Shared file recommendation method, system and device, medium and program product

Abstract

The invention discloses a shared file recommendation method, system and device, a medium and a program product, and the method comprises the steps: firstly obtaining a shared target group content theme sent by a cloud disk server, then calculating a recommended content detection ratio based on the shared target group content theme, and when the recommended content detection ratio is greater than a preset detection ratio threshold value, sending a recommended file to the cloud disk server; based on the file theme content, generating a file recommendation target list through a file theme content matching model; and when the recommended content detection ratio is smaller than a preset detection ratio threshold value, generating a file recommendation target list based on file modification time. By adopting the method and the device, more accurate and more efficient file recommendation can be realized.

Description

Technical Field

[0001] This invention relates to the field of Internet technology, and in particular to a method, system, device, medium, and program product for recommending shared files. Background Technology

[0002] With the widespread adoption of cloud storage technology, cloud drives have become an important tool for users to store, manage, and share files. To improve the convenience of file sharing, existing cloud drive systems generally integrate file recommendation functions. Currently, cloud drives commonly use a file recommendation method based on file creation time when sharing files. When a user selects a file to share in the cloud drive, the system activates the file selection function. This function first obtains the creation or modification time of all files in the target directory, then sorts them in reverse chronological order by modification time, thus filtering out the most recently modified files. Next, the system filters based on file type, excluding file formats that are not supported for sharing, and finally selects a subset of the top-ranked files to generate a recommendation list for users to view and choose to share.

[0003] However, in community settings such as "sharing groups" and "circles" where members share common attributes or interests, the files they share often revolve around specific themes or content. For example, a photography hobby group mainly shares image files, while a project work group mainly shares project documents. If file recommendations are based solely on modification time, the inherent connection between the file content and the target group's current discussion topics or historical interests is completely ignored. The recommended files may be irrelevant to the group's actual needs, resulting in low accuracy, poor user adoption rates, and severely impacting sharing efficiency and user experience. Summary of the Invention

[0004] The purpose of this invention is to provide a method, system, device, medium, and program product for recommending shared files, which can provide more accurate and efficient file recommendations that understand the content topics of a group.

[0005] To achieve the above objectives, this invention provides a file sharing recommendation method applied to a cloud storage terminal, comprising: Retrieve the content topic sent by the cloud disk server to the target group for sharing; Based on the content theme of the target group being shared, the recommended content detection ratio is calculated; When the detection ratio of the recommended content is greater than the preset detection ratio threshold, a list of recommended file targets is generated based on the file topic content and through the file topic content matching model. When the detection ratio of the recommended content is less than the preset detection ratio threshold, a list of recommended target files is generated based on the file modification time.

[0006] Optionally, calculating the recommended content detection ratio based on the content topic of the sharing target group includes: Based on the content theme of the target group to be shared, calculate the predicted execution response time and the computing power support requirements that characterize the user's operational tolerance for the recommendation task. The recommended content detection ratio is calculated based on the predicted execution response time and the computing power support requirements.

[0007] Optionally, the prediction execution response time is calculated based on the quantified value of the total file computing load time and the model processing time extension; wherein, the model processing time extension is the additional time consumed when the cloud disk terminal calls the file topic extraction model or the file content matching model due to the adaptation of the model size and the computing power of the cloud disk terminal; The computing power support requirements are determined based on the average latency of users' historical recommendations and the average response time of cloud disk servers.

[0008] Optionally, the step of generating a list of recommended files based on file topic content using a file topic content matching model includes: Obtain the latest file topic content and cloud disk terminal resource information, and calculate the pre-evaluation efficiency of the file topic content matching model. Obtain the effective threshold for recommendations based on users' historical behavior calculated by the cloud disk server; If the pre-evaluation efficiency is not greater than the effective recommendation threshold, then the file topic content matching model is executed to generate a list of file recommendation targets; If the pre-evaluation efficiency is greater than the recommended effective threshold, then the file topic content is sent to the cloud disk server to instruct the cloud disk server to execute the file topic content matching model and generate a list of recommended file targets.

[0009] Optionally, the latest document topic content is obtained through the following steps: Get the cloud drive terminal file list and the corresponding historical file topic table; By comparing the file list on the cloud drive terminal with the historical file topic table, files without recorded topics are identified. For files without recorded topics, the file topics are extracted using a file topic extraction model to update the historical file topic table and obtain the latest file topic content.

[0010] Optionally, the step of generating a list of recommended file targets based on file modification time includes: Sort the files in the target folder in reverse order of their modification time; Extract the top N files to form the file recommendation target list; where N is a positive integer calculated based on the user interface display area size and user acceptance.

[0011] Optionally, the topic of the shared target group content sent by the cloud disk server includes: Send the last update time of the content topic of the target group to the cloud disk server, which is cached on the cloud disk terminal; The cloud disk server returns the content topic for the target group to share based on the last update time.

[0012] Optionally, after generating the list of recommended targets for the files, the following may also be included: Output the generated list of recommended files to the user. In response to the user's selection of a file from the file recommendation target list and the execution of a file sharing operation, the system submits the recommendation result selection log to the cloud disk server.

[0013] To achieve the above objectives, the present invention further provides a file sharing recommendation method, applied to a cloud disk server, comprising: Based on the context of the chat history shared by users in the group, determine the topic of the content to be shared with the target group; Send the content topic of the target group to the cloud drive terminal; Receive file content from the cloud drive terminal; Based on the file topic content, execute the file topic content matching model to generate a list of recommended file targets; Send the recommended target list of the file to the cloud drive terminal.

[0014] Optionally, determining the topic of content to be shared with the target group based on the chat history context of the user-shared group includes: The last update time of the content topic of the target group being shared is sent by the cloud storage terminal; The effective update interval is calculated based on the last update time and the chat history context of the current user's shared group; If the effective update interval does not exceed the preset update threshold, the shared target group content topic cached in the cloud disk server will be determined as the shared target group content topic to be sent. If the effective update interval exceeds the preset update threshold, a new target group content topic is calculated based on the shared target group content topic cached in the cloud disk server and the newly added part in the chat history context, and the obtained new target group content topic is determined as the target group content topic to be sent.

[0015] Optionally, calculating the effective update interval based on the last update time and the chat history context of the current user's shared group includes: Based on the last update time, determine the incremental content portion generated later than the last update time from the chat history context; The effective update interval is calculated based on the content density information of the incremental content section and the current user's speaking time span and number of speakings in the incremental content section.

[0016] Optionally, the file sharing recommendation method, applied to a cloud disk server, further includes: Based on the similarity quantification weight matching degree between the content topic of the target group to be shared and the content topic of the target group associated with each behavior in the user's historical behavior record, the effective threshold for recommending the user's historical behavior is calculated. The recommended effective threshold is sent to the cloud disk terminal.

[0017] Optionally, the file sharing recommendation method, applied to a cloud disk server, further includes: Receive the selected recommendation results log sent from the cloud disk terminal; Based on the recommended logs, the shared content topic extraction model and the file topic content matching model are trained and updated.

[0018] To achieve the above objectives, the present invention also provides a file sharing recommendation system applied to a cloud storage terminal, comprising: The module for obtaining content topics for sharing with the target group is used to obtain the content topics for sharing with the target group sent by the cloud disk server. The recommended content detection ratio calculation module is used to calculate the recommended content detection ratio based on the content theme of the target sharing group; The first file recommendation target list generation module is used to generate a file recommendation target list based on the file topic content and through a file topic content matching model when the detection ratio of the recommended content is greater than a preset detection ratio threshold. The second file recommendation target list generation module is used to generate a file recommendation target list based on the file modification time when the detection ratio of the recommended content is less than a preset detection ratio threshold.

[0019] Optionally, the recommended content detection ratio calculation module is used for: Based on the content theme of the target group to be shared, calculate the predicted execution response time and the computing power support requirements that characterize the user's operational tolerance for the recommendation task. The recommended content detection ratio is calculated based on the predicted execution response time and the computing power support requirements.

[0020] Optionally, the first file recommendation target list generation module is used for: Obtain the latest file topic content and cloud disk terminal resource information, and calculate the pre-evaluation efficiency of the file topic content matching model. Obtain the effective threshold for recommendations based on users' historical behavior calculated by the cloud disk server; If the pre-evaluation efficiency is not greater than the effective recommendation threshold, then the file topic content matching model is executed to generate a list of file recommendation targets; If the pre-evaluation efficiency is greater than the recommended effective threshold, then the file topic content is sent to the cloud disk server to instruct the cloud disk server to execute the file topic content matching model and generate a list of recommended file targets.

[0021] Optionally, the second file recommendation target list generation module is used for: Sort the files in the target folder in reverse order of their modification time; Extract the top N files to form the file recommendation target list; where N is a positive integer calculated based on the user interface display area size and user acceptance.

[0022] Optionally, the module for obtaining the content topic of the sharing target group is used to: Send the last update time of the content topic of the target group to the cloud disk server, which is cached on the cloud disk terminal; The cloud disk server returns the content topic for the target group to share based on the last update time.

[0023] Optionally, the file sharing recommendation system, applied to a cloud storage terminal, further includes a file recommendation target list output module, used for: Output the generated list of recommended files to the user. In response to the user's selection of a file from the file recommendation target list and the execution of a file sharing operation, the system submits the recommendation result selection log to the cloud disk server.

[0024] To achieve the above objectives, the present invention also provides a file sharing recommendation system applied to a cloud disk server, comprising: The module for determining the topic of content to be shared with the target group is used to determine the topic of content to be shared with the target group based on the context of the chat history of the user sharing the group. The target group content topic sending module is used to send the target group content topic to the cloud disk terminal; The file topic content receiving module is used to receive file topic content from the cloud disk terminal. The third file recommendation target list generation module is used to generate a file recommendation target list based on the file topic content by executing a file topic content matching model. The file recommendation target list sending module is used to send the file recommendation target list to the cloud disk terminal.

[0025] Optionally, the module for determining the content topic of the sharing target group is used to: The last update time of the content topic of the target group being shared is sent by the cloud storage terminal; The effective update interval is calculated based on the last update time and the chat history context of the current user's shared group; If the effective update interval does not exceed the preset update threshold, the shared target group content topic cached in the cloud disk server will be determined as the shared target group content topic to be sent. If the effective update interval exceeds the preset update threshold, a new target group content topic is calculated based on the shared target group content topic cached in the cloud disk server and the newly added part in the chat history context, and the obtained new target group content topic is determined as the target group content topic to be sent.

[0026] Optionally, the shared file recommendation system, applied to a cloud disk server, further includes a recommendation validity threshold calculation module, used for: Based on the similarity quantification weight matching degree between the content topic of the target group to be shared and the content topic of the target group associated with each behavior in the user's historical behavior record, the effective threshold for recommending the user's historical behavior is calculated. The recommended effective threshold is sent to the cloud disk terminal.

[0027] Optionally, the file sharing recommendation system, applied to a cloud storage server, further includes a model update and training module for: Receive the selected recommendation results log sent from the cloud disk terminal; Based on the recommended logs, the shared content topic extraction model and the file topic content matching model are trained and updated.

[0028] To achieve the above objectives, the present invention also provides a file sharing recommendation device, comprising: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the file sharing recommendation method as described above.

[0029] To achieve the above objectives, the present invention also provides a computer-readable storage medium storing a computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the file sharing recommendation method described in any of the above embodiments.

[0030] To achieve the above objectives, the present invention also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the file sharing recommendation method as described in any of the above embodiments.

[0031] Compared to existing technologies, the present invention provides a file sharing recommendation method, system, device, medium, and program product. This method achieves accurate file recommendations by associating content themes with shared target groups, improving the relevance of recommended content to group needs. Furthermore, it adapts the recommended content detection ratio to the computing power level of the cloud disk terminal, ensuring recommendation accuracy when terminal capabilities allow, and guaranteeing recommendation response efficiency through time sorting when terminal capabilities are insufficient. This effectively balances recommendation accuracy and execution efficiency, reduces the operational cost for users to filter and share files, and improves the user experience and recommendation adoption rate in cloud disk sharing scenarios. Attached Figure Description

[0032] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a flowchart illustrating a file sharing recommendation method provided in an embodiment of the present invention; Figure 2 This is another flowchart illustrating a file sharing recommendation method provided in an embodiment of the present invention; Figure 3 This is another flowchart illustrating a file sharing recommendation method provided in an embodiment of the present invention; Figure 4 This is a structural block diagram of a file sharing recommendation system provided in an embodiment of the present invention; Figure 5 This is another structural block diagram of a file sharing recommendation system provided in an embodiment of the present invention; Figure 6 This is a structural block diagram of a file sharing recommendation device provided in an embodiment of the present invention. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] See Figure 1 , Figure 1 This is a flowchart illustrating a file sharing recommendation method provided in an embodiment of the present invention. The file sharing recommendation method is applied to a cloud storage terminal and includes steps S1 to S4: Step S101: Obtain the content topic of the target group to be shared sent by the cloud disk server; Step S102: Calculate the recommended content detection ratio based on the content topic of the target sharing group; Step S103: When the detection ratio of the recommended content is greater than the preset detection ratio threshold, a list of recommended file targets is generated based on the file topic content and through the file topic content matching model. Step S104: When the detection ratio of the recommended content is less than the preset detection ratio threshold, generate a list of recommended target files based on the file modification time.

[0036] This invention not only achieves accurate file recommendations by associating content themes with shared target groups, improving the relevance of recommended content to group needs, but also adapts the recommended content detection ratio to the computing power level of the cloud disk terminal. When the terminal capability allows, it ensures recommendation accuracy, and when the terminal capability is insufficient, it uses time sorting as a fallback to ensure recommendation response efficiency. This effectively balances recommendation accuracy and execution efficiency, reduces the operational cost for users to filter and share files, and improves the user experience and recommendation adoption rate in cloud disk sharing scenarios.

[0037] In an optional embodiment, step S101 includes steps S1011 to S1012: Step S1011: Send the last update time of the content topic of the target group being shared, which is cached on the cloud disk terminal, to the cloud disk server; Step S1012: Receive the content topic of the sharing target group returned by the cloud disk server based on the last update time.

[0038] For example, when a user prepares to share a file through a cloud storage terminal, the cloud storage terminal automatically sends the last update time (e.g., 2024-05-20 14:30) of the local cached content topic of the target group to the cloud storage server. After receiving the data, the cloud storage server can compare the time with the latest chat context in the group to determine if the original cached topic needs to be updated and return the updated topic. If there is no new relevant content in the group, the original cached topic will be returned directly.

[0039] By sending the last update time to the cloud drive server, the server can determine whether to update the topic based on that time. This ensures the timeliness and accuracy of the content topics shared with the target group, avoids redundant calculations on the terminal, and improves the response efficiency of topic acquisition.

[0040] In an optional embodiment, step S102 includes steps S1021 to S1022: Step S1021: Based on the content topic of the target group for sharing, calculate the predicted execution response time and the computing power support requirements that characterize the user's operational tolerance for the recommendation task. In one optional embodiment, the prediction execution response time is calculated based on the quantified value of the total file computing load time and the model processing time extension; wherein, the model processing time extension is the additional time consumed when the cloud disk terminal calls the file topic extraction model or the file content matching model due to the adaptation of the model size and the computing power of the cloud disk terminal; The computing power support requirements are determined based on the average latency of users' historical recommendations and the average response time of cloud disk servers.

[0041] For example, the cloud drive terminal obtains the content topic of the target group to share. Then, considering the file topic matching and filtering requirements corresponding to the content topic (such as the file type to be matched and the topic relevance requirements), as well as the terminal's own computing power, storage resources, and other hardware conditions, the estimated time required to perform the "file topic matching and filtering with the group topic" operation locally on the terminal is calculated, which is the local predicted execution response time. : ; in, This represents the total computational load time quantization value for the file, and its calculation logic is as follows: ; Where n represents the number of files without a theme in the cloud drive terminal. This represents the size of the i-th file. This represents the complexity of recognizing the i-th file. This represents the overall computing power of the cloud storage terminal, which is derived by aggregating the computing power of the CPU (Central Processing Unit), GPU (Graphics Processing Unit), and dedicated AI chips. This refers to cloud disk storage, which is derived by aggregating the CPU, GPU, dedicated AI chip cache, and memory.

[0042] in, This indicates an extension in model processing time, and its computational logic is as follows: ; in, Indicates the model size. This indicates the length of the model parameter vector.

[0043] Meanwhile, the computing power support requirements characterize the user's tolerance for ergonomic operation. The calculation logic is as follows: ; in, The penalty for response time stability fluctuations is determined by network bandwidth and stability. The adjustment coefficient is used to compensate for fluctuations in response time stability. Based on historical benchmarks, its calculation logic is as follows: ; in, This indicates the average latency that users accept in their historical recommendations. Recommended average response time for cloud disk servers. This is the normalization factor for delayed response.

[0044] The calculation logic for the attenuation coefficient of the device is as follows: ; in, This indicates user patience, inferred from user profiles; This represents the performance index of the user's terminal device. It is obtained by querying the basic information of the cloud disk terminal device and quantifying the performance parameters, based on the user's cloud disk terminal model obtained from the current cloud disk terminal.

[0045] It is worth noting that, by splitting the prediction execution response time (integrating the total computational load of the file with the model size and the additional time generated by the adaptation of terminal computing power) and the computing power support requirements (combining historical benchmarks such as historical recommendation latency and server response time, along with attenuation coefficients and response stability penalties related to user patience and device performance), the calculation of the recommended content detection ratio can be more in line with actual usage scenarios. This helps cloud disk terminals to rationally select recommendation strategies, ensuring recommendation accuracy while avoiding waste of terminal computing power or excessive user waiting time, effectively improving the adaptability and user experience of shared file recommendations.

[0046] Step S1022: Calculate the recommended content detection ratio based on the predicted execution response time and the computing power support requirements.

[0047] Specifically, based on predicted execution response time Computing power support requirements Calculate the recommended content detection ratio : ; in, This represents the expansion factor for the proportion of response time.

[0048] It is worth noting that the embodiments of the present invention correlate the predicted execution response time with the computing power support requirements of the user terminal operation, and scientifically quantify the degree of compatibility between the two. This not only avoids the lag problem caused by forcibly executing topic matching recommendations when the terminal computing power is insufficient, but also ensures the accuracy of recommendations when the terminal computing power is sufficient, effectively balancing the quality and execution efficiency of shared file recommendations.

[0049] In an optional embodiment, step S103 includes steps S1031 to S1034: Step S1031: Obtain the latest file topic content and cloud disk terminal resource information, and calculate the pre-evaluation efficiency of the file topic content matching model. In one alternative embodiment, the latest document topic content is obtained through the following steps: Get the cloud drive terminal file list and the corresponding historical file topic table; By comparing the file list on the cloud drive terminal with the historical file topic table, files without recorded topics are identified. For files without recorded topics, the file topics are extracted using a file topic extraction model to update the historical file topic table and obtain the latest file topic content.

[0050] For example, the preset detection ratio threshold can be set to 1. Therefore, when the terminal recommends content with a detection ratio of 1... When this happens, the folder's historical file topic table is read first. and get the list of files in the folder. By comparing the historical file subject table With file list Retrieve a list of empty subject files that have not undergone main file checks. Extracting hybrid models using terminal file themes Extract empty files and list of main files. The subject of the Chinese document Among them, for the list of empty files and topic files. The i-th empty file topic file The corresponding file topic The extraction method is as follows: ; Finally, the i-th empty file topic file and its corresponding file themes Combined topic items Add to the history document topic list From this, you can obtain the latest document content.

[0051] For example, in a scientific research sharing group, when a user prepares to share a file through a cloud drive terminal, the terminal first obtains the file list of the local folder and the corresponding historical file topic table. After comparison, it is found that the newly added "Sample Analysis Report.docx" does not have a recorded topic. The terminal then uses the file topic extraction model to extract the topic "Project X Sample Data Analysis" of the file and adds it to the historical file topic table to obtain the latest file topic content containing all file topics.

[0052] This invention, by identifying files with unrecorded topics and extracting topics, ensures the integrity and timeliness of the historical file topic table, providing data support for accurate matching and recommendation based on shared target group content topics. At the same time, it avoids repeatedly processing files with recorded topics, saving terminal computing power and processing time, and further improving recommendation efficiency and accuracy.

[0053] Furthermore, based on the updated latest historical document subject table Local resource information on cloud storage terminals Calculate the topic content matching model of the local executable file. Pre-evaluation efficiency The calculation logic is as follows: ; in, To improve the efficiency of computing power by increasing memory usage. This indicates retrieving the subject table of historical documents. The i-th term corresponding file Size, Represents a document topic content matching model The number of parameters multiplied by the number of quantization bits.

[0054] It is worth noting that by integrating the updated file topic table data, terminal local memory and computing resources, the pre-evaluation efficiency of the local execution topic matching model is quantified, providing a basis for subsequent judgment on "local execution or server execution model". This avoids the response delay caused by forced execution when the terminal computing power is insufficient, and makes full use of terminal resources to improve recommendation efficiency, further optimizing the adaptability and user experience of shared file recommendations.

[0055] Step S1032: Obtain the effective threshold for recommending users' historical behavior calculated by the cloud disk server; Step S1033: If the pre-evaluation efficiency is not greater than the recommendation validity threshold, then execute the file topic content matching model to generate a file recommendation target list; For example, if the recommended effective threshold obtained from the cloud disk server is Then when At that time, the cloud disk terminal first loads the full updated and latest historical file topic table. ,for correspond document By executing the document subject content matching model By reasoning, we can obtain... Content topics shared with the target group Matching degree : ; After deducing all the latest historical document subject lists Matching degree of all files in Then, sort them by matching degree from largest to smallest, and take the top ones. The corresponding files are used as the final file recommendation target list. Recommendations will be made.

[0056] Among them, the number of recommendations The quantity calculation method is as follows: ; in, Indicates obtaining , The maximum value in, Indicates user acceptance. Indicates the size of the visible recommendation display area. This indicates the size of the display area for a single recommended item.

[0057] Step S1034: If the pre-evaluation efficiency is greater than the recommended effective threshold, then send the file topic content to the cloud disk server to instruct the cloud disk server to execute the file topic content matching model and generate a file recommendation target list.

[0058] For example, when At that time, the cloud drive terminal will update the latest historical file topic list. The incremental portion is synchronized to the cloud disk server, where the cloud disk server executes the file topic content matching model. Inference is then performed. It should be noted that the inference process of the file topic content matching model and the generation of the file recommendation target list are explained. The process is the same as above, so I will not repeat it here. The only difference is the execution entity of the file subject content matching model.

[0059] However, the cloud storage server will recommend files to the target list. Returning to the cloud drive terminal, the cloud drive terminal receives the data and selects the recommended target list. Display recommended information.

[0060] In an optional embodiment, step S104 includes steps S1041 to S1042: Step S1041: Sort the files in the target folder in reverse order by modification time; Step S1042: Extract the top N files to form the file recommendation target list; where N is a positive integer calculated based on the user interface display area size and user acceptance.

[0061] For example, when the recommended content detection ratio is... When using the recommended method of sorting by modification time in reverse order, sort the files in the target folder in reverse order and extract the files from the top of the time list. One, as a recommended target list Recommendations are made, including the number of recommendations. The quantity calculation method is the same as before. Consistent, .

[0062] It should be noted that when the recommended content detection ratio is... If the timing of a recommendation is too slow, it indicates that the terminal's computing power is insufficient or the execution time for topic matching may exceed the user's tolerance. By adopting a recommendation method that sorts the files by modification time in reverse order, complex topic matching calculations can be quickly avoided, preventing terminal lag or slow response. At the same time, recently modified files are often more in line with the user's current sharing needs, balancing the timeliness of recommendations with execution efficiency. This simple and efficient fallback strategy reduces the user's screening cost and can ensure a basic user experience in sharing scenarios.

[0063] In one alternative embodiment, after generating the list of recommended target files, the method further includes: Output the generated list of recommended files to the user. In response to the user's selection of a file from the file recommendation target list and the execution of a file sharing operation, the system submits the recommendation result selection log to the cloud disk server.

[0064] For example, after generating a list of recommended files that fit the group's theme, the cloud storage terminal displays the list to the user. The user selects the file to be shared and performs the sharing operation. The cloud storage terminal then submits a recommendation result selection log containing information such as the selected file name and theme matching degree to the cloud storage server.

[0065] This invention provides real and effective training data for the cloud disk server to subsequently optimize the content topic extraction model and file topic content matching model by selecting the submitted recommendation results in the log, thereby continuously improving the recommendation accuracy and further enhancing the user experience in sharing scenarios.

[0066] See Figure 2 , Figure 2This is a flowchart illustrating a file sharing recommendation method provided in an embodiment of the present invention. The file sharing recommendation method is applied to a cloud disk server and includes steps S201 to S205: Step S201: Based on the chat history context of the user-shared group, determine the topic of the content to be shared with the target group; Step S202: Send the content topic of the target group to the cloud disk terminal; Step S203: Receive the file subject content from the cloud disk terminal; Step S204: Based on the file topic content, execute the file topic content matching model to generate a file recommendation target list; Step S205: Send the recommended target list of files to the cloud disk terminal.

[0067] This invention, through a cloud storage server, dynamically determines the content topic of the target group to be shared based on the chat history context of the user-shared group, and executes an efficient file topic content matching model to generate a recommendation list. This can significantly improve the accuracy and relevance of file recommendations in scenarios such as shared groups and circles. At the same time, the execution path of the recommendation strategy can be adaptively adjusted according to the computing power and network conditions of the cloud storage terminal, thereby ensuring the quality of recommendations while optimizing the use of computing resources and ensuring a smooth user experience.

[0068] In an optional embodiment, step S201 includes steps S2011 to S2014: Step S2011: Receive the last update time of the content topic of the target group being shared from the cloud disk terminal; Step S2012: Calculate the effective update interval based on the last update time and the chat history context of the current user's shared group; In one optional embodiment, step S2012 includes: Based on the last update time, determine the incremental content portion generated later than the last update time from the chat history context; The effective update interval is calculated based on the content density information of the incremental content section and the current user's speaking time span and number of speakings in the incremental content section.

[0069] For example, after a user enters the communication and sharing page of a shared group or circle, the cloud disk server first loads the locally cached content topics of the target group in the shared group or circle. and receiving data sent by the cloud disk terminal Last updated time Then based on the last update time With chat history context Calculate the effective update interval : ; in, The temporal density of contextual content information is calculated as follows: ; in, Indicates the context of the chat history The generation time is greater than the last update time. Part of Indicates the context of the chat history The Middle i The length of the content of the bar Indicates the current time.

[0070] The proportion of user context locations across time domains is represented by the following calculation logic: ; in, Indicates the context of the chat history Quantity, Indicates the context of the chat history The content currently posted by the user. Indicates the context of the chat history The shortest time, Indicates the context of the chat history The longest period of time.

[0071] It is worth noting that, by focusing on incremental content in chat records that is later than the last update time, and combining its content density with the current user's speaking time span and number of speaking in the incremental part to calculate the effective update interval, the core new information that may cause topic changes can be captured. This avoids the waste of resources caused by repeatedly calculating all historical chat records, and makes the calculation of the effective update interval more in line with the actual communication scenario, providing a basis for judging whether the subsequent topic needs to be updated. It can balance calculation efficiency and topic timeliness.

[0072] Step S2013: If the effective update interval does not exceed the preset update threshold, then the shared target group content topic cached in the cloud disk server is determined as the shared target group content topic to be sent. For example, when updating the valid interval Preset update threshold At that time, it directly returns the shared content topic of the target group in the shared group or circle cached in the cloud disk server. .

[0073] Step S2014: If the effective update interval exceeds the preset update threshold, then based on the shared target group content topic cached in the cloud disk server and the newly added part in the chat history context, a new shared target group content topic is calculated through the shared content topic extraction model, and the obtained new shared target group content topic is determined as the shared target group content topic to be sent.

[0074] For example, when updating the valid interval Preset update threshold At that time, the content topics of shared groups and circles cached in the cloud disk server are shared by the target groups. Chat history context The generation time is greater than the last update time. Part Using the input parameters, execute the topic extraction model for shared content. Obtain content topics for sharing with the target group. Updated topic vectors : ; in, ， This indicates the inference task of performing model M, with the following input parameters: .

[0075] Then use the updated topic vectors Replace the original topic of the content shared with the target group. and using the current time Replace last updated time Return the updated topic vector to the cloud disk server. As a shared group or circle, the target group for sharing content topics .

[0076] Therefore, the cloud disk server receives the last update time of the topic sent by the cloud disk terminal, calculates the effective update interval by combining the chat history context, and compares it with a preset threshold. It reuses cached topics as needed or updates topics based on new content. This can avoid the waste of resources caused by full calculation and improve the efficiency of topic return, thereby ensuring the timeliness and accuracy of the content topics shared with the target group.

[0077] In one optional embodiment, the file sharing recommendation method, applied to a cloud disk server, further includes: Based on the similarity quantification weight matching degree between the content topic of the target group to be shared and the content topic of the target group associated with each behavior in the user's historical behavior record, the effective threshold for recommending the user's historical behavior is calculated. The recommended effective threshold is sent to the cloud disk terminal.

[0078] For example, simultaneously based on the content theme of the target group being shared. Calculate user history behavior Recommended effective threshold The specific calculation logic is as follows: ; in, Represents the i-th user's historical behavior. The target group's content theme; Representing computational topics Content themes related to the target group The similarity quantization weight matching degree is obtained by the magnitude of the difference vector between the vectors.

[0079] This invention calculates the effective recommendation threshold based on the similarity weight matching degree between the content topic of the target group and the topic associated with the user's historical behavior, and sends it to the cloud disk terminal. This ensures that the recommendation standard takes into account both the common needs of the group and the individual behavioral preferences of the user, avoiding the one-sidedness of recommending based solely on the group topic. It also provides a quantitative basis for the terminal to subsequently screen the target list of recommended files, reducing the computing power consumption of the terminal to calculate the threshold independently. This can improve the overall efficiency of the recommendation process and make the recommendation results more in line with user expectations.

[0080] In one optional embodiment, the file sharing recommendation method, applied to a cloud disk server, further includes: Receive the selected recommendation results log sent from the cloud disk terminal; Based on the recommended logs, the shared content topic extraction model and the file topic content matching model are trained and updated.

[0081] For example, after a user selects a recommended list and completes the sharing operation, the cloud storage terminal sends a recommendation result selection log to the cloud storage server, containing information such as the selected file name, topic matching score, and user operation time. The cloud storage server aggregates similar logs submitted by the cloud storage terminal, using the correspondence between high-matching and frequently selected file topics and group content topics as positive samples, and the special cases of low-matching but selected topics as supplementary samples. Iterative training and parameter tuning are then performed on the sharing content topic extraction model and the file topic content matching model to further correct model parameter biases, improve the accuracy of group content topic extraction and the precision of file topic matching, thereby improving the user's search and sharing efficiency in file sharing scenarios.

[0082] In summary, see Figure 3 , Figure 3 This is another flowchart of a file sharing recommendation method provided in an embodiment of the present invention. For example... Figure 3As shown in the figure, the file sharing recommendation method provided by this embodiment of the invention includes the following interaction steps in the cloud disk terminal and cloud disk server: After a user enters the communication and sharing page of a shared group or circle, the cloud storage server first loads the chat history context of that shared group or circle. Execute the topic extraction model for shared content. Determine the target groups and content themes for sharing in shared groups and circles. And the data is fed back to the cloud disk terminal.

[0083] Cloud drive terminals receive content from target groups and topics. Afterwards, the cloud disk terminal computes content topics. Local prediction execution response time Simultaneously calculate the user terminal's gesture computing power requirements. Execution response time based on local prediction Support for user terminal operation and motion-sensing computing power requirements Calculate the detection ratio of recommended content on the terminal. .

[0084] When the terminal recommended content detection ratio When using the recommended method of sorting by modification time in reverse order, sort the files in the checked folder in reverse order and extract the files from the top of the time list. One, as the recommendation target Recommendations will be made.

[0085] When the terminal recommended content detection ratio At that time, the cloud disk terminal reads the local folder file history topic table. and get the list of files in the folder. By comparing the historical file subject table With file list Retrieve a list of empty subject files that have not undergone main file checks. Extracting hybrid models using terminal file themes Extract empty files and list of main files. The subject of the Chinese document Add to the history document topic list middle.

[0086] The cloud storage terminal is based on the content theme of the target group being shared. Updated latest historical document topic table Terminal local resource information Calculate the topic content matching model of the local executable file. Pre-evaluation efficiency Simultaneously, it obtains the effective threshold for recommendations based on users' historical behavior calculated by the cloud disk server. .

[0087] when At that time, the cloud disk terminal executes the file topic content matching model. Finally, a list of recommended target files is obtained. Recommendations will be made.

[0088] when At that time, the cloud drive terminal will update the latest historical file topic list. The incremental portion is synchronized to the cloud disk server, where the cloud disk server executes the file topic content matching model. Finally, a list of recommended target files is obtained. Recommendations will be made.

[0089] The user will receive the final list of recommended targets. After selecting a file, the cloud drive terminal executes the file sharing and upload process, and simultaneously submits the recommended result selection log to the cloud drive server. The cloud disk server periodically selects logs based on recommended results. As training and update data, the model extracts topics from shared content. Matching model with document subject content Perform a second training update.

[0090] In summary, the file sharing recommendation method provided by this invention first combines the group chat topic to obtain a recommendation strategy based on the content detection score. When the terminal's computing power or response time does not meet the complex matching requirements (detection ratio ≤ 1), a lightweight reverse modification time recommendation is adopted to avoid complex calculations and prevent terminal lag. When the terminal has the processing conditions (detection ratio > 1), the local file topic is completed first, and then the matching is promoted, balancing efficiency and basic user experience. Secondly, a comparison mechanism between pre-evaluation efficiency and recommendation effectiveness threshold is introduced to reasonably allocate the model execution subject. When the terminal's computing power is sufficient, the matching is performed locally; when the computing power is insufficient, the topic is synchronized to the server for execution. This reduces the computing power pressure on the terminal and utilizes the server's computing power to ensure matching accuracy, optimizing the collaboration efficiency between the cloud and the terminal. Finally, based on the user recommendation result selection log, the topic extraction and matching model is iteratively trained to continuously improve the recommendation accuracy.

[0091] See Figure 4 , Figure 4 This is a structural block diagram of a file sharing recommendation system provided in an embodiment of the present invention. The file sharing recommendation system is applied to a cloud storage terminal and includes: The target group content topic acquisition module 11 is used to acquire the target group content topic sent by the cloud disk server. The recommended content detection ratio calculation module 12 is used to calculate the recommended content detection ratio based on the content theme of the sharing target group; The first file recommendation target list generation module 13 is used to generate a file recommendation target list based on the file topic content and through a file topic content matching model when the detection ratio of the recommended content is greater than a preset detection ratio threshold. The second file recommendation target list generation module 14 is used to generate a file recommendation target list based on the file modification time when the detection ratio of the recommended content is less than a preset detection ratio threshold.

[0092] In an optional embodiment, the recommended content detection ratio calculation module 12 is used to: Based on the content theme of the target group to be shared, calculate the predicted execution response time and the computing power support requirements that characterize the user's operational tolerance for the recommendation task. The recommended content detection ratio is calculated based on the predicted execution response time and the computing power support requirements.

[0093] In an optional embodiment, the first file recommendation target list generation module 13 is used to: Obtain the latest file topic content and cloud disk terminal resource information, and calculate the pre-evaluation efficiency of the file topic content matching model. Obtain the effective threshold for recommendations based on users' historical behavior calculated by the cloud disk server; If the pre-evaluation efficiency is not greater than the effective recommendation threshold, then the file topic content matching model is executed to generate a list of file recommendation targets; If the pre-evaluation efficiency is greater than the recommended effective threshold, then the file topic content is sent to the cloud disk server to instruct the cloud disk server to execute the file topic content matching model and generate a list of recommended file targets.

[0094] In an optional embodiment, the second file recommendation target list generation module 14 is used to: Sort the files in the target folder in reverse order of their modification time; Extract the top N files to form the file recommendation target list; where N is a positive integer calculated based on the user interface display area size and user acceptance.

[0095] In one optional embodiment, the content topic acquisition module 11 for sharing target groups is configured to: Send the last update time of the content topic of the target group to the cloud disk server, which is cached on the cloud disk terminal; The cloud disk server returns the content topic for the target group to share based on the last update time.

[0096] In one optional embodiment, the shared file recommendation system, applied to a cloud storage terminal, further includes a file recommendation target list output module, used for: Output the generated list of recommended files to the user. In response to the user's selection of a file from the file recommendation target list and the execution of a file sharing operation, the system submits the recommendation result selection log to the cloud disk server.

[0097] It should be noted that the file sharing recommendation system provided in this embodiment of the invention is used to execute all the process steps of the file sharing recommendation method for cloud disk terminals described above. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0098] See Figure 5 , Figure 5 This is another structural block diagram of a file sharing recommendation system provided in an embodiment of the present invention. The file sharing recommendation system is applied to a cloud disk server and includes: The target group content topic determination module 21 is used to determine the target group content topic based on the chat history context of the user's shared group. The target group content topic sending module 22 is used to send the target group content topic to the cloud disk terminal; The file subject content receiving module 23 is used to receive file subject content from the cloud disk terminal. The third file recommendation target list generation module is used to generate a file recommendation target list based on the file topic content by executing a file topic content matching model. The file recommendation target list sending module 24 is used to send the file recommendation target list to the cloud disk terminal.

[0099] In one optional embodiment, the content topic determination module 21 for sharing target groups is configured to: The last update time of the content topic of the target group being shared is sent by the cloud storage terminal; The effective update interval is calculated based on the last update time and the chat history context of the current user's shared group; If the effective update interval does not exceed the preset update threshold, the shared target group content topic cached in the cloud disk server will be determined as the shared target group content topic to be sent. If the effective update interval exceeds the preset update threshold, a new target group content topic is calculated based on the shared target group content topic cached in the cloud disk server and the newly added part in the chat history context, and the obtained new target group content topic is determined as the target group content topic to be sent.

[0100] In one optional embodiment, the shared file recommendation system, applied to a cloud disk server, further includes a recommendation validity threshold calculation module, used for: Based on the similarity quantification weight matching degree between the content topic of the target group to be shared and the content topic of the target group associated with each behavior in the user's historical behavior record, the effective threshold for recommending the user's historical behavior is calculated. The recommended effective threshold is sent to the cloud disk terminal.

[0101] In one optional embodiment, the file sharing recommendation system, applied to a cloud storage server, further includes a model update and training module for: Receive the selected recommendation results log sent from the cloud disk terminal; Based on the recommended logs, the shared content topic extraction model and the file topic content matching model are trained and updated.

[0102] It should be noted that the file sharing recommendation system provided in this embodiment of the invention is used to execute all the process steps of the file sharing recommendation method for cloud disk servers described in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0103] See Figure 6 , Figure 6 This is a structural block diagram of a file-sharing recommendation device provided in an embodiment of the present invention. The file-sharing recommendation device includes a processor 31, a memory 32, and a computer program stored in the memory 32 and executable on the processor 31. When the processor 31 executes the computer program, it implements the steps in the various file-sharing recommendation method embodiments described above.

[0104] For example, the computer program may be divided into one or more modules / units, which are stored in the memory 32 and executed by the processor 31 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the file-sharing recommendation device.

[0105] The file sharing recommendation device may include, but is not limited to, a processor 31 and a memory 32. Those skilled in the art will understand that the schematic diagram is merely an example of a file sharing recommendation device and does not constitute a limitation on the device. It may include more or fewer components than illustrated, or combine certain components, or use different components. For example, the file sharing recommendation device may also include input / output devices, network access devices, buses, etc.

[0106] The processor 31 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor 31 is the control center of the file-sharing recommendation device, connecting various parts of the entire file-sharing recommendation device through various interfaces and lines.

[0107] The memory 32 can be used to store the computer programs and / or modules. The processor 31 implements various functions of the file sharing recommendation device by running or executing the computer programs and / or modules stored in the memory 32 and calling the data stored in the memory 32. The memory 32 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory 32 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0108] Wherein, if the modules / units integrated into the file-sharing recommendation device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor 31, it can implement the steps of the various method embodiments described above. Wherein, the computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0109] Furthermore, embodiments of the present invention also provide a computer program product, including a computer program / instructions, which, when executed by a processor, implement the file sharing recommendation method as described in any of the above embodiments.

[0110] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for recommending shared files, characterized in that, Applied to cloud disk terminals, including: Retrieve the content topic sent by the cloud disk server to the target group for sharing; Based on the content theme of the target group being shared, the recommended content detection ratio is calculated; When the detection ratio of the recommended content is greater than the preset detection ratio threshold, a list of recommended file targets is generated based on the file topic content and through the file topic content matching model. When the detection ratio of the recommended content is less than the preset detection ratio threshold, a list of recommended target files is generated based on the file modification time.

2. The file sharing recommendation method as described in claim 1, characterized in that, The step of calculating the recommended content detection ratio based on the content topic of the target sharing group includes: Based on the content theme of the target group to be shared, calculate the predicted execution response time and the computing power support requirements that characterize the user's operational tolerance for the recommendation task. The recommended content detection ratio is calculated based on the predicted execution response time and the computing power support requirements.

3. The file sharing recommendation method as described in claim 2, characterized in that, The prediction execution response time is calculated based on the quantified value of the total file computing load time and the model processing time extension; wherein, the model processing time extension is the additional time consumed when the cloud disk terminal calls the file topic extraction model or the file content matching model due to the adaptation of the model size and the computing power of the cloud disk terminal. The computing power support requirements are determined based on the average latency of users' historical recommendations and the average response time of cloud disk servers.

4. The file sharing recommendation method as described in claim 1, characterized in that, The process of generating a list of recommended files based on file topic content using a file topic content matching model includes: Obtain the latest file topic content and cloud disk terminal resource information, and calculate the pre-evaluation efficiency of the file topic content matching model. Obtain the effective threshold for recommendations based on users' historical behavior calculated by the cloud disk server; If the pre-evaluation efficiency is not greater than the effective recommendation threshold, then the file topic content matching model is executed to generate a list of file recommendation targets; If the pre-evaluation efficiency is greater than the recommended effective threshold, then the file topic content is sent to the cloud disk server to instruct the cloud disk server to execute the file topic content matching model and generate a list of recommended file targets.

5. The file sharing recommendation method as described in claim 4, characterized in that, The latest document topic content is obtained through the following steps: Get the cloud drive terminal file list and the corresponding historical file topic table; By comparing the file list on the cloud drive terminal with the historical file topic table, files without recorded topics are identified. For files without recorded topics, the file topics are extracted using a file topic extraction model to update the historical file topic table and obtain the latest file topic content.

6. The file sharing recommendation method as described in claim 1, characterized in that, The file recommendation target list generated based on file modification time includes: Sort the files in the target folder in reverse order of their modification time; Extract the top N files to form the file recommendation target list; where N is a positive integer calculated based on the user interface display area size and user acceptance.

7. The file sharing recommendation method as described in claim 1, characterized in that, The topic of the shared target group content sent by the cloud disk server includes: Send the last update time of the content topic of the target group to the cloud disk server, which is cached on the cloud disk terminal; The cloud disk server returns the content topic for the target group to share based on the last update time.

8. The file sharing recommendation method as described in claim 1, characterized in that, After generating the list of recommended targets for the file, it also includes: Output the generated list of recommended files to the user. In response to the user's selection of a file from the file recommendation target list and the execution of a file sharing operation, the system submits the recommendation result selection log to the cloud disk server.

9. A method for recommending shared files, characterized in that, Applied to cloud disk servers, including: Based on the context of the chat history shared by users in the group, determine the topic of the content to be shared with the target group; Send the content topic of the target group to the cloud drive terminal; Receive file content from the cloud drive terminal; Based on the file topic content, execute the file topic content matching model to generate a list of recommended file targets; Send the recommended target list of the file to the cloud drive terminal.

10. The file sharing recommendation method as described in claim 9, characterized in that, The process of determining the topic of content to be shared with the target group based on the chat history context of the user-shared group includes: The last update time of the content topic of the target group being shared is sent by the cloud storage terminal; The effective update interval is calculated based on the last update time and the chat history context of the current user's shared group; If the effective update interval does not exceed the preset update threshold, the shared target group content topic cached in the cloud disk server will be determined as the shared target group content topic to be sent. If the effective update interval exceeds the preset update threshold, a new target group content topic is calculated based on the shared target group content topic cached in the cloud disk server and the newly added part in the chat history context, and the obtained new target group content topic is determined as the target group content topic to be sent.

11. The file sharing recommendation method as described in claim 10, characterized in that, The calculation of the effective update interval based on the last update time and the chat history context of the current user's shared group includes: Based on the last update time, determine the incremental content portion generated later than the last update time from the chat history context; The effective update interval is calculated based on the content density information of the incremental content section and the current user's speaking time span and number of speakings in the incremental content section.

12. The file sharing recommendation method as described in claim 9, characterized in that, Also includes: Based on the similarity quantification weight matching degree between the content topic of the target group to be shared and the content topic of the target group associated with each behavior in the user's historical behavior record, the effective threshold for recommending the user's historical behavior is calculated. The recommended effective threshold is sent to the cloud disk terminal.

13. The file sharing recommendation method as described in claim 10, characterized in that, Also includes: Receive the selected recommendation results log sent from the cloud disk terminal; Based on the recommended logs, the shared content topic extraction model and the file topic content matching model are trained and updated.

14. A file sharing recommendation system, characterized in that, Applications to cloud storage terminals include: The module for obtaining content topics for sharing with the target group is used to obtain the content topics for sharing with the target group sent by the cloud disk server. The recommended content detection ratio calculation module is used to calculate the recommended content detection ratio based on the content theme of the target sharing group; The first file recommendation target list generation module is used to generate a file recommendation target list based on the file topic content and through a file topic content matching model when the detection ratio of the recommended content is greater than a preset detection ratio threshold. The second file recommendation target list generation module is used to generate a file recommendation target list based on the file modification time when the detection ratio of the recommended content is less than a preset detection ratio threshold.

15. The file sharing recommendation system as described in claim 14, characterized in that, The recommended content detection ratio calculation module is used for: Based on the content theme of the target group to be shared, calculate the predicted execution response time and the computing power support requirements that characterize the user's operational tolerance for the recommendation task. The recommended content detection ratio is calculated based on the predicted execution response time and the computing power support requirements.

16. The file sharing recommendation system as described in claim 14, characterized in that, The first document recommends a target list generation module, which is used for: Obtain the latest file topic content and cloud disk terminal resource information, and calculate the pre-evaluation efficiency of the file topic content matching model. Obtain the effective threshold for recommendations based on users' historical behavior calculated by the cloud disk server; If the pre-evaluation efficiency is not greater than the effective recommendation threshold, then the file topic content matching model is executed to generate a list of file recommendation targets; If the pre-evaluation efficiency is greater than the recommended effective threshold, then the file topic content is sent to the cloud disk server to instruct the cloud disk server to execute the file topic content matching model and generate a list of recommended file targets.

17. The file sharing recommendation system as described in claim 14, characterized in that, The second document recommends a target list generation module, which is used for: Sort the files in the target folder in reverse order of their modification time; Extract the top N files to form the file recommendation target list; where N is a positive integer calculated based on the user interface display area size and user acceptance.

18. The file sharing recommendation system as described in claim 14, characterized in that, The module for obtaining content topics from the target group is used for: Send the last update time of the content topic of the target group to the cloud disk server, which is cached on the cloud disk terminal; The cloud disk server returns the content topic for the target group to share based on the last update time.

19. The file sharing recommendation system as described in claim 14, characterized in that, It also includes a file recommendation target list output module, used for: Output the generated list of recommended files to the user. In response to the user's selection of a file from the file recommendation target list and the execution of a file sharing operation, the system submits the recommendation result selection log to the cloud disk server.

20. A file-sharing recommendation system, characterized in that, Applied to cloud disk servers, including: The module for determining the topic of content to be shared with the target group is used to determine the topic of content to be shared with the target group based on the context of the chat history of the user sharing the group. The target group content topic sending module is used to send the target group content topic to the cloud disk terminal; The file topic content receiving module is used to receive file topic content from the cloud disk terminal. The third file recommendation target list generation module is used to generate a file recommendation target list based on the file topic content by executing a file topic content matching model. The file recommendation target list sending module is used to send the file recommendation target list to the cloud disk terminal.

21. The file sharing recommendation system as described in claim 20, characterized in that, The module for determining the content topic of the sharing target group is used for: The last update time of the content topic of the target group being shared is sent by the cloud storage terminal; The effective update interval is calculated based on the last update time and the chat history context of the current user's shared group; If the effective update interval does not exceed the preset update threshold, the shared target group content topic cached in the cloud disk server will be determined as the shared target group content topic to be sent. If the effective update interval exceeds the preset update threshold, a new target group content topic is calculated based on the shared target group content topic cached in the cloud disk server and the newly added part in the chat history context, and the obtained new target group content topic is determined as the target group content topic to be sent.

22. The file sharing recommendation system as described in claim 20, characterized in that, It also includes a recommended effective threshold calculation module, used for: Based on the similarity quantification weight matching degree between the content topic of the target group to be shared and the content topic of the target group associated with each behavior in the user's historical behavior record, the effective threshold for recommending the user's historical behavior is calculated. The recommended effective threshold is sent to the cloud disk terminal.

23. The file sharing recommendation system as described in claim 21, characterized in that, It also includes a model update training module, used for: Receive the selected recommendation results log sent from the cloud disk terminal; Based on the recommended logs, the shared content topic extraction model and the file topic content matching model are trained and updated.

24. A file sharing recommendation device, characterized in that, include: A processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the file sharing recommendation method as described in any one of claims 1 to 8, or the file sharing recommendation method as described in any one of claims 9 to 13.

25. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the file sharing recommendation method as described in any one of claims 1 to 8, or the file sharing recommendation method as described in any one of claims 9 to 13.

26. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the file sharing recommendation method as described in any one of claims 1 to 8, or the file sharing recommendation method as described in any one of claims 9 to 13.

Images