signals, system signals, historical user data, and / or other data that define past activities or habits that influence the likelihood of a user adsorbing a particular item. Recommended items may include applications, files, websites, operating system menus, contacts, or other digital content including associated graphical representations displayed in the environment. For example, these graphical representations may generally be referred to as user interface elements or windows. Furthermore, items can be freely moved within the environment and presented in a separate or overlapping (e.g., stacked) manner.

[0032] The disclosed technology addresses several technical problems associated with adsorption-assisted recommendation generation and presentation. For example, the disclosed system overcomes the inefficiencies and limitations of existing adsorption-assisted solutions that do not provide intelligent adsorption-assisted recommendations. This typically leads to a degraded user experience, resulting in reduced productivity. As described in more detail below, the disclosed technology can also improve the performance of various personal computing devices. By providing intelligent adsorption-assisted recommendations, the disclosed technology saves computational resources, such as processing cycles, memory resources, and network resources, through a streamlined user experience.

[0033] Traditional approaches to snapping-in assist solutions are limited in several ways. First, only currently open items displayed in the environment are eligible to be shown in the snapping-in assist user interface (UI). Therefore, if a user wants to snap to an item that is not currently open, they must manually locate and open the desired item. Second, existing snapping-in assist UIs only present currently open items, lacking additional organization or underlying intelligence. These limitations typically lead to reduced productivity and a degraded user experience.

[0034] In contrast, the disclosed technique provides intelligent snap-in-the-flavor recommendations generated by a machine learning model based on a diverse set of factors. Generating snap-in-the-flavor recommendations in this way ensures that recommended items are relevant to the user's current task, enabling the user to quickly organize their display environment and achieve maximum productivity. Furthermore, by leveraging machine learning models and various factor weights, the snap-in-the-flavor recommendations described herein can improve over time and deliver a personalized experience for individual users. The disclosed technique further streamlines the user experience by providing a layout that visually presents snap-in-the-flavor recommendations in a hierarchy that draws the user's attention to high-ranking items.

[0035] In another example of the technical effects of this disclosure, the intelligent adsorption-assisted recommendation described herein improves the performance of personal computing devices. This is possible because adsorption recommendation utilizes user input to refine and customize the generation and display of recommended items to improve over time. By tailoring the user experience to individual users, the disclosed system allows users to organize the display environment even faster as the system adapts to specific habits and contexts. Therefore, by continuously receiving user input and improving the user experience, the system can improve performance by conserving various computing resources.

[0036] The following descriptions, with reference to Figures 1 to 11, illustrate various examples, scenarios, and aspects of implementing intelligent adsorption-assisted recommendation generation and presentation.

[0037] Figure 1A A display environment (e.g., a display screen) 100 is shown, in which a calendar application 102 is the first item snapped to the left side of the display screen as a result of user input, occupying a first area of ​​the display environment. The display environment may encompass a display device that is part of a personal computing device (such as a laptop, tablet, desktop computer, etc.). Alternatively, the display environment may be a display device that is part of a shared facility (e.g., a meeting room or "central" facility, kiosk, etc.). In various examples, the user input that triggers the snapping of the calendar application 102 may be a gesture of dragging the calendar application 102 to the edge of the display environment. The gesture can be performed using a pointing device (such as a mouse) and executing a dragging motion to the edge of the display environment. Alternatively, user input can be performed using a finger or stylus on a touch-sensitive display and similarly executing a dragging motion to the edge of the display environment.

[0038] In response to user input performing a snap gesture, the system can display multiple recommended items, such as a spreadsheet application 104, an active communication session 106, and a communication application 108, in a second area of ​​the display environment. These recommended items 104, 106, and 108 allow the user to select and snap a specific item to the calendar application 102. In this example, and as described above, each recommended item has an associated confidence score calculated based on various factors, such as frequency of use, recentity of use, etc. As will be described in further detail below, recommended items 104, 106, and 108 are part of a ranked list of recommended items generated by machine learning model 110 and are selected for presentation in display environment 100 based on the confidence scores. It should be understood that machine learning model 110 can be implemented in any manner. In some examples, machine learning model 110 is implemented as part of a personal computing device providing display environment 100. Alternatively, machine learning model 110 can be implemented in a remote and / or network-based server environment communicating with the personal computing device.

[0039] In a specific example, the system can detect that the first adsorbed item is calendar application 102. Therefore, the system can determine that the user frequently uses spreadsheet application 104 simultaneously with calendar application 102 to perform data entry tasks, and may adsorb calendar application 102 and spreadsheet application 104 together. Therefore, the system can assign a high confidence score to spreadsheet application 104 and include spreadsheet application 104 in the recommended items. In another example, the system can detect that the user is currently engaged in communication session 106. In this case, the communication session may have recently been displayed in the foreground of the display environment. Therefore, the system can determine that communication session 106 may require additional attention from the user. Therefore, the system can assign a high confidence score to communication session 106 and present communication session 106 as a recommended item to the user. In a similar manner, the system can determine that the user typically uses calendar application 102 and communication application 108 in combination to schedule meetings. Similar to spreadsheet application 104, the system can assign a high confidence score to communication application 108 and recommend communication application 108 to the user.

[0040] Turning Figure 1B The user subsequently selected spreadsheet application 104 to snap next to calendar application 102. Therefore, the system enlarged spreadsheet application 104 to fill a second area of ​​the display environment. Multiple areas (e.g., two, three, four) and the size and position of each area to which items are snapped can be pre-configured as an organized layout or grid. Additionally, machine learning model 110 can receive and store user selection data 112 defining the user's selection of spreadsheet application 104. As described above and further described in detail below, machine learning model 110 can analyze user selection data 112 to understand future snapping assistance suggestions and customize the user experience based on the specific habits and context of individual users. In various examples, machine learning model 110 can initially be trained using user data from many users (e.g., a global user library) and reflects large-scale habits within a general user group. Furthermore, individual instances of machine learning model 110 can be associated with specific users or user devices and gradually adapted to specific habits to provide a customized experience.

[0041] It should be understood that, although Figure 1B The example shown depicts items being snapped into an organized grid within a single display device, but the snapping auxiliary interface can utilize multiple display devices. For example, in a multi-monitor configuration, a first item can be snapped into place to fill a first area in a first display device, and a second item can be snapped into place to fill a second area in a second display device. Additionally, although Figure 1BThe example shown illustrates a side-by-side format, but the arrangement and number of items in an organized layout can be automatically configured by the system or modified by the user to create a custom preferred grid. For example, a user can use a default layout (such as...) Figure 1B The dual-item layout shown 102 snaps items together, with the two items displayed at similar sizes. Users can adjust the grid boundaries to increase the size of the calendar application 102 and decrease the size of the spreadsheet application 104. Additionally, users can specify a preferred grid format to snap items together using custom boundaries, the number of items, etc.

[0042] Turn now Figure 2 The machine learning model 202 (e.g., machine learning model 110) that generates intelligent adsorption-assisted recommendations will be interpreted. To generate adsorption-assisted recommendations, the machine learning model 202 receives multiple items 204. As mentioned above, items can be standalone applications, files, websites, contacts, in-app activities, etc. In contrast to existing solutions, the items provided to the machine learning model 202 are not limited to open and currently displayed items.

[0043] Additionally, the machine learning model 202 receives a set of factors 206 to calculate a confidence score for each item 204. Factors 206 may include a variety of data and signals obtained from various items, data repositories, and the operating system providing the display environment. As mentioned above, these factors may include various characteristics of the item, such as frequency of use, recentity of use, location in the display environment, download location, etc. These factors may also include relationships and associations between the item and the first attracted item, such as indicators of how frequently the item is associated with other items. Figure 1A Factors that are used simultaneously with the first item as discussed in the examples. Other examples may include factors that indicate that the item belongs to a similar category to the first item (e.g., productivity app, media app, etc.), factors that are related to items with similar titles or keywords, factors that indicate that the item is frequently snapped up together with the first item, and so on.

[0044] Furthermore, the machine learning model 202 can receive a set of factor weights 208 used to emphasize or de-emphasize individual factors. Weights 208 can be predetermined values ​​used to prioritize certain factors over others. For example, factors indicating that an item frequently attaches next to the first item can be weighted more heavily than factors related to similar titles or keywords. Factor weights 208 can also be adjusted by the machine learning model 202 using user activity data 210, which defines various activities such as certain habits in app usage, different contexts in which the attachment-assisted UI is invoked (e.g., productivity, leisure, entertainment), etc. As mentioned above, user activity data 210 can be an aggregation of habits from multiple users (e.g., within a family, organization, or globally) to allow the machine learning model to adapt to large-scale trends in attachment-assisted usage. Alternatively, user activity data 210 can define activities for individual users, enabling the machine learning model 202 to customize attachment-assisted features to suit individual users and provide a customized experience.

[0045] After receiving item 204, factor 206, factor weight 208, and user activity data 210, machine learning model 202 can calculate a confidence score for each item in the project and generate a set of rated items 212. The system can then rank the rated items to generate a ranked list 214 of items. As described in more detail below, the system can then select a set of recommended items 216 from the ranked list to present to the user in the adsorption-assisted UI. Since the ranked list 214 may contain a large number of items, presenting the entire list may be impractical. In these cases, the system can select the highest-ranked items to display. As described above, the system can use a threshold confidence score or a threshold number of items to select recommended items 216 and adjust the individual thresholds. As described, when the number of recommended items 216 is lower than the threshold number of items, the system can lower the threshold confidence score.

[0046] The system can subsequently receive items selected from the recommended items 216 to be snapped into the first item (such as...). Figure 1A The user input is shown next to the calendar application 102 (as illustrated). Additionally, the machine learning model 202 can receive data defining the selected items 218 to understand future item ratings. For example, the machine learning model 202 can use the selected items 218 to adjust factor weights 208 and update the subsequent set of rated items 212. In this way, the selected items become user activity data 210, and the system can continuously improve over time to consistently recommend useful items that are relevant to the user's current tasks.

[0047] Turn now Figure 3A This shows an example of confidence score calculation and project ranking. Figure 3A The examples discussed in the text are based on Figure 1A and Figure 1B The example shown and Figure 2 The operation described herein. In this example, the set of items 302 and the set of factors 304A-304F are shown in Table 306. Additionally, confidence score 308 is calculated based on the set of parameters 310. In this case, factors are given equal weight, individual factor scores are generated and summed to produce the final score, and the threshold score used to select items for recommendation is set to twenty.

[0048] Based on parameter 310, factor scores 304A to 304F for items A through E are listed in Table 306. In Table 306, based on factor 206, the factor score 304 for each item in items A through E is determined based on the analysis of item 204 performed by machine learning model 202. For example, machine learning model 202 may include methods for quantifying various factors of item 204 into corresponding factor scores 304. The total score 312 for each item (also known as the confidence score 308) can then be generated using the individual's factor scores 304.

[0049] Since parameter 310 indicates that the same weights are applied to the factor scores 304, the total score 312 can be determined as the sum of the factor scores 304 for the corresponding items. The total score 312 for each item is then compared to a threshold score of twenty. Items with a total score 312 higher than the threshold score are identified as recommended items 314 and, in accordance with... Figure 1A The adsorption-assisted UI shown is presented to the user. According to one aspect of the implementation, items are ranked based on their confidence score of 308 before an application threshold score of 20 is applied. In this way, the recommended items are also a ranking list, in which items determined by the machine learning model to be most likely to be adsorbed are listed at the top of the list.

[0050] exist Figure 3A In the example shown, confidence score 308 is calculated based on usage frequency score 304A and usage proximity score 304B, which indicate the level of user interaction with a particular item. These factors also include a Z-order score 304D, which indicates whether the item is close to the foreground of the display environment. Finally, these factors include grouping score 304C, adsorption frequency score 304E, and configuration score 304F, which may individually or collectively indicate the level of association between items 302A through 302E and the first adsorbed item. It should be understood that factor 206 considered in calculating confidence score 308 may include ratios... Figure 3AThe system can consider far more factors than those shown. For example, it can consider the project's creation date (e.g., older projects may generally be less relevant than newer ones), the project's status (e.g., whether the project is currently open), the project's starting location (such as where the project was downloaded or stored), or the project's type or category (e.g., productivity, entertainment, etc.). The system can even consider the hardware parameters of the computing device presenting the display environment, such as the screen resolution, screen orientation (e.g., portrait or landscape mode), and the number of screens.

[0051] exist Figure 3A In the example shown, item A is the highest-ranking item due to its high frequency of use score 304A and recentity of use score 304B, indicating that it is frequently used by users and that its recent use increases the likelihood of adsorption. Item A also has a high grouping score 304C, indicating that it is frequently used consecutively or concurrently with the first item in the task flow, but not necessarily in the adsorption configuration. For example, Figure 1A The communication application 108 shown can have a high grouping score because it is frequently used concurrently with the calendar application 102 to find and schedule meetings. Additionally, item A has a high snapping frequency score 304E, indicating that the user frequently snaps item A into the display environment, regardless of its position in the snapping configuration. Similarly, item B has high scores in usage frequency 304A, usage proximity 304B, and snapping frequency 304E. Furthermore, item B has a high Z-order score, indicating that item B was recently displayed in the foreground of the display environment (e.g., stacked on top of other items in the display environment). For example, Figure 1A The active communication session 106 shown can have a high Z-order score because the user has recently been interacting with the window in the foreground.

[0052] Item D is the final recommended item, primarily due to its high configuration score, indicating that Item D is typically displayed with a specific configuration and / or in a position adjacent to the snapped-in items. For example, in Figure 1A In this system, it can detect that when a user snaps the calendar application 102 to the left side of the display environment, he or she typically snaps the spreadsheet application 104 to the right. Based on these past snapping configurations, the system can assign a higher configuration score to the spreadsheet application 104 accordingly.

[0053] exist Figure 3B In the example shown, the threshold confidence score defined by parameter 310 remains at twenty; however, item C is now among the recommended items. This is because it corresponds to... Figure 2The factor weights 316 and 318 of factor weight 208 are added as shown. In this case, the grouping score 304C and the configuration score 304F are heavily weighted, indicating the system's emphasis on the relationship between the item and the first attracted item. Item C, which has a moderate level of correlation with the first attracted item as indicated by the configuration score 304F, benefits from the heavy weighting of the configuration score and is therefore eligible to be included in the recommended items. The weights can be generated based on the analysis of user activity data 210 and selected items 218 by the machine learning model 202. As discussed, the weights can be adjusted by the machine learning model over time to adapt to user habits, thereby refining the generation of recommended items and customizing the user experience.

[0054] It should be understood that Figure 3A and Figure 3B The examples shown are merely illustrative and should not be construed as limiting. A set of various factors 206, weighting schemes, and / or mechanisms for scoring item 212 can be utilized to determine a ranking list 214 and recommended items 314 based on the preferences of the individual or entity defining parameter 310. For example, if the individual or entity setting the parameter values ​​the relationship between an item and the first attracted item, then if a weighted score is to be calculated, the grouping score 304C and the configuration score 304F should be considered and assigned higher weights.

[0055] Turn now Figure 4 Additional aspects of the system used to generate and display recommended items will be shown and discussed. In this example, such as Figure 1A The display environment 100 shown is again illustrated, with the calendar app 102 snapped into a first area on the left side of the display environment. Therefore, the snapped auxiliary UI is displayed in a second area on the right side of the display environment. Figure 1A In a similar way to the environment, due to, Figures 2 to 3B The high confidence scores calculated by the machine learning model 202 described herein indicate that active communication session 106 and communication application 108 are displayed as recommended items. However, in Figure 4 In this example, supplementary table 402 displays multiple additional recommended items to the user. As described above, and in contrast to existing adsorption-assisted solutions, machine learning model 202 can recommend additional items that are not currently open or displayed in the display environment but may have a high confidence score. In this example, these items may include application activities (such as design review meetings 404), documents (such as functional specifications 406, quarterly sales 408, and quarterly reports 410), contacts (such as Daichi F. 412 and the design team 414), as well as websites and other items.

[0056] By expanding the diversity of items that can be presented in the Adsorption Assist UI to include files, events, contacts, and more, the system can provide more specific and potentially helpful recommendations. Furthermore, by including items not currently open in the display environment and providing users with shortcuts to relevant items, the system allows users to quickly set their preferred workspace, further streamlining the user experience. Therefore, these enhancements to the Adsorption Assist UI enable users to reach peak productivity in less time compared to existing Adsorption Assist solutions, thereby improving the performance of personal computing devices.

[0057] Based on the various aspects of the system as described above, additional recommended items 404 to 414, along with activity communication session 106 and communication application 108, are part of the ranking list of items. Furthermore, additional items 404 to 414 are selected based on associated confidence scores for presentation in the adsorption auxiliary UI. For example, when it is detected that a user has adsorbed calendar application 102, the system can determine that the user frequently uses calendar application 102 as part of a task flow for scheduling and attending meetings. Therefore, the system can recommend communication application 108 to the user, as it is also part of a meeting task flow. However, more specifically, the system can also detect that the user is part of an upcoming design review meeting 416, as specified in calendar application 102. Therefore, the system can recommend items that may be related to a specific meeting, such as design review meeting 404, functional specifications 406, and design team 414. Furthermore, it should be understood that... Figure 4 The recommended items shown are those that meet or exceed the criteria in the ranking list. Figure 3A and 3B The example shows the items associated with the threshold confidence score. The ranking list of items can contain far more items than are recommended to the user, and the recommendation set can be modified by adjusting the threshold score, factor weights, etc.

[0058] In addition, although Figure 4 The example shown presents recommended items in order of confidence score (e.g., Communication Session 106 has the highest score, and Design Team 414 has the lowest score), but the system can use factor 206 to organize the recommended items in the ad-hoc UI. For example, the recommended items Communication Session 106 and Communication Application 108 could currently be open and displayed in the display environment. Conversely, items 404 through 414 might not currently be open, but received a high confidence score and are therefore recommended to the user. To convey the difference between currently open and closed items to the user, the system can group closed items in supplementary table 402.

[0059] Turn now Figure 5Aspects of the routine 500 for generating intelligent adsorption-assisted recommendations are shown and described below. It should be understood that the operations of the methods disclosed herein are not presented in any particular order, and it is possible and contemplated to perform some or all of the operations in alternative orders. For ease of description and illustration, these operations have been presented in the order of demonstration. Operations may be added, omitted, and / or performed simultaneously without departing from the scope of the appended claims.

[0060] It should also be understood that the method shown can end at any time and does not need to be fully executed. Some or all of the operations of the method and / or substantially equivalent operations can be performed by executing computer-readable instructions included on a computer storage medium, as defined below. The term "computer-readable instructions" and variations thereof, as used in the specification and claims, are used broadly herein to include routines, applications, application modules, program modules, programs, components, data structures, algorithms, etc. Computer-readable instructions can be implemented on a variety of system configurations, including single-processor or multi-processor systems, minicomputers, mainframe computers, personal computers, handheld computing devices, microprocessor-based programmable consumer electronics, combinations thereof, etc.

[0061] Therefore, it should be understood that the logical operations described herein are implemented as (1) a series of computer-implemented actions or program modules running on a computing system and / or (2) machine logic circuits or circuit modules with internal interconnections within the computing system. The implementation is a choice dependent on the performance and other requirements of the computing system. Therefore, the logical operations described herein are referred to differently as states, operations, structural devices, actions, or modules. These operations, structural devices, actions, and modules can be implemented using software, firmware, special-purpose digital logic, and any combination thereof.

[0062] For example, the operation of routine 500 is described herein as being implemented at least in part by a module that runs the features disclosed herein, which may be a dynamic link library (DLL), a static link library, a function produced by an application programming interface (API), a compiler, an interpreter, a script, or any other set of executable instructions. Data may be stored in one or more memory components as data structures. Data can be retrieved from a data structure by addressing a link or reference to the data structure.

[0063] Although the illustrations below refer to the components of the accompanying drawings, it will be understood that the operation of routine 500 can also be implemented in many other ways. For example, routine 500 can be implemented at least in part by the processor or local circuitry of another remote computer. Additionally, one or more operations of routine 500 can alternatively or additionally be implemented at least in part by a chipset that works alone or in combination with other software modules. In the examples described below, one or more modules of the computing system can receive and / or process the data disclosed herein. Any services, circuitry, or applications suitable for providing the techniques disclosed herein can be used in the operations described herein.

[0064] refer to Figure 5 Routine 500 begins with operation 502, where the display environment shows the first set of items. As mentioned above, items can include application activities, files, contacts, websites, etc.

[0065] When the operation reaches operation 504, the user can provide the option to place the first item in the first multiple items in a place such as... Figure 1A The input in the snap-in configuration of the calendar application 102 shown is as follows. As described above, user input can be a gesture of dragging a first item to the edge of the display environment. The gesture can be performed using a pointing device (such as a mouse) and executing a dragging motion to the edge of the display environment. Alternatively, user input can be performed using a finger or stylus on a touch-sensitive display and similarly executing a dragging motion to the edge of the display environment.

[0066] In operation 506, the system responds to the user input of operation 504 by calculating a confidence score 308 for the first plurality of items 204 (e.g., items already existing in the display environment). The confidence score may be based on the user input and one or more factors 206, such as recentity of use, item location in the display environment, level of association with the first item, etc.

[0067] In subsequent operation 508, the system calculates a confidence score 308 for the second set of multiple items 204 (e.g., items not currently open in the display environment). Similar to the first set of multiple items, the confidence score is based on user input and one or more factors 206.

[0068] Upon reaching operation 510, the system can generate a sorted list 214 of items based on the confidence score 308. The sorted list of items may include items from both the first set of multiple items and the second set of multiple items.

[0069] In operation 512, the system can select a set of recommended items 216 from the ranked list 214. Recommended items can have confidence scores that meet a threshold. In various examples, the threshold can be a threshold confidence score, where items with associated confidence scores that meet or exceed the threshold are recommended. In another example, the threshold can be the number of items. For example, if the threshold number of items is five, the top five items in the ranked list are selected for recommendation.

[0070] Finally, in operation 514, recommended item 216 is displayed in the second area of ​​the adsorption configuration, such as... Figure 1A and Figure 4 The example shown. The items displayed in the recommended items can be part of a first set of multiple items (e.g., items currently open in the display environment) or part of a second set of multiple items (e.g., items that are not open).

[0071] Turn now Figure 6 The various aspects of the display engine 602, which provides an intuitive presentation of intelligent adsorption-assisted recommendations, will be explained. To display adsorption-assisted recommendations, the display engine 602 receives data from... Figure 2 The machine learning model 202 shown represents a set of recommended items 216. Additionally, the display engine 602 receives a set of predefined recommended layouts 604 to organize and present the recommended items. As described above, the layouts can present items using a visual hierarchy that highlights items with high positions within the set of recommended items 216. In this way, the user's attention can be drawn to items more likely to be snapped up, further streamlining the user experience. In various examples, and as will be described in more detail below, these default layouts 604 can include various ways of presenting recommended items, such as organizing items into multiple areas within the layout, sorting items across multiple pages, or using a scrollable menu to present many recommended items. Furthermore, the display engine 602 receives a set of user preferences 606, which defines various parameters related to individual users, such as a preference for certain layout formats over others, defining data on past item selections, etc. It should be understood that highlighting items is not limited to the size or position of items within the snapping auxiliary UI. For example, highlighting can include emphasizing certain items.

[0072] Using a set of recommended items 216, a default layout 604, and user preferences 606, the display engine 602 can generate a layout 608 to visually present the recommended items 216 in a snap-in-the-loop UI. The display engine 602 can generate the layout based on a variety of factors. For example, the selection of a specific layout 608 can be based on the size of the list of recommended items 216. In a specific example, if the list of recommended items contains a large number of items, the display engine 602 can choose a multi-page layout to present the items in a visually intuitive and organized manner. In another example, the layout selection can be based on the requirements of a visual hierarchy. Consider a list of recommended items 216 where many items have high confidence scores for highlighting within a visual hierarchy. In this case, the display engine 602 can choose a scrollable layout to present all the recommended items.

[0073] From layout 608, the user can then select an item to snap to next to the first snapped-to item. The data defining the item selection 610 can subsequently be received by the display engine 602 to refine the future layout 608. As will be explained in detail below, the user can select an item to highlight within the visual hierarchy, indicating his or her agreement with the item's ranking and the display engine's presentation. In this case, the display engine 602 can continue generating the layout with minimal changes. Alternatively, the user can select an item with a lower ranking and therefore less prominent within the visual hierarchy. Thus, the display engine 602 can adjust the future layout 608 to take user feedback into account. In a specific example, the display engine 602 can adjust the factor weights 208 to adjust the future confidence score. By reflecting the ranking of recommended items in the visual hierarchy, the system can guide users to the items they are most likely to snap to. Therefore, the user experience is further streamlined, and the time required for individual users to set up an efficient workspace is reduced.

[0074] Now for reference Figure 7 The example shows that environment 700 uses, for example Figure 6 One such layout 608 is shown. As in the previous example, the user has snapped the calendar application 102 to the left side of the display environment 700. In response, the system displays a set of recommended items 702 to 712 on the right side of the display environment 700. (See reference...) Figure 6 As described, recommended items are presented using a visual hierarchy that highlights high-ranking items to attract the user's attention. In this example, active communication session 702 ranks high among the recommended items. Therefore, communication session 702 is shown near the top of the display area and displayed in a magnified format relative to other items (such as spreadsheet application 704). In this way, the user's attention is directed to communication session 702 and they can more quickly engage with the item to set up their desired workspace.

[0075] Additionally, spreadsheet application 704 can be ranked higher than items 706 to 712 but lower than communication session 702, which translates to a moderately prominent position in the layout. In a similar manner to communication session 702, spreadsheet application 704 is displayed above items 706 to 712 and in a medium-sized format relative to the other items in the layout. Finally, supplementary items 706 to 712 are grouped together in supplementary table 714. Items can be grouped together in this way to indicate similar confidence score rankings that elicit similar levels of attention from items 706 to 712.

[0076] Turn now Figure 8 Another example layout for presenting intelligent adsorption-assisted recommendations is shown. Similar to the previous example, in response to user input on the adsorption calendar application 102, a set of recommended items 802 to 814 is displayed on the left side of the display environment 800. Similar to... Figure 7 For example, recommended items 802 through 814 are organized into a visual hierarchy based on their respective positions within the ranking list of items. In this instance, active communication session 802 and spreadsheet application 804 rank high within recommended item 216 and are therefore prominently displayed at the top of the display in a magnified format. Furthermore, communication session 802 and spreadsheet application 804 can be displayed side-by-side to indicate similar confidence scores in addition to their high positions within the ranking list.

[0077] The layout of recommended items may also include a second area displaying the email application 806, oriented towards the center of the snap-in auxiliary UI and displayed in a medium-sized format. In this way, the system can convey to the user that the email application 806 may have some relevance to the snap-in calendar application 102, but less so compared to the communication session 802 and the spreadsheet application 804. Additionally, the layout may include a third area 816 displaying supplementary items 808 to 814 that meet the confidence score threshold used for recommendations, such as... Figure 3A and 3B As shown in the diagram. Items in the third region can be grouped in the third region 816 to indicate similar confidence scores (e.g., score differences within a predefined similarity score difference). Alternatively, items can be grouped within the third region 816 to clearly communicate to the user the associated difference in confidence scores between the highlighted items 802 to 806 and the grouped items 808 to 814.

[0078] Additionally, items can be grouped using criteria other than confidence scores. For example, items in the recommended layout can be grouped based on various factors, such as the level of association with item 102 mentioned above. In a specific example, users can participate in meetings, such as... Figure 4The example illustrates design meeting 416. In this example, display engine 602 can group items related to design meeting 416, such as communication session 802, functional specification document 810, and design team contact 812. Despite differences in confidence scores, items grouped in this way can be prominently displayed in recommendation layout 608 to draw the user's attention to items relevant to the current task at hand. It should be understood that the examples discussed are merely illustrative, and items can be grouped in the recommendation layout in any other way to visually appeal to the user's attention.

[0079] Figure 8 The example recommendation layout also includes an interactive UI element 818 to indicate that additional recommended items are available to the user. This can include a multi-page layout, where selecting the UI element 818 will bring up a new page of recommended items within a third area 816. As mentioned above, the display engine 602 can select this particular layout when the list of recommended items 216 has a certain size. In this example, there may be many lower-ranked items with similar confidence scores, thus requiring a multi-page layout to visually present the collection of recommended items 216.

[0080] Turn now Figure 9 Another example desktop 900 is shown, illustrating a layout for visually displaying intelligent adsorption-assisted recommendations. (Compared to...) Figure 8 The layout shown is similar, with active communication session 902 and spreadsheet application 904 prominently displayed in an enlarged format at the top of the adsorption-assisted UI due to their high confidence scores and associated rankings. However, for Figure 9 The example layout's list of recommended items 216 can include many items with moderate confidence scores, which translate into a medium-sized display in the visual hierarchy. To display the recommended items in an intuitive and organized manner, the display engine 602 can choose a scrollable layout with scrollbars 906 for navigating the list of recommended items. In various examples, the highlighted items (such as communication sessions 902 and project spreadsheet applications 904) can be part of a first set of items with confidence scores that meet a threshold. Conversely, items 908 through 914 can be part of a second set of items with confidence scores that do not meet a threshold. Therefore, the second set of items can be displayed at a lower position in the visual hierarchy relative to the first set of items in the snap-in-the-wall UI.

[0081] In this example, communication application 908, documents 910 and 912, and email application 914 are shown below communication session 902 and spreadsheet application 904 to indicate a lower relative ranking. Additionally, several more items can be displayed further below in a scrollable layout format, organized by their associated ranking within the list of recommended items. As described above, display engine 602 can select this particular layout based on the size of the list of recommended items 216 and the requirements of the resulting visual hierarchy. In this example, many items' confidence scores are converted into a medium-sized format within the visual hierarchy, requiring a scrollable layout format to present the recommended items in an organized manner.

[0082] exist Figures 7 to 9 In each example discussed, the user can select an item snapped next to the calendar app 102. For example... Figure 6 As shown, the system can detect and store user selections accordingly for reference when generating the future layout 608. For example, if the user selects a highlighted item, the system can record the user input as reinforcement of the current user preference 606. Therefore, the system can continue generating the layout with minimal adjustments. Conversely, if the user selects a lower-ranked item, the system can adjust the user preference 606 to refine the future layout 608, thereby customizing the layout for the individual user and ensuring that the highlighted items are relevant and useful.

[0083] Furthermore, the data defining user choices can include various metrics to measure user engagement with a particular recommended layout. For example, item selection data 610 can include data defining how long a user spends selecting an item. In some examples, a user might spend a long time selecting an item, indicating that even items the user expects to be visible within the visual hierarchy may not be easily visible. Display engine 602 can take this information into account to adjust future layouts 608, such as by increasing the overall size of recommended items. Item selection data 610 can additionally include a variety of information, from the number of pages the user searches through (e.g., ...). Figure 8 The layout), and the user's layout (e.g., Figure 9 The display engine 602 can use all the information from the item selection data 610 to refine the recommended layout and quickly help users find the desired items to snap to.

[0084] Turn now Figure 10 The various aspects of the routine 1000 used to present the intelligent adsorption-assisted recommendation are shown and described below.

[0085] refer to Figure 10Routine 1000 begins with operation 1002, in which multiple items are displayed in a display environment. The display environment may include a desktop provided by a personal computing device (e.g., a laptop, tablet), a large-format monitor (such as in a conference room), a smartphone, etc.

[0086] In operation 1004, the user provides input to place the first item in the first display area of ​​the snap-on configuration. As described above, the user input may include a gesture of dragging the first item to the edge of the display environment. The gesture can be performed using a pointing device (such as a mouse) and by performing a dragging motion that sweeps across the edge of the display environment. Alternatively, the user input can be performed using a finger or stylus on a touch-sensitive display and similarly by performing a dragging motion to the edge of the display environment.

[0087] In operation 1006, the system receives a list of recommended items generated by a machine learning model (e.g., machine learning model 202). As described above, the recommended items are ranked based on associated confidence scores calculated using various factors and factor weights.

[0088] In subsequent operation 1008, the system generates a recommended layout to visually present the ranked list of recommended items. The recommended layout is organized using a visual hierarchy to highlight high-ranking items within the list. Prominence within the recommended layout can include the item's position in the display environment, the size of the displayed item, and so on. In this way, the user's attention is drawn to items with high confidence scores.

[0089] Finally, in operation 1010, the system uses the generated recommended layout to present a list of recommended items to the user in the second area of ​​the snap-in configuration.

[0090] Figure 11 Additional details of an example computer architecture 1100 capable of executing the program components described herein are shown. Therefore, Figure 11 The computer architecture 1100 shown illustrates an architecture for server computers, mobile phones, PDAs, smartphones, desktop computers, netbooks, tablets, and / or laptops. Computer architecture 1100 can be utilized to execute any aspect of the software components presented herein.

[0091] Figure 10The illustrated computer architecture 1100 includes a central processing unit 1102 (“CPU”), system memory 1104, and a system bus 1110 coupling the memory 1104 to the CPU 1102. The system memory 1104 includes random access memory 1106 (“RAM”) and read-only memory (“ROM”) 1108. A basic input / output system containing basic routines that help transfer information between components within the computer architecture 1100 (such as during startup) is stored in the ROM 1108. The computer architecture 1100 also includes a mass storage device 1112 for storing an operating system 1114, other data, one or more applications 1116, and one or more modules 1118, such as a machine learning model 202 and a display engine 602.

[0092] Mass storage device 1112 is connected to CPU 1102 via a mass storage controller (not shown) connected to bus 1110. Mass storage device 1112 and its associated computer-readable media provide non-volatile storage for computer architecture 1100. Although the description of computer-readable media herein refers to mass storage devices such as solid-state drives, hard disks, or CD-ROM drives, those skilled in the art will understand that computer-readable media can be any available computer storage or communication media that can be accessed by computer architecture 1100.

[0093] Communication media include computer-readable instructions, data structures, program modules, or other data in modulated data signals (such as carrier waves or other transmission mechanisms), and include any transmission medium. The term "modulated data signal" refers to a signal whose characteristics are altered or configured in a way that encodes information in the signal. By way of example and not limitation, communication media include wired media (such as wired networks or direct-line connections) and wireless media (such as acoustic, RF, infrared, and other wireless media). Any combination of the foregoing should also be included within the scope of computer-readable media.

[0094] By way of example and not limitation, a computer storage medium or computer storage medium may include volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information such as computer-readable instructions, data structures, program modules and other data. For example, computer media includes, but is not limited to, RAM, ROM, EPROM, EEPROM, flash memory or other solid-state storage technologies, CD-ROM, digital versatile optical disc (“DVD”), HD-DVD, BLU-RAY, or other optical storage, cassette tape, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible by computer architecture 1100. For the purposes of the claims, the phrases “non-transitory computer storage medium,” “computer storage medium,” “computer-readable storage medium,” and variations thereof do not include waves, signals and / or other transient and / or intangible communication media themselves.

[0095] Depending on the configuration, computer architecture 1100 can operate in a networked environment using a logical connection to a remote computer via network 1120 and / or another network (not shown). Computer architecture 1100 can be connected to network 1120 via a network interface unit 1122 connected to bus 1110. It should be understood that network interface unit 1122 can also be utilized to connect to other types of networks and remote computer systems. Computer architecture 1100 may also include an input / output controller 1124 for receiving and processing input from various other devices, including a keyboard, mouse, or stylus. Figure 11 (Not shown in the image). Similarly, the input / output controller 1124 can output to a display screen, printer, or other type of output device (not shown in the image). Figure 11 (The output is not shown in the image.)

[0096] In various examples, computer architecture 1100 may include a display device 1126 configured to present information in a visual form. Specifically, display device 1126 may present graphical user interface (“GUI”) elements, text, images, video, notifications, virtual buttons, virtual keyboards, message sending and receiving data, internet content, device status, time, date, calendar data, preferences, map information, location information, and any other information that can be presented visually. In some configurations, display device 1126 is a liquid crystal display (“LCD”) utilizing any active or passive matrix technology and any backlight technology (if used). In some configurations, display device 1126 is an organic light-emitting diode (“OLED”) display. Other display types are conceivable. Display device 1126 may also be a touchscreen configured to detect the presence and location of a touch. Display device 1126 may be a resistive touchscreen, a capacitive touchscreen, a surface acoustic wave touchscreen, an infrared touchscreen, an optical imaging touchscreen, a dispersive signal touchscreen, an acoustic impulse recognition touchscreen, or may utilize any other touchscreen technology.

[0097] It should be understood that the software components described herein, when loaded into and executed, can transform the processing unit 1102 and the entire computer architecture 1100 from a general-purpose computing system into a dedicated computing system tailored to facilitate the functionality presented herein. The processing unit 1102 can be composed of any number of transistors or other discrete circuit elements, which can individually or collectively assume any number of states. More specifically, the processing unit 1102 can operate as a finite state machine in response to executable instructions contained within the software modules disclosed herein. These computer-executable instructions can transform the processing unit 1102 by specifying how it transitions between states, thereby transforming the transistors or other discrete hardware elements constituting the processing unit 1102.

[0098] Encoding the software modules presented herein can also transform the physical structure of the computer-readable medium presented herein. In different implementations of this specification, the specific transformation of the physical structure can depend on various factors. Examples of these factors may include, but are not limited to, the technology used to implement the computer-readable medium, whether the computer-readable medium is characterized as primary or secondary storage, etc. For example, if the computer-readable medium is implemented as a semiconductor-based memory, the software disclosed herein can be encoded on the computer-readable medium by transforming the physical state of the semiconductor memory. For example, the software can transform the state of transistors, capacitors, or other discrete circuit elements constituting the semiconductor memory. The software can also transform the physical state of such components to store data thereon.

[0099] As another example, the computer-readable medium disclosed herein can be implemented using magnetic or optical techniques. In such an implementation, when software is encoded therein, the software presented herein can transform the physical state of the magnetic or optical medium. These transformations may include altering the magnetic properties of a specific location within a given magnetic medium. These transformations may also include altering the physical characteristics or properties of a specific location within a given optical medium to change the optical properties of those locations. Other transformations of the physical medium are possible without departing from the scope and spirit of this specification; the foregoing examples are provided merely for the purpose of discussion.

[0100] In light of the foregoing, it should be understood that numerous types of physical transformations occur within computer architecture 1100 to store and execute the software components presented herein. It should also be understood that computer architecture 1100 may include other types of computing devices, including handheld computers, embedded computer systems, personal digital assistants, and other types of computing devices known to those skilled in the art. It is also contemplated that computer architecture 1100 may not include… Figure 11 All components shown in the diagram may include Figure 10 Other components not explicitly shown in the document, or those that can be utilized with Figure 11 The architecture shown is completely different from the one shown.

[0101] The disclosure presented herein also covers the topics set forth in the following clauses.

[0102] Example Term A, a computer-implemented method for providing recommended items for display in an adsorption configuration, comprising: displaying a first plurality of items in a display environment; receiving user input placing a first item from the first plurality of items in a first display area of ​​the adsorption configuration, wherein the adsorption configuration includes an organized layout in which at least two items are positioned or resized in the display environment; determining, based on the user input and using a machine learning model, a confidence score for each of the first plurality of items other than the first item, based on one or more factors associated with each item in the first plurality of items, wherein the one or more factors include frequency of use, recentity of use, item position in the display environment, and level of association with the first item; determining, based on the user input and using the machine learning model, a confidence score for each of a second plurality of items not currently displayed in the display environment, based on one or more factors; generating a ranking list of the first plurality of items and the second plurality of items based on the confidence scores; determining, from the ranking list, a set of recommended items having confidence scores that satisfy a threshold; and displaying the set of recommended items in a second area of ​​the adsorption configuration.

[0103] Example Clause B, Example Clause A, wherein the item includes at least one of the following: application, website, file, application group, in-app activity, or operating system user interface element.

[0104] The method of Example Clause C, Example Clause A, or Example Clause B, wherein one or more factors also include the project creation date, the project starting location, the project type, and one or more project keywords.

[0105] Example clause D, any of the methods in Example clauses A through C, wherein one or more weights are applied to one or more factors to emphasize or weaken individual factors.

[0106] The methods in Example Term E and Example Term D involve one or more weights being determined based on historical user data of past user choices for recommended items across user groups.

[0107] Example clause F, any of example clauses A through E, wherein the threshold includes a threshold confidence score, and the set of recommended items includes items with confidence scores that meet or exceed the threshold confidence score.

[0108] Example clause G, any of example clauses A through E, wherein the threshold includes a predefined number of the highest-ranked items, and the set of recommended items is determined by selecting a predefined number of the highest-ranked items from the ranking list.

[0109] A method of any one of Example Clause H and Example Clauses A through E, wherein the threshold includes a threshold confidence score, the method further comprising: determining that the number of items in the set of recommended items with confidence scores higher than the threshold confidence score is less than a threshold number of items; reducing the threshold confidence score based on determining that the number of items in the set of recommended items with confidence scores higher than the threshold confidence score is less than the threshold number of items; determining that the number of items with confidence scores higher than the reduced threshold confidence score is equal to or higher than the threshold number of items; and generating a candidate set of recommended items based on the number of items with confidence scores higher than the reduced threshold confidence score, wherein displaying the set of recommended items includes displaying the candidate set of recommended items in a second display area of ​​the adsorption configuration.

[0110] Example Item I, a computational system for providing recommended items for display in an adsorption configuration, includes: one or more processing units; and a computer-readable medium encoded with computer-readable instructions to cause the one or more processing units to: display a first plurality of items in a display environment; receive user input for placing a first item from the first plurality of items in a first display area of ​​the adsorption configuration, wherein the adsorption configuration includes an organized layout in which at least two items are positioned or resized in the display environment; determine a confidence score for each item in the first plurality of items other than the first item based on the user input and using a machine learning model, based on one or more factors associated with each item in the first plurality of items; determine a confidence score for each item in a second plurality of items not currently displayed in the display environment, based on the user input and using a machine learning model, based on one or more factors; generate a ranking list of the first plurality of items and the second plurality of items based on the confidence scores; determine a set of recommended items with confidence scores that meet a threshold from the ranking list; and display the set of recommended items in a second area of ​​the adsorption configuration.

[0111] The system of Example Clause J and Example Clause I, wherein the item includes at least one of the following: application, website, file, application group, in-application activity, or operating system user interface element.

[0112] The system of Example Clause K, Example Clause I, or Example Clause J includes one or more factors such as frequency of use, recentity of use, project location in the display environment, level of association with the first project, project creation date, project starting location, project type, and one or more project keywords.

[0113] Example clause L, any of example clauses I through K, where one or more weights are applied to one or more factors to emphasize or weaken individual factors.

[0114] Example clause M, a system of any one of example clauses I to L, wherein the threshold includes a threshold confidence score, and the set of recommended items includes items with confidence scores exceeding the threshold confidence score.

[0115] A system of any one of Exemplary Clause N, Exemplary Clauses I to L, wherein the threshold includes a predefined number of top-ranked items, and the set of recommended items is determined by selecting a predefined number of top-ranked items from the ranking list.

[0116] A system according to any one of Exemplary Clause O, Exemplary Clauses I to L, wherein the threshold includes a threshold confidence score, the computer-readable medium having instructions that further cause the system to: determine that the number of items in the set of recommended items with confidence scores higher than the threshold confidence score is lower than a threshold number of items; reduce the threshold confidence score based on the determination that the number of items in the set of recommended items with confidence scores higher than the threshold confidence score is lower than the threshold number of items; determine that the number of items with confidence scores higher than the reduced threshold confidence score is equal to or higher than the threshold number of items; and generate a candidate set of recommended items based on the number of items with confidence scores higher than the reduced threshold confidence score, wherein displaying the set of recommended items includes displaying the candidate set of recommended items in a second display area of ​​the adsorption configuration.

[0117] Example Clause P, a computer-readable storage medium having computer-readable instructions encoded thereon to enable a computing system to: display a first plurality of items in a display environment; receive user input placing a first item from the first plurality of items in a first display area of ​​an adsorption configuration, wherein the adsorption configuration includes an organized layout in which at least two items are positioned or resized in the display environment; determine, based on the user input and using a machine learning model, a confidence score for each item in the first plurality of items other than the first item, based on one or more factors associated with each item in the first plurality of items; determine, based on the user input and using a machine learning model, a confidence score for each item in a second plurality of items not currently displayed in the display environment, based on one or more factors; generate a ranking list of the first plurality of items and the second plurality of items based on the confidence scores; determine, from the ranking list, a set of recommended items having confidence scores that meet a threshold; and display the set of recommended items in a second area of ​​the adsorption configuration.

[0118] The computer-readable storage medium of Example Clause Q and Example Clause P, wherein the item includes at least one of the following: application, website, file, application group, in-application activity, or operating system user interface element.

[0119] The computer-readable storage medium of Example Clause R, Example Clause P, or Example Clause Q includes one or more factors such as frequency of use, recentity of use, project location in the display environment, level of association with the first project, project creation date, project starting location, project type, and one or more project keywords.

[0120] The computer-readable storage medium of any of the example clauses S, P to R, wherein one or more weights are applied to one or more factors to emphasize or weaken individual factors.

[0121] A computer-readable storage medium of any one of Exemplary Clause T, Exemplary Clauses P to S, wherein the threshold includes a threshold confidence score, the computer-readable medium having instructions that further cause the system to: determine that the number of items in the set of recommended items with confidence scores higher than the threshold confidence score is lower than a threshold number of items; reduce the threshold confidence score based on the determination that the number of items in the set of recommended items with confidence scores higher than the threshold confidence score is lower than the threshold number of items; determine that the number of items with confidence scores higher than the reduced threshold confidence score is equal to or higher than the threshold number of items; and generate a candidate set of recommended items based on the number of items with confidence scores higher than the reduced threshold confidence score, wherein displaying the set of recommended items includes displaying the candidate set of recommended items in a second display area of ​​an adsorption configuration.

[0122] It should be understood that any reference to elements such as "first" and "second" in the summary of the invention and / or the detailed description is not intended and should not be construed as necessarily corresponding to any reference to elements such as "first" and "second" in the claims. Rather, any use of "first" and "second" in the summary of the invention, the detailed description, and / or the claims can be used to distinguish two different instances of the same element (e.g., two different items, two different regions, etc.).

[0123] Finally, although various configurations have been described in a language specific to structural features and / or methodological actions, it should be understood that the subject matter defined in the appended statements is not necessarily limited to the specific features or actions described. Rather, specific features and actions are disclosed as exemplary forms for implementing the claimed subject matter.

Claims

1. A computer-implemented method for providing recommended items for display in an adsorption configuration, comprising: Display multiple first items in the display environment; Receive adsorption user input, the adsorption user input placing a first item from the plurality of first items in a first display area of ​​the adsorption configuration, wherein the adsorption configuration includes an organized layout in which at least two items are positioned or resized in the display environment; Based on the absorbed user input and using a machine learning model, a first confidence score is determined based on multiple first factors, wherein the first confidence score represents a first relevance to the first item for each other first item among the multiple first items, and wherein the multiple first factors include: the frequency with which the first item is used together with the other first items, the similarity level between the content associated with the first item and the content associated with the other first items, and the first display position of the first item with respect to the second display position of the other first items; Based on the absorbed user input and using the machine learning model, a second confidence score is determined based on multiple second factors, wherein the second confidence score represents a second relevance of the first item to each of multiple second items not currently displayed in the display environment, and wherein the multiple second factors include: the frequency with which the first item and the second item are used together, and the level of similarity between the content associated with the first item and the content associated with the second item; A ranking list of the plurality of first items and the plurality of second items is generated based on the first confidence score and the second confidence score; From the ranking list, determine a set of recommended items with a first confidence score or a second confidence score that meets a threshold; and The set of recommended items is displayed in the second display area of ​​the adsorption configuration.

2. The method of claim 1, wherein the project includes at least one of the following: an application, a website, a file, or an operating system user interface element.

3. The method of claim 1, wherein each of the plurality of first factors and the plurality of second factors further includes project creation date, project starting location, project type, and one or more project keywords.

4. The method of claim 1, wherein weights are applied to each of the plurality of first factors and the plurality of second factors to emphasize or de-emphasize individual factors.

5. The method of claim 4, wherein the weights are determined based on historical user data of past user choices for recommended items across user groups.

6. The method of claim 1, wherein the threshold comprises a threshold confidence score established to identify sufficient correlation between two items, and the set of recommended items comprises items having a first confidence score or a second confidence score that meets or exceeds the threshold confidence score.

7. The method of claim 1, wherein the threshold comprises a predefined number of top-ranked items, and the set of recommended items is determined by selecting the predefined number of top-ranked items from the ranking list.

8. The method of claim 1, wherein the threshold includes a threshold confidence score, and the method further comprises: The number of items in the set of recommended items that have a first confidence score or a second confidence score higher than the threshold confidence score is less than the threshold number of items; The threshold confidence score is reduced based on the determination that the number of items in the set of recommended items with a first confidence score or a second confidence score higher than the threshold confidence score is lower than the threshold number of items; The number of items with a first confidence score or a second confidence score higher than the reduced threshold is equal to or higher than the threshold number of items; as well as A candidate set of recommended items is generated based on the number of items having a first confidence score or a second confidence score higher than the reduced threshold confidence score, wherein the set of recommended items is displayed in the second display area of ​​the adsorption configuration, and the candidate set of recommended items is displayed in the second display area of ​​the adsorption configuration.

9. The method of claim 1, wherein the first item of the plurality of first items is adsorbed to fill the first display area of ​​the organized layout of the adsorption configuration.

10. A computational system for providing recommended items for display in an adsorption configuration, comprising: One or more processing units; as well as A computer-readable medium having computer-readable instructions encoded thereon to cause the one or more processing units to: Display multiple first items in the display environment; Receive adsorption user input, the user input placing a first item from the plurality of first items in a first display area of ​​the adsorption configuration, wherein the adsorption configuration includes an organized layout in which at least two items are positioned or resized in the display environment; Based on the absorbed user input and using a machine learning model, a first confidence score is determined based on multiple first factors, wherein the first confidence score represents a first relevance to the first item for each other first item among the multiple first items, and wherein the multiple first factors include: the frequency with which the first item is used together with the other first items, the similarity level between the content associated with the first item and the content associated with the other first items, and the first display position of the first item with respect to the second display position of the other first items; Based on the absorbed user input and using the machine learning model, a second confidence score is determined based on multiple second factors, wherein the second confidence score represents a second confidence score of the second relevance of the first item to each of the multiple second items not currently displayed in the display environment, and wherein the multiple second factors include: the frequency with which the first item and the second item are used together, and the level of similarity between the content associated with the first item and the content associated with the second item; A ranking list of the plurality of first items and the plurality of second items is generated based on the first confidence score and the second confidence score; From the ranking list, determine a set of recommended items with a first confidence score or a second confidence score that meets a threshold; and The set of recommended items is displayed in the second display area of ​​the adsorption configuration.

11. The system of claim 10, wherein the item includes at least one of the following: an application, a website, a file, or an operating system user interface element.

12. The system of claim 10, wherein each of the plurality of first factors and the plurality of second factors includes a project creation date, a project start location, a project type, and one or more project keywords.

13. The system of claim 10, wherein weights are applied to each of the plurality of first factors and the plurality of second factors to emphasize or de-emphasize individual factors.

14. The system of claim 10, wherein the threshold comprises a threshold confidence score established to identify sufficient correlation between two items, and the set of recommended items comprises items having a first confidence score or a second confidence score exceeding the threshold confidence score.

15. The system of claim 10, wherein the threshold comprises a predefined number of top-ranked items, and the set of recommended items is determined by selecting the predefined number of top-ranked items from the ranking list.

16. The system of claim 10, wherein the threshold includes a threshold confidence score, and the computer-readable medium has instructions that further cause the system to: The number of items in the set of recommended items that have a first confidence score or a second confidence score higher than the threshold confidence score is less than the threshold number of items; The threshold confidence score is reduced based on the determination that the number of items in the set of recommended items with a first confidence score or a second confidence score higher than the threshold confidence score is lower than the threshold number of items; The number of items with a first confidence score or a second confidence score higher than the reduced threshold is equal to or higher than the threshold number of items; as well as A candidate set of recommended items is generated based on the number of items having a first confidence score or a second confidence score higher than the reduced threshold confidence score, wherein the set of recommended items is displayed in the second display area of ​​the adsorption configuration, and the candidate set of recommended items is displayed in the second display area of ​​the adsorption configuration.

17. A computer-readable storage medium having computer-readable instructions encoded thereon to enable a computing system to: Display multiple first items in the display environment; Receive adsorption user input, the adsorption user input placing a first item from the plurality of first items in a first display area of ​​the adsorption configuration, wherein the adsorption configuration includes an organized layout in which at least two items are positioned or resized in the display environment; Based on the absorbed user input and using a machine learning model, a first confidence score is determined based on multiple first factors, wherein the first confidence score represents a first confidence score of first relevance to the first item for each of the multiple first items, and wherein the multiple first factors include: The frequency with which the first item is used together with the other first items is related to the level of similarity between the content of the first item and the content associated with the other first items, and the first display position of the first item with respect to the second display position of the other first items; Based on the absorbed user input and using the machine learning model, a second confidence score is determined based on multiple second factors, wherein the second confidence score represents a second relevance of the first item to each of multiple second items not currently displayed in the display environment, and wherein the multiple second factors include: the frequency with which the first item and the second item are used together, and the level of similarity between the content associated with the first item and the content associated with the second item; A ranking list of the plurality of first items and the plurality of second items is generated based on the first confidence score and the second confidence score; From the ranking list, determine a set of recommended items with a first confidence score or a second confidence score that meets a threshold; and The set of recommended items is displayed in the second display area of ​​the adsorption configuration.

18. The computer-readable storage medium of claim 17, wherein the item comprises at least one of the following: an application, website, or operating system user interface element.

19. The computer-readable storage medium of claim 17, wherein each of the plurality of first factors and the plurality of second factors includes a project creation date, a project start location, a project type, and one or more project keywords.

20. The computer-readable storage medium of claim 17, wherein the threshold includes a threshold confidence score, the computer-readable storage medium having instructions that further cause the system to: The number of items in the set of recommended items that have a first confidence score or a second confidence score higher than the threshold confidence score is less than the threshold number of items; The threshold confidence score is reduced based on the determination that the number of items in the set of recommended items with a first confidence score or a second confidence score higher than the threshold confidence score is lower than the threshold number of items; The number of items with a first confidence score or a second confidence score higher than the reduced threshold is equal to or higher than the threshold number of items; as well as A candidate set of recommended items is generated based on the number of items having a first confidence score or a second confidence score higher than the reduced threshold confidence score, wherein the set of recommended items is displayed in the second display area of ​​the adsorption configuration, and the candidate set of recommended items is displayed in the second display area of ​​the adsorption configuration.

Images