Platform management method, terminal and storage medium

Abstract

This application relates to the field of platform management technology and specifically discloses a platform management method, terminal, and storage medium. The method first parses platform setting information to break it down into various platform setting contents. Deep learning-based artificial intelligence technology is then introduced to perform semantic parsing and fine-grained semantic query interaction on the various platform setting contents and user-entered preference descriptions. The method dynamically modulates the various platform setting contents based on user preference information, thereby intelligently generating platform page identification information that meets the user's personalized needs. This application enables personalized customization of platform pages to meet the specific needs of different users.

Description

Technical Field

[0001] The present application relates to the technical field of platform management, and more specifically, to a platform management method, terminal, and storage medium. Background Art

[0002] In today's rapidly developing digital age, various platforms play a crucial role in the daily operations and lives of businesses and users. As businesses continue to expand and their operations become increasingly complex, the number of information systems supporting their operations continues to increase. Numerous applications, such as financial management, marketing management, and production management, have been built and put into use, significantly improving business efficiency.

[0003] To provide a better user experience, platforms need to be able to adjust their platform interface and service content based on the personalized needs of users. Generally speaking, the identification information settings of platform pages mainly rely on predefined templates and limited options manually selected by users, such as the title of the platform homepage, logo image information, and the color tones and fonts of different pages. In actual application scenarios, different users often have unique preferences for the display style and functional layout of platform pages. For example, some users may prefer a simple and clear page design, while others may require a richer and more detailed information display.

[0004] However, existing technologies simply rely on platform settings for identification, lacking the ability to deeply explore and accurately respond to users' personalized needs. Furthermore, given that user preferences may change over time, a good platform should allow for dynamic adjustment of these settings and be able to securely and reliably save the latest configuration so that it takes effect immediately upon the next visit.

[0005] Therefore, an optimized platform management method, terminal and storage medium are needed. Summary of the Invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide a platform management method, terminal and storage medium, which first parses the platform setting information to break it down into various platform setting contents, and then further introduces artificial intelligence technology based on deep learning to perform semantic parsing and fine-grained semantic query interaction on the various platform setting contents and the preference descriptions entered by the user, so as to dynamically modulate the various platform setting contents according to the user preference information, thereby intelligently generating platform page identification information that meets the user's personalized needs, which can realize personalized customization of the platform page to meet the specific needs of different users.

[0007] Accordingly, according to one aspect of the present application, a platform management method is provided, which includes: obtaining platform setting information, and determining identification information of a platform page based on the platform setting information, including: parsing the platform setting information to obtain a set of platform setting content; semantically encoding each platform setting content in the set of platform setting content to obtain a set of platform setting content semantic coding vectors; obtaining a text description of user preferences; semantically encoding the text description of user preferences to obtain a user preference semantic coding vector; performing fine-grained cross-domain association learning on the set of platform setting content semantic coding vectors and the user preference semantic coding vector to obtain a set of user preference modulated platform setting content semantic coding vectors; and generating identification information of the platform page based on the set of user preference modulated platform setting content semantic coding vectors.

[0008] Preferably, semantically encoding the text description of the user preference to obtain the user preference semantic encoding vector includes: using a semantic encoder based on the Bert model to semantically encode the text description of the user preference to obtain the user preference semantic encoding vector.

[0009] Preferably, fine-grained cross-domain association learning is performed on the set of platform setting content semantic coding vectors and the user preference semantic coding vectors to obtain a set of user preference modulated platform setting content semantic coding vectors, including: performing association strength measurement on each platform setting content semantic coding vector in the set of user preference semantic coding vectors and the platform setting content semantic coding vectors to obtain a set of user preference-platform setting content fine-grained ablation factors; based on the set of user preference-platform setting content fine-grained ablation factors, fine-grained ablation modulation is performed on the set of platform setting content semantic coding vectors to obtain the set of user preference modulated platform setting content semantic coding vectors.

[0010] Preferably, the association strength measurement is performed on each platform setting content semantic coding vector in the set of the user preference semantic coding vector and the platform setting content semantic coding vector to obtain a set of user preference-platform setting content fine-grained ablation factors, including: performing cross-domain query interaction based on the attention mechanism on each platform setting content semantic coding vector in the set of the user preference semantic coding vector and the platform setting content semantic coding vector to obtain a set of user preference-platform setting content cross-domain query interaction feature vectors; and inputting each user preference-platform setting content cross-domain query interaction feature vector in the set of user preference-platform setting content cross-domain query interaction feature vectors into an ablation measurement function to obtain the set of user preference-platform setting content fine-grained ablation factors.

[0011] Preferably, a cross-domain query interaction based on an attention mechanism is performed on each platform setting content semantic coding vector in the set of the user preference semantic coding vector and the platform setting content semantic coding vector to obtain a set of user preference-platform setting content cross-domain query interaction feature vectors, including: performing a linear transformation on the user preference semantic coding vector to obtain a query vector and a value vector; performing a linear transformation on the platform setting content semantic coding vector to obtain a key vector; and inputting the query vector, the value vector and the key vector into a cross-domain interaction encoder based on an imitation converter structure to obtain the user preference-platform setting content cross-domain query interaction feature vector.

[0012] Preferably, based on the set of user preference-platform setting content fine-grained ablation factors, the set of platform setting content semantic coding vectors is subjected to fine-grained ablation modulation to obtain the set of user preference modulated platform setting content semantic coding vectors, including: inputting the set of user preference-platform setting content fine-grained ablation factors into an ablation effect coding module containing a normalization function and a masking function to obtain a set of user preference-platform setting content fine-grained ablation weight factors; based on the set of user preference-platform setting content fine-grained ablation weight factors, weighted modulation is performed on the set of platform setting content semantic coding vectors to obtain the set of user preference modulated platform setting content semantic coding vectors.

[0013] According to another aspect of the present application, a terminal device is provided, comprising: a memory for storing instructions; a processor coupled to the memory, the processor being configured to execute the platform management method described above based on the instructions stored in the memory.

[0014] According to another aspect of the present application, a storage medium is provided, on which a platform management program is stored. When the platform management program is executed by a processor, the platform management method as described above is implemented.

[0015] This application has at least the following technical effects:

[0016] Compared with the existing technology, the platform management method, terminal, and storage medium provided by this application first parse the platform setting information to break it down into various platform setting contents. Then, it further introduces deep learning-based artificial intelligence technology to perform semantic parsing and fine-grained semantic query interaction on each platform setting content and the preference description entered by the user. In this way, the platform setting contents are dynamically modulated according to the user's preference information, thereby intelligently generating platform page identification information that meets the user's personalized needs. This application can realize the personalized customization of platform pages to meet the specific needs of different users. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0018] Figure 1 Flowchart of a platform management method according to an embodiment of the present application.

[0019] Figure 2 Schematic diagram of data flow of the platform management method according to an embodiment of the present application.

[0020] Figure 3 Flowchart of step S150 in the platform management method according to an embodiment of the present application.

[0021] Figure 4 is a block diagram of a terminal device according to an embodiment of the present application.

[0022] Reference numerals:

[0023] 10. Terminal device; 11. Processor; 12. Memory; 13. Input device; 14. Output device. DETAILED DESCRIPTION

[0024] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0025] It should be noted that the acquisition and processing of all information or data in this application are carried out in compliance with the relevant national data protection laws and policies and with the authorization given by the owner of the corresponding device.

[0026] As mentioned in the background technology above, in the existing technology, identification is simply determined based on platform setting information, which lacks the ability to deeply explore and accurately respond to users' personalized needs. In addition, considering that user preferences may change over time, a good platform should allow dynamic adjustment of these settings and be able to safely and reliably save the latest configuration so that it will take effect immediately the next time you visit. In response to this technical problem, this application proposes a platform management method, specifically, Figure 1 Flowchart of a platform management method according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of the platform management method according to the embodiment of the present application. Figure 1 and Figure 2As shown, the platform management method according to the embodiment of the present application includes the steps of: S110, parsing the platform setting information to obtain a set of platform setting contents; S120, semantically encoding each platform setting content in the set of platform setting contents to obtain a set of platform setting content semantic encoding vectors; S130, obtaining a text description of user preferences; S140, semantically encoding the text description of user preferences to obtain a user preference semantic encoding vector; S150, performing fine-grained cross-domain association learning on the set of platform setting content semantic encoding vectors and the user preference semantic encoding vector to obtain a set of user preference modulated platform setting content semantic encoding vectors; S160, generating identification information of the platform page based on the set of user preference modulated platform setting content semantic encoding vectors.

[0027] Specifically, platform settings information refers to the data set used to define the platform's appearance (e.g., color scheme, font style), functional layout (e.g., menu location, toolbar visibility), and branding elements (e.g., logo, slogan, etc.). In practice, this information is typically entered by administrators or users during initial installation or use, or can be imported from other systems via APIs.

[0028] In practice, to accurately adjust the platform page's identification information based on user preferences, it's necessary to first comprehensively and deeply acquire and parse the platform's existing settings. This step is crucial because it provides the foundational data for subsequent semantic encoding and association learning.

[0029] Specifically, you first need to access the data source that stores the platform settings information. For many modern web applications and services, this information is usually stored in one or more databases, which may be in SQL or NoSQL format, depending on the specific architecture of the platform. For example, an e-commerce website may store its product categories, page templates, color themes, and other information in a relational database, while behavioral data such as user clickstreams and dwell time may use a non-relational database to efficiently handle large-scale concurrent write operations. Therefore, obtaining platform settings information requires establishing a stable and reliable connection to these data sources and ensuring that the required data can be retrieved safely and quickly.

[0030] Next, different methods are needed to extract and organize different types of platform setting information. For static and structured information, such as page titles, logo links, etc., they can be directly extracted from the database through query statements (SQL queries or other applicable query languages). This type of information often has a clear correspondence between field names and values, making it easy to parse. However, for more complex information, such as page layout rules, animation effect parameters, etc., it may be nested in configuration files in JSON or XML format, or even exist in the form of code in the front-end JavaScript framework. At this point, you need to use a parsing library (such as the json module or lxml library in Python) to decode these unstructured data and extract the key elements.

[0031] In addition, with the development of technology, more and more platforms have begun to support the provision of services through API interfaces, which has also become an important way to obtain platform setting information. The API interface allows third-party developers or internal systems to request specific resources programmatically, thereby simplifying the data exchange process across systems. In the scenario of this application, the latest version of platform setting information can be obtained by calling the relevant API, including but not limited to the latest UI component library version number, recommended style guide, etc. This not only ensures the freshness of the data, but also helps to quickly respond to platform changes.

[0032] In addition to the above work, we also need to consider the possibility that platform settings information will change over time. The platform's administrators or operations team may regularly update page elements, adjust functional layouts, or introduce new features to adapt to market trends and user needs. This requires not only the ability to capture the current platform settings information once, but also the ability to continuously monitor and promptly detect and record any changes. One feasible approach is to establish an event-driven mechanism. Whenever a major change occurs on the platform, the corresponding callback function is triggered to re-evaluate the latest settings status. In addition, periodic inspection tasks can be set to regularly synchronize the latest platform settings information to ensure that the system is always up to date.

[0033] Finally, the process of obtaining platform configuration information inevitably raises privacy and security issues. This is especially true when involving user personal data or sensitive business information, which must comply with strict laws and regulations, such as the GDPR (General Data Protection Regulation). Therefore, the principle of minimization should be prioritized when designing the data collection process, collecting only the minimum amount of information necessary to complete the task. Measures such as encrypted transmission and access control should be implemented to ensure data security. Furthermore, users should be clearly informed of the data usage and their consent should be obtained to maintain transparency and trust.

[0034] In the above-mentioned platform management method, the step S110 parses the platform setting information to obtain a collection of platform setting contents. Specifically, considering that the platform setting information contains a variety of different setting items, in order to more accurately personalize each platform setting element, the present application further parses the platform setting information to identify and distinguish different setting contents to form a collection of platform setting contents. For example, the color scheme can be further subdivided into background color, text color, button color, etc.; the functional layout can be subdivided into menu bar position, button size, icon style, etc.; brand elements can be subdivided into logo design, slogan font, etc.

[0035] During the specific implementation process, in order to convert the original and possibly disorganized platform setting information into a series of structured, easy-to-understand and process elements, it is required not only to accurately identify the different setting items, but also to be able to distinguish and classify these setting items to form an organized content collection, so that each setting element can be personalized more accurately in subsequent steps.

[0036] Before parsing platform settings information, data preprocessing is required. First, ensure the completeness and accuracy of the acquired platform settings information. This may involve interacting with multiple data sources, such as extracting records from relevant database tables or reading specific fields from configuration files. Furthermore, the acquired data must be formatted and cleaned to remove any invalid, duplicate, or malformed data.

[0037] For example, for settings stored in a database, some fields may contain empty values ​​or may not conform to the preset format due to untimely data updates or data entry errors. In this case, these abnormal values ​​need to be processed through data cleaning algorithms, such as filling empty values ​​with default values ​​or making reasonable repairs based on the context of the data and other relevant data. For data read from configuration files, there may be format inconsistencies, such as some color values ​​using hexadecimal and others using RGB format. They need to be unified into a standard format for subsequent parsing and processing.

[0038] Furthermore, platform settings information needs to be categorized and labeled to better identify different types of settings. This can be achieved by establishing a metadata dictionary, categorizing common settings into categories such as visual, layout, interaction, and branding, and assigning each setting a unique identifier and description. This allows for quick and accurate identification of the category of each data segment during parsing, improving parsing efficiency and accuracy.

[0039] A common method for parsing platform settings information is rule-based parsing, which analyzes and decomposes platform settings information one by one according to a set of predefined parsing rules to obtain a collection of platform settings content.

[0040] For the parsing of color schemes, the rules can be defined as: find all color-related fields, and assign color values ​​to corresponding subcategories based on the keywords in the field names (such as "background", "text", "button", etc.). For example, if there is a field called "pageBackgroundColor" and its value is "#FFFFFF", it can be identified as the background color, and "#FFFFFF" can be recorded as the background color value in the corresponding position in the platform settings content collection. For the parsing of functional layouts, the rules can be set as: based on specific layout element keywords (such as "menu bar", "button", "icon", etc.), extract its position, size, style and other related attributes. For example, for the "menuBarPosition" field, if its value is "top", the menu bar position is recorded as the top.

[0041] When parsing brand elements, for information related to logo design, the rule might be to search for fields containing the keyword "logo" and extract information such as dimensions, graphic format, and color composition. For example, the "logoWidth" and "logoHeight" fields record the width and height of the logo, respectively, while the "logoColorPalette" field stores the color palette used by the logo. This information is combined to form a complete logo design description and added to the platform settings content collection.

[0042] Rule-based parsing methods are simple, direct, and easy to implement, making them particularly suitable for platform settings information with a relatively fixed structure and standardized format. However, they also have significant drawbacks. When the structure of platform settings information changes or when it doesn't conform to the pre-set rules, the accuracy and completeness of the parsing may be affected, necessitating the constant updating and maintenance of the parsing rules.

[0043] To overcome the limitations of rule-based parsing methods, machine learning can be introduced to parse platform settings information. Machine learning models can automatically identify and extract different settings by learning from a large number of labeled platform settings information samples, thus improving adaptability and generalization.

[0044] First, a suitable machine learning model, such as a decision tree or deep learning model, needs to be built. The input platform settings information needs to be converted into a feature form suitable for the model. This can be achieved through feature engineering. For example, color values ​​can be converted into RGB component values ​​as features. Textual descriptions of settings (such as function layout descriptions) can be converted into numerical features using methods such as bag-of-words models and TF-IDF. During the training phase, the model is trained using a dataset of annotated platform settings information. The model learns the characteristic patterns and regularities of different settings. For example, by learning from a large number of examples, the model can automatically identify different color scheme settings and the feature representations of individual elements in the function layout. During the actual parsing process, the platform settings information to be parsed is input into the trained model, and the model outputs the corresponding parsing results, namely, a collection of platform settings content. The advantage of machine learning methods is that they can automatically adapt to the changes and complexity of platform settings information, and they are more effective in parsing irregular and diverse settings information.

[0045] When parsing platform settings information, optimization strategies and error handling mechanisms can be employed to improve efficiency and accuracy. One optimization strategy involves caching the results of previously parsed platform settings information fragments or common settings patterns. The next time the same or similar information is encountered, the parsing results are retrieved directly from the cache, avoiding repeated computations. Furthermore, parallel computing technology can be used to parallelize the parsing of different types of settings information (such as visual elements and functional layouts), fully leveraging the performance of multi-core processors to increase parsing speed.

[0046] Error handling mechanisms are also crucial. Because platform setting information may contain various uncertainties and errors, such as missing data, format errors, and incomplete setting descriptions, these errors need to be detected and handled promptly during the parsing process. When encountering missing data, reasonable inferences and fill-in can be made based on the platform's default settings or the user's historical setting habits. For format errors, attempt to convert and repair the format. If repair is not possible, record the error information and notify relevant personnel for manual intervention. For incomplete setting descriptions, interaction with the user (such as popping up a prompt box to ask the user for specific information about the relevant settings) or reference the platform's general setting specifications to supplement and improve.

[0047] Finally, the platform settings content obtained through the above parsing process needs to be verified and consolidated to ensure its accuracy and completeness. This verification process can be carried out in various ways, such as by comparing the parsed settings with the actual platform display to verify whether the parsed settings accurately reflect the actual presentation of the platform page; or by comparing them with a known sample of correct parsed results, and evaluating the quality of the parsed results by calculating metrics such as precision and recall. Regarding data consolidation, the various parsed platform settings content needs to be rationally organized and structured to form a complete and organized platform settings content collection. For example, settings for various aspects such as visual elements, functional layout, and branding elements should be arranged according to a certain hierarchy to facilitate subsequent semantic encoding and personalization. At the same time, consistency and coordination between different settings should be ensured to avoid conflicting settings information, such as inconsistent color combinations and confusing functional layouts.

[0048] Through the detailed implementation process of parsing the platform setting information above, different elements in the platform setting information can be more accurately identified and distinguished, providing a solid data foundation for the subsequent dynamic adjustment of identification information based on user preferences.

[0049] In the above-mentioned platform management method, step S120 semantically encodes each platform setting content in the set of platform setting content to obtain a set of semantically encoded platform setting content vectors. Specifically, considering that platform setting content may be text, labels, or other forms of data, it is difficult to directly process it using a computer. Therefore, the present application further semantically encodes each platform setting content in the set of platform setting content to convert each platform setting content into a numerical vector representation, thereby generating a set of semantically encoded platform setting content vectors, enabling a deep learning model to perform data analysis and processing on the platform setting content. In a specific implementation, for text-based platform setting content, natural language processing techniques such as word embedding or the BERT model can be used to convert the text content into a vector representation rich in semantic information. For label-based data, methods such as one-hot encoding are used for conversion. For image data such as logo designs, image processing techniques such as convolutional neural networks (CNNs) are used for feature extraction. Through the above processing, each platform setting content can be converted into a unified numerical vector format, providing a basis for subsequent personalized adjustments.

[0050] In the above-mentioned platform management method, the step S130 obtains a text description of the user's preferences. In the technical solution of the present application, the text description of the user's preferences is a text description directly input by the user, which includes the user's personalized needs for the platform interface style, functional layout, brand elements, etc. By introducing the user's preference information, the platform setting content can be adjusted more accurately to meet the user's personalized needs. For example, if the user prefers a simple interface style, the background color, text color, button color, etc. can be automatically adjusted to light tones; if the user prefers detailed information display, the page layout can be adjusted to add more content display area.

[0051] In the above-mentioned platform management method, the step S140 performs semantic encoding on the text description of the user preference to obtain a user preference semantic encoding vector. Similarly, in order to convert the textual description of the user preference into a computer-processable format, the present application uses natural language processing technology to perform semantic encoding on the text description of the user preference to extract the deep contextual semantic information in the user preference description and generate a user preference semantic encoding vector. In a specific example of the present application, a semantic encoder based on the Bert model is used to perform semantic encoding on the text description of the user preference to obtain the user preference semantic encoding vector. Specifically, the Bert model is a pre-trained language representation model that can capture the bidirectional contextual information of each word in the text through its internal bidirectional Transformer architecture, thereby fully understanding the meaning of each word in a specific context and providing a more accurate semantic representation. Based on this, the present application can accurately capture the user's specific preferences and deeply understand the contextual information of the user preference description by using the Bert model to semantically encode the text description of the user preference, thereby providing high-quality semantic encoding feature representation for the text description of the user preference and providing strong support for subsequent personalized customization.

[0052] In the aforementioned platform management method, step S150 performs fine-grained cross-domain association learning on the set of platform setting content semantic coding vectors and the user preference semantic coding vectors to obtain a set of user preference-modulated platform setting content semantic coding vectors. That is, to achieve a deep integration of platform setting content and user preference information, the present application further utilizes user preference information to dynamically modulate various platform setting contents by performing cross-domain association learning on the set of platform setting content semantic coding vectors and the user preference semantic coding vectors.

[0053] Figure 3 FIG. 1 is a flow chart of step S150 in the platform management method according to an embodiment of the present application. Figure 3As shown, the step S150 includes: S151, performing association strength measurement on each platform setting content semantic coding vector in the set of the user preference semantic coding vector and the platform setting content semantic coding vector to obtain a set of user preference-platform setting content fine-grained ablation factors; S152, based on the set of user preference-platform setting content fine-grained ablation factors, performing fine-grained ablation modulation on the set of platform setting content semantic coding vectors to obtain the set of user preference modulated platform setting content semantic coding vectors.

[0054] Specifically, step S151 includes: first, performing a cross-domain query interaction based on an attention mechanism on each platform setting content semantic encoding vector in the set of the user preference semantic encoding vector and the platform setting content semantic encoding vector to obtain a set of user preference-platform setting content cross-domain query interaction feature vectors. More specifically, the cross-domain query interaction based on the attention mechanism includes: performing a linear transformation on the user preference semantic encoding vector to obtain a query vector and a value vector; performing a linear transformation on the platform setting content semantic encoding vector to obtain a key vector; and inputting the query vector, the value vector, and the key vector into a cross-domain interaction encoder based on an imitation transformer structure to obtain the user preference-platform setting content cross-domain query interaction feature vector, which is expressed as follows:

[0055]

[0056]

[0057]

[0058]

[0059]

[0060] in, represents the user preference semantic encoding vector, represents a set of semantic coding vectors of the platform settings content, 、 、 and Respectively represent the first, second, and third in the set of semantic coding vectors of the platform setting content and Each platform sets the content semantic encoding vector, Setting the number of vectors in the set of content semantic encoding vectors for the platform, 、 and denote the query embedding matrix, value embedding matrix and key embedding matrix respectively, 、 and Represent different bias terms, represents the matrix multiplication operation, 、 and Represent the query vector, value vector and the The corresponding key vector, is the characteristic scale value of the key vector, is the normalized exponential function, represents the transpose of a vector, Indicates the and stated The feature vector of user preference-platform setting content cross-domain query interaction.

[0061] First, a query vector and a value vector are constructed based on the user preference semantic encoding vector. A key vector is constructed based on the semantic encoding vectors of each platform setting content in the set of platform setting content semantic encoding vectors. The correlation between user preference information and platform setting content is captured through cross-domain query interaction based on a transformer structure. In this process, the transformer structure effectively handles information interaction between different domains and strengthens the correlation between user preferences and platform setting content through an attention mechanism, thereby establishing a direct connection between user preferences and each platform setting content.

[0062] Then, each user preference-platform setting content cross-domain query interaction feature vector in the set of user preference-platform setting content cross-domain query interaction feature vectors is input into the ablation metric function to obtain a set of user preference-platform setting content fine-grained ablation factors, which is expressed as follows:

[0063]

[0064] in, represents the ablation metric function, Indicates taking the maximum value, and Respectively represent the The characteristic mean and characteristic variance of represents the regularization term, Indicates the The corresponding user preference-platform setting content fine-grained ablation factor.

[0065] That is, by further introducing the ablation metric function, the cross-domain query interaction characteristics between user preference information and the settings of each platform are evaluated and quantified. Among them, the ablation metric function is similar to the ablation analysis in experimental design, which is used to judge the impact of removing a certain specific relationship on the final interaction effect. It helps to identify the degree of influence of user preference information on the settings of each platform, thereby ensuring that the platform settings that are highly relevant to user preferences are given priority while reducing the impact of redundant platform settings. For example, if the user preference information indicates that the user prefers a simple interface style, then the settings related to the interface style in the platform setting information will be given a higher weight. In this way, the personalized customization process can be ensured to be more efficient and accurate.

[0066] Specifically, step S152 includes: first, inputting the set of user preference-platform setting content fine-grained ablation factors into an ablation effect encoding module including a normalization function and a masking function to obtain a set of user preference-platform setting content fine-grained ablation weight factors, which is expressed as:

[0067]

[0068] in, represents the exponential function with base e, Indicates the The corresponding normalized user preference-platform setting content fine-grained ablation factor, is the mask function, is the mask threshold, For the The corresponding user preference-platform setting content fine-grained ablation weight factor.

[0069] That is, to standardize the weights and exclude irrelevant items, the set of fine-grained ablation factors for user preference-platform setting content is further normalized and masked. The normalization function is used to convert all ablation factors to a standard range (e.g., between 0 and 1) for direct comparison; while the masking function is used to increase the ablation factors of platform settings that are significantly correlated with user preferences, while reducing or excluding the ablation factors of platform settings that are irrelevant or less correlated with user preferences, thereby focusing resources and attention on the platform settings that are most closely aligned with user preferences.

[0070] Then, based on the set of user preference-platform setting content fine-grained ablation weight factors, the set of platform setting content semantic coding vectors is weighted modulated to obtain the set of user preference modulated platform setting content semantic coding vectors.

[0071]

[0072] in, 、 、 and are respectively 、 、 and stated The corresponding user preference-platform setting content fine-grained ablation weight factor, A collection of semantic encoding vectors representing user preference modulation platform settings content.

[0073] Specifically, based on the generated set of fine-grained ablation weight factors for the user preference-platform setting content, a fine-grained ablation modulation is performed on the original set of semantic encoding vectors of the platform setting content. This strengthens the platform setting content that is highly consistent with user preferences while also preserving the uniqueness of the original platform setting content, thereby achieving personalized customization and optimization of the platform setting content. In this way, users can enjoy platform services that better meet their personal needs.

[0074] In the above-mentioned platform management method, the step S160 generates the identification information of the platform page by modulating the set of semantic coding vectors of the content of the platform setting based on the user preference. In a specific example of the present application, after adding a prompt word to the end of the set of semantic coding vectors of the content of the platform setting for the user preference modulation, the identification information dynamic adjuster based on the large model is input to obtain the identification information of the platform page, and the prompt word is to adjust the identification information based on the user preference. That is, by utilizing the deep learning ability of the large model, the set of semantic coding vectors of the content of the platform setting for the user preference modulation is further feature-learned and refined to generate platform page identification information that matches the user's personalized needs. Here, the role of the prompt word is to guide and constrain. By explicitly specifying the prompt word as "adjusting identification information based on user preferences", it can be ensured that the large model always takes the user's personalized needs as the primary consideration when generating the platform page identification information, which helps to maintain a high degree of consistency between the output results and the user's expectations, thereby providing an interface design that is more in line with the user's preferences. In a specific example of the present application, the large model is a GPT series model.

[0075] Preferably, after adding a prompt word to the end of the set of content semantic coding vectors of the user preference modulation platform, inputting a large model-based identification information dynamic adjuster to obtain the identification information of the platform page includes:

[0076] First, the set of user preference modulation platform setting content semantic coding vectors is concatenated to obtain a user preference modulation platform setting content semantic concatenated coding vector;

[0077] Next, the distance between each pair of eigenvalues ​​of the user preference modulation platform setting content semantic cascade coding vector is calculated, such as the L2 distance, and the square root of the distance is taken to obtain the user preference modulation platform setting content semantic cascade distance representation matrix, that is,

[0078]

[0079] in, Indicates user preference modulation platform settings content semantic cascade distance representation matrix The element value at the position, Represents the user preference modulation platform setting content semantic cascade coding vector, 、 Respectively represent the user preference modulation platform setting content semantic cascade coding vector Position and The eigenvalues ​​at the positions, represents the distance metric function;

[0080] Then, the user preference modulation platform setting content semantic cascade autocorrelation matrix of the user preference modulation platform setting content semantic cascade coding vector is obtained as a row vector, that is:

[0081]

[0082] in, Representing the user preference modulation platform setting content semantic cascade autocorrelation matrix;

[0083] Secondly, the user preference modulation platform setting content semantic cascade coding vector is matrix multiplied with the user preference modulation platform setting content semantic cascade distance representation matrix to obtain the user preference modulation platform setting content semantic cascade first-level mapping vector, that is:

[0084]

[0085] in, Represents user preference modulation platform setting content semantic cascade distance representation matrix, Represents the user preference modulation platform setting content semantic cascade first-level mapping vector;

[0086] Furthermore, the user preference modulation platform setting content semantic cascade first-level mapping vector is matrix-multiplied by the matrix product of the user preference modulation platform setting content semantic cascade distance representation matrix and the user preference modulation platform setting content semantic cascade self-correlation matrix to obtain the user preference modulation platform setting content semantic cascade multi-level mapping vector, that is:

[0087]

[0088] in, Represents user preference modulation platform setting content semantic cascade multi-level mapping vector;

[0089] Finally, the user preference modulation platform setting content semantic cascade multi-level mapping vector is interpolated with the user preference modulation platform setting content semantic cascade associated eigenvector (if the eigenvalue is insufficient, interpolation or zero padding) composed of the eigenvalue of the user preference modulation platform setting content semantic cascade auto-association matrix to obtain an optimized user preference modulation platform setting content semantic cascade coding vector; after adding a prompt word to the end of the optimized user preference modulation platform setting content semantic cascade coding vector, it is input into the large model-based identification information dynamic adjuster to obtain the identification information of the platform page.

[0090] Among them, considering that the set of semantic coding vectors of the platform setting content represents the semantic coding features of each platform setting content, and the user preference semantic coding vector represents the semantic coding features of user preference, when performing cross-domain attention joint encoding on them, the semantic coding features of the user preference are used to perform fine-grained ablation modulation on the semantic coding features of each platform setting content. However, the semantic flow field offset between the semantic coding features of user preference and the semantic coding features of each platform setting content will cause the set of user preference modulated platform setting content semantic coding vectors to have a lack of fine-grained interactive feature instance judgment, thereby affecting the accuracy of the platform page identification information obtained by adding the prompt word and inputting the identification information dynamic adjuster based on the large model.

[0091] Therefore, the linear target mapping representation of the similarity distance representation matrix of the content semantic cascade coding vector of the user preference modulation platform setting is obtained by cascading the set of content semantic coding vectors of the user preference modulation platform setting, to perform a secondary target mapping representation based on the multi-level distribution hierarchy on the complete similarity instantiation of the self-association of the content semantic cascade coding vector of the user preference modulation platform setting, and compensate for the negative impact factor of the association mismatch through the association fusion kernel bias, so as to improve the instance judgment degree of the eigenvalue of the content semantic cascade coding vector of the user preference modulation platform setting under the similarity constraint, that is, the significance degree of the eigenvalue as an instance for the generative regression judgment, and improve the accuracy of the identification information of the platform page obtained by adding the prompt word to the tail of the content semantic cascade coding vector of the user preference modulation platform setting and inputting the identification information dynamic adjuster based on the large model.

[0092] In summary, the platform management method according to the embodiment of the present application is explained. It first parses the platform setting information to break it down into various platform setting contents. Then, it further introduces deep learning-based artificial intelligence technology to perform semantic parsing and fine-grained semantic query interaction on each platform setting content and the preference description entered by the user. In this way, the various platform setting contents are dynamically modulated according to the user preference information, thereby intelligently generating platform page identification information that meets the user's personalized needs. In this way, personalized customization of platform pages can be achieved to meet the specific needs of different users.

[0093] Exemplary terminal device, reference Figure 4 To describe the terminal device according to an embodiment of the present application. Figure 4 FIG is a block diagram of a terminal device according to an embodiment of the present application. Figure 4 As shown, the terminal device 10 includes one or more processors 11 and a memory 12 .

[0094] The processor 11 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the terminal device 10 to perform desired functions.

[0095] The memory 12 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), a hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may execute the program instructions to implement the platform management methods of the various embodiments of the present application described above and / or other desired functions. Various content, such as platform setting information and text descriptions of user preferences, may also be stored in the computer-readable storage medium.

[0096] In one example, the terminal device 10 may further include an input device 13 and an output device 14 , and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0097] The input device 13 may include, for example, a keyboard, a mouse, and the like.

[0098] The output device 14 can output various information to the outside, including identification information of the platform page, etc. The output device 14 can include, for example, a display, a speaker, a printer, a communication network and its connected remote output device, etc.

[0099] Of course, to simplify, Figure 4 Only some of the components related to the present application in the terminal device 10 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, the terminal device 10 may further include any other appropriate components according to specific application scenarios.

[0100] Exemplary computer program product and computer-readable storage medium. In addition to the above-mentioned methods and systems, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the platform management method according to various embodiments of the present application described in the above-mentioned exemplary method section of this specification.

[0101] The computer program product may be written in any combination of one or more programming languages ​​to implement the program code for performing the operations of the embodiments of the present application, including object-oriented programming languages ​​such as Java, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0102] In addition, an embodiment of the present application may also be a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, enables the processor to execute the steps of the platform management method according to various embodiments of the present application described in the above exemplary method section of this specification.

[0103] The computer-readable storage medium may be any combination of one or more readable storage media. The readable storage medium may be a readable signal storage medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0104] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the composition and steps of each example according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0105] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A platform management method, characterized in that: include: Obtaining platform setting information and determining identification information of the platform page according to the platform setting information includes: Parsing the platform setting information to obtain a set of platform setting contents; performing semantic coding on each platform setting content in the set of platform setting content to obtain a set of platform setting content semantic coding vectors; Get a text description of the user's preferences; Performing semantic encoding on the text description of the user preference to obtain a user preference semantic encoding vector; Performing fine-grained cross-domain association learning on the set of platform setting content semantic coding vectors and the user preference semantic coding vectors to obtain a set of user preference modulation platform setting content semantic coding vectors; Modulating a set of semantic coding vectors of platform setting content based on the user preference to generate identification information of the platform page; After adding a prompt word to the end of the set of semantic coding vectors of the user preference modulation platform setting content, inputting the identification information dynamic adjuster based on the large model to obtain the identification information of the platform page, including: Cascading the set of user preference modulation platform setting content semantic coding vectors to obtain a user preference modulation platform setting content semantic cascade coding vector; Calculating the distance between each pair of eigenvalues ​​of the user preference modulation platform setting content semantic cascade coding vector, and taking the square root of the distance to obtain a user preference modulation platform setting content semantic cascade distance representation matrix; Obtaining a user preference modulation platform setting content semantic cascade autocorrelation matrix of the user preference modulation platform setting content semantic cascade coding vector as a row vector; Performing matrix multiplication on the user preference modulation platform setting content semantic cascade coding vector and the user preference modulation platform setting content semantic cascade distance representation matrix to obtain a user preference modulation platform setting content semantic cascade primary mapping vector; Performing matrix multiplication of the user preference modulation platform setting content semantic cascade primary mapping vector and the matrix product of the user preference modulation platform setting content semantic cascade distance representation matrix and the user preference modulation platform setting content semantic cascade self-correlation matrix to obtain a user preference modulation platform setting content semantic cascade multi-level mapping vector; The user preference modulation platform setting content semantic cascade multi-level mapping vector is interpolated with the user preference modulation platform setting content semantic cascade association eigenvector composed of the eigenvalues ​​of the user preference modulation platform setting content semantic cascade self-association matrix to obtain an optimized user preference modulation platform setting content semantic cascade coding vector; after adding a prompt word to the end of the optimized user preference modulation platform setting content semantic cascade coding vector, it is input into the large model-based identification information dynamic adjuster to obtain the identification information of the platform page.

2. The platform management method according to claim 1, characterized in that: Performing semantic encoding on the text description of the user preference to obtain a user preference semantic encoding vector, comprising: The text description of the user preference is semantically encoded using a semantic encoder based on the Bert model to obtain the user preference semantic encoding vector.

3. The platform management method according to claim 2, characterized in that: Performing fine-grained cross-domain association learning on the set of platform setting content semantic coding vectors and the user preference semantic coding vectors to obtain a set of user preference modulation platform setting content semantic coding vectors, including: performing association strength measurement on each platform setting content semantic coding vector in the set of the user preference semantic coding vector and the platform setting content semantic coding vector to obtain a set of user preference-platform setting content fine-grained ablation factors; Based on the set of user preference-platform setting content fine-grained ablation factors, the set of platform setting content semantic coding vectors is fine-grained ablation modulated to obtain the set of user preference modulated platform setting content semantic coding vectors.

4. The platform management method according to claim 3, characterized in that: Performing an association strength measurement on each platform setting content semantic coding vector in the set of the user preference semantic coding vector and the platform setting content semantic coding vector to obtain a set of user preference-platform setting content fine-grained ablation factors, including: Performing a cross-domain query interaction based on an attention mechanism on each platform setting content semantic encoding vector in the set of the user preference semantic encoding vector and the platform setting content semantic encoding vector to obtain a set of user preference-platform setting content cross-domain query interaction feature vectors; Each user preference-platform setting content cross-domain query interaction feature vector in the set of user preference-platform setting content cross-domain query interaction feature vectors is input into an ablation metric function to obtain a set of user preference-platform setting content fine-grained ablation factors.

5. The platform management method according to claim 4, characterized in that: Performing a cross-domain query interaction based on an attention mechanism on each platform setting content semantic encoding vector in the set of the user preference semantic encoding vector and the platform setting content semantic encoding vector to obtain a set of user preference-platform setting content cross-domain query interaction feature vectors, including: Performing a linear transformation on the user preference semantic encoding vector to obtain a query vector and a value vector; Performing a linear transformation on the platform setting content semantic encoding vector to obtain a key vector; The query vector, the value vector and the key vector are input into a cross-domain interaction encoder based on an imitation transformer structure to obtain the user preference-platform setting content cross-domain query interaction feature vector.

6. The platform management method according to claim 5, characterized in that: Based on the set of user preference-platform setting content fine-grained ablation factors, fine-grained ablation modulation is performed on the set of platform setting content semantic coding vectors to obtain the set of user preference modulated platform setting content semantic coding vectors, including: Inputting the set of user preference-platform setting content fine-grained ablation factors into an ablation effect encoding module including a normalization function and a masking function to obtain a set of user preference-platform setting content fine-grained ablation weight factors; Based on the set of user preference-platform setting content fine-grained ablation weight factors, the set of platform setting content semantic coding vectors is weighted modulated to obtain the set of user preference modulated platform setting content semantic coding vectors.

7. A terminal device, characterized in that: include: a memory for storing instructions; A processor is coupled to the memory, and the processor is configured to execute the platform management method according to any one of claims 1 to 6 based on instructions stored in the memory.

8. A storage medium, characterized in that: The storage medium stores a platform management program, and when the platform management program is executed by the processor, the platform management method according to any one of claims 1 to 6 is implemented.

Images