A website content management method, system, device and storage medium based on large model machine reading

Abstract

This invention relates to the field of data management technology and discloses a method, system, device, and storage medium for website content management based on large-scale machine reading. The method includes: relying on the website backend interface, automatically generating structured schema data and an llms.txt site description file after matching the content type based on the semantic tag set and content hierarchy graph within the reading results of the reading content data and associated metadata; establishing source management records based on associated indexes; comparing the source management records with the source description records to determine the content's credibility and version status, and completing corresponding data interaction operations according to the access request path; setting a fixed management cycle, periodically iteratively updating the site description file and the structured data of changed content, and dynamically adjusting operating parameters based on statistical feedback from content access and citations, thereby achieving intelligent management of website content for large-scale machine reading; this invention can improve the efficiency of website content management based on large-scale machine reading.

Description

Technical Field

[0001] This invention relates to the field of data management technology, and in particular to a method, system, device, and storage medium for website content management based on large-scale machine reading. Background Technology

[0002] Current website content management models for large-scale machine reading are relatively traditional. Most platforms can only complete basic content publishing and simple tagging, lacking standardized semantic structuring capabilities. Website content cannot generate dedicated semantic data and site description files adapted for large-scale model crawling and parsing. When reading website content, large models struggle to accurately identify content levels, entity attributes, and relationships, which not only reduces the parsing efficiency of machine reading but also easily leads to problems such as content comprehension bias and incomplete information extraction. Overall, the adaptation effect of content to large-scale models is poor.

[0003] Existing technologies lack robust mechanisms for source verification, version control, and dynamic iteration, making it impossible to systematically check the credibility of website content sources and data version status. Furthermore, data such as content updates and access popularity cannot drive adjustments to management strategies. Over the long term, this can lead to issues such as invalid content continuously consuming resources, delayed responses to high-value content, and the inability to effectively block illegal paths. Consequently, the reliability and operational efficiency of website content when accessed by large models are compromised, and the overall management system exhibits significant shortcomings in intelligence and adaptability. Summary of the Invention

[0004] This disclosure provides a method, system, device, storage medium, and computer program for website content management based on large-scale machine reading.

[0005] Firstly, this disclosure provides a website content management method based on large-scale machine reading models, including: Step 1: Through the website backend interface, perform structured parsing of the semantics of the collected reading content data and the metadata associated with the content to obtain the reading results of the reading content data; Step 2: Using a predefined machine reading model, based on the semantic tag set and content hierarchy map in the reading results, automatically generate structured schema data and llms.txt site description file that match the content type of the reading results; Step 3: Add the structured schema data to the header or body area of ​​the website's front-end page as embedded code or related references. Based on the association index between the readable semantic description and the reading content data in the machine reading model, establish a source management record for the reading content data. Step 4: Compare and correlate the source management records with the source description records in the reading results. Based on the comparison results, determine the credibility of the description of the reading content data and the data version status. Based on the request path parsed from the reading request, complete the corresponding operation of the readable data in the machine reading model. Step 5: Repeat steps 1 to 4 within a fixed management cycle to generate a new site-level llms.txt file and update the structured schema data for each changed content data. Based on the reading feedback of website content and content citation statistics, dynamically adjust the condition parameters within the management cycle.

[0006] In a preferred embodiment, the step of automatically generating structured schema data and an llms.txt site description file that match the content type of the reading results based on the semantic tag set and content hierarchy map in the reading results includes: Based on the semantic tag set and content hierarchy graph in the reading results, the content type in the reading content data is identified. The content types include articles, product descriptions, case descriptions, service introductions, or press releases. Based on the content type, the system calls the pre-built Schema.org vocabulary template to generate structured schema data in the corresponding format. The structured schema data includes the title, text summary, publication time, author information, entity attributes, and the relationships between them and the content data. While generating structured schema data, an llms.txt generator is invoked to collect site semantic descriptions, content scope, and reliable content paths from the structured reading results corresponding to all reading content data of the current website; The llms.txt generator generates a site-level llms.txt file based on the collected information. The llms.txt file contains at least the following: a list of website content scope, a list of key page URLs, a set of trusted content paths, a set of prohibited paths, content update description text, and site semantic description text. The generated llms.txt file is deployed to the website's root directory, where it, together with the structured schema data, constitutes a machine-readable dataset for large models.

[0007] In a preferred embodiment, adding structured schema data to the header or body area of ​​the website's front-end page via embedded code or associative references, and establishing an association index between readable semantic descriptions and reading content data in a machine reading model, includes: Convert structured schema data into embedded code blocks in JSON-LD format, and insert the embedded code blocks into the script tags in the HTML header or body area of ​​the website front-end page, so that the website front-end page directly exposes machine-readable semantic descriptions when accessed by large models or artificial intelligence crawling programs. In the backend content library of the website content management system, create associated index records for the reading content data. The associated index records contain the unique identifier of the reading content data, the storage path of the structured schema data, and the file address of the readable semantic description output by the machine reading model.

[0008] In a preferred embodiment, establishing a source management record for reading content data includes: Based on the associated index records, a source management record is generated for each piece of reading content data. The source management record includes the source URL, data type, credibility level, original collection time, and last verification time of the cited materials. The source description records output by the machine reading model are compared one by one with the source URLs of the cited materials in the source management records; When an external reference URL in the source description record does not appear in the source management record, the external reference URL will be automatically added as a new source to the source management record, and the trust level of the new source will be temporarily set to pending verification. When the trust level in the information source management record is lower than the preset verification trigger level, an information source verification prompt is generated. The information source verification prompt is used to notify website administrators or automatic verification programs to re-verify the information source.

[0009] In a preferred embodiment, the step of associating and comparing the source management record with the source description record in the reading result, determining the credibility of the description and the data version status of the reading content data based on the comparison result, and completing the corresponding operation of the readable data in the machine reading model according to the request path parsed from the reading request includes: Each source URL in the source management record is matched with the external reference URL in the source description record. If a match is successful and the credibility level in the source management record is verified, the credibility of the description of the reading content data is marked as high credibility. If the match fails or the credibility level is pending verification, the credibility of the reading content data will be marked as pending review. Based on the original collection time and the last verification time in the information source management record, determine whether the reading content data has undergone content changes. If it has undergone changes, increment the version number in its data version status and retain the content snapshot of the previous version. If it has not undergone changes, maintain the original version status. When a read request is received from a large model or AI crawling program, the request path is parsed from the read request and matched against the set of prohibited read paths in the llms.txt site description file. If the match is successful, an access denied response is returned. If the requested path matches the trusted content path or key page list in the llms.txt site description file, the reading content data corresponding to the requested path, along with the associated description credibility and data version status, are obtained and returned to the machine reading model along with the structured schema data. This allows the machine reading model to perform readable data referencing or filtering operations based on the description credibility and data version status.

[0010] In a preferred embodiment, steps 1 to 4 are repeated at a fixed management cycle to generate a new site-level llms.txt file and update the structured schema data for each changed content data, including: At the end of each management cycle, steps 1 to 4 are executed again to obtain the changed reading content data for the current cycle from the website content management terminal. The changed reading content data includes content addition, content modification, or content deletion. For changes to the reading content data, the predefined machine reading model is called again to generate new structured reading results, and the corresponding structured schema data is regenerated based on the new structured reading results. The original structured schema data in the backend content library is replaced with the regenerated structured schema data. Collect the site semantic description, content scope, and trusted content path from the latest structured reading results of all reading content data on the website within the current management cycle. Call the llms.txt generator to regenerate a site-level llms.txt file and overwrite the old version file in the website root directory with the newly generated llms.txt file.

[0011] In a preferred embodiment, the step of dynamically adjusting the condition parameters within the management period based on reading feedback and content citation statistics of the reading website content includes: Based on the statistical results of the frequency of content requests and citations on the website, dynamically adjust at least one of the following conditional parameters when executing steps 1 to 4 within the current management cycle: The verification trigger condition for the trust level in the source management record marks the source corresponding to the reading content data with zero citations in the previous period as a low-activity source, thereby reducing the automatic decay rate of the trust level. The sensitivity of content change detection is improved. When reading content data that was requested more frequently than the average in the previous period changes, the update response time of its structured schema data is shortened, and it is given priority to be regenerated in the current period. The rules for compiling the list of key pages in the llms.txt file are as follows: reading content data that ranks highly in both request frequency and citation count in the previous period is automatically added to the list of key pages, and content paths that have not been requested for a long time are removed from the list. The dynamically adjusted condition parameters will be used as the basis for repeatedly executing steps 1 to 4 within the current management cycle, and will be persistently stored in the configuration library of the website content management terminal.

[0012] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention can perform semantic structured parsing of website content and metadata, automatically generating standardized structured data and site description documents, enabling pages to possess standardized machine-readable semantic information, and significantly improving the recognition, parsing, and reading performance of large models for website content. Simultaneously, relying on associated indexes, it completes source archiving, achieving automated determination of content source credibility and data version, ensuring the standardization and accuracy of externally output content.

[0013] 2. This invention iteratively updates content data at fixed intervals and dynamically adjusts management parameters based on content access frequency and citation statistics to continuously optimize content update response speed, source verification rules, and site path configuration. This entire operational mechanism enables automated and adaptive website content management, effectively improving content operation and maintenance efficiency and stably maintaining data quality and access control capabilities during large model reading processes. Attached Figure Description

[0014] The present disclosure will be described in more detail below based on embodiments and with reference to the accompanying drawings: Figure 1 The flowchart illustrates a website content management method based on large-scale machine reading according to Embodiment 1 of the present invention. Figure 2 This diagram shows the functional modules of a website content management system based on large-scale machine reading according to Embodiment 2 of the present invention. Figure 3 The diagram shows the structural composition of an electronic device that implements a website content management method based on large-scale machine reading, according to Embodiment 3 of the present invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the technical solutions of this disclosure, and to fully understand and implement the process of how this disclosure applies technical means to solve technical problems and achieve corresponding technical effects, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, not all embodiments. The embodiments of this disclosure and the various features within them can be combined with each other without conflict, and the resulting technical solutions are all within the protection scope of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort should fall within the protection scope of this disclosure.

[0016] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0017] Example 1 Figure 1 This is a flowchart illustrating a website content management method based on large-scale machine reading, provided as an embodiment of this disclosure. Figure 1 As shown, a website content management method based on large-scale machine reading includes: In this embodiment of the invention, step 1 involves performing structured parsing of the semantics of the collected reading content data and content-related metadata through the website backend interface to obtain the reading results of the reading content data. A dedicated data receiving channel is built based on the pre-deployed website backend interface. This channel sets a fixed data transmission format as the access benchmark. External collection devices push reading content data and content-related metadata in a unified manner according to this format. The backend interface continuously monitors the data access status. When a complete data packet is detected, it is immediately split and decomposed to separate the independent reading content data part and the content-related metadata part. The two types of data are bound and stored according to the corresponding relationship at the time of original collection, ensuring that a single piece of reading content data can accurately match all the corresponding content-related metadata.

[0018] For the split and bound reading content data, the content is decomposed according to the preset semantic segmentation rules. These rules use the content's theme and constituent sections as the dividing criteria to divide the complete reading content data into independent text sections. Each text section is assigned a unique semantic category. Staff members pre-define a list of semantic categories for all categories of content based on the website's content operation guidelines. All segmented text sections are semantically categorized and labeled according to this list. The labeled text sections retain the original text content and the corresponding semantic labeling information.

[0019] The metadata associated with the content obtained from the synchronous splitting is classified and sorted according to the data attribute categories. The metadata category division is based on the data description object and is divided into four fixed categories: content attribute category, source attribute category, time attribute category, and associated object category. Each piece of metadata will be assigned to the corresponding category, and the correspondence between the metadata and the bound reading content data is recorded to ensure that each piece of metadata can accurately point to its corresponding reading content section.

[0020] The semantically annotated reading content segments and the associated metadata after classification are integrated and reorganized. The reorganization work takes a single piece of original reading content data as the integration unit. All text segments, segment semantic annotations, and supporting classification metadata under the unit are combined in an orderly manner according to the preset structural arrangement rules. The structural arrangement rules are fixed as a three-layer structure of basic content layer, semantic annotation layer, and metadata supplement layer. The three layers of content are arranged and combined in sequence to form a complete set of structured information.

[0021] According to the website's pre-defined reading result formatting specifications, the integrated structured information set is formatted. These specifications clearly define the carrier format, information arrangement order, and information display scope of the reading results. During the formatting process, duplicate text content and redundant metadata entries are removed, and only valid information directly related to the core semantics of the reading content is retained. After the complete formatting operation, the reading results with specified names are finally generated. These results fully carry the original reading content, section semantic information, and all related metadata information.

[0022] The beneficial effects include achieving precise separation and binding of reading content data and associated metadata based on standardized data reception and segmentation methods, ensuring accurate correspondence between the two types of data. Content semantic segmentation, annotation, and metadata classification are completed according to established rules, allowing various types of information to form a clear hierarchy and classification. Information integration and format standardization are achieved through a fixed structure, outputting standardized and uniform reading results, providing a complete, orderly, and reliable data foundation for subsequent website content management operations.

[0023] Step 2: Using a predefined machine reading model, based on the semantic tag set and content hierarchy map in the reading results, automatically generate structured schema data and llms.txt site description file that match the content type of the reading results; In this embodiment of the invention, the step of automatically generating structured schema data and an llms.txt site description file that match the content type of the reading results based on the semantic tag set and content hierarchy graph in the reading results includes: Based on the semantic tag set and content hierarchy graph in the reading results, the content type in the reading content data is identified. The content types include articles, product descriptions, case descriptions, service introductions, or press releases. Based on the content type, the system calls the pre-built Schema.org vocabulary template to generate structured schema data in the corresponding format. The structured schema data includes the title, text summary, publication time, author information, entity attributes, and the relationships between them and the content data. While generating structured schema data, an llms.txt generator is invoked to collect site semantic descriptions, content scope, and reliable content paths from the structured reading results corresponding to all reading content data of the current website; The llms.txt generator generates a site-level llms.txt file based on the collected information. The llms.txt file contains at least the following: a list of website content scope, a list of key page URLs, a set of trusted content paths, a set of prohibited paths, content update description text, and site semantic description text. The generated llms.txt file is deployed to the website's root directory, where it, together with the structured schema data, constitutes a machine-readable dataset for large models.

[0024] The system retrieves the parsed reading results, extracts the complete semantic tag set and content hierarchy map, and determines the content type by referring to the platform's pre-set content type matching benchmark. This benchmark establishes a unique correspondence between different semantic tag combinations and content hierarchy distribution patterns and five types of content: articles, product descriptions, case descriptions, service introductions, and press releases. After comparing the tag combinations and hierarchy structure, the system determines the specific content type to which the current reading content data belongs.

[0025] Based on the determined content type, the system retrieves the Schema.org glossary template that has been configured in advance. Different content types correspond to independent template files. Fixed information filling positions are defined within the template. The title content, text summary content, publication time information, author information, entity attribute content, and the relationships between content data in the reading results are filled into the corresponding positions in the template one by one. The template integrates all the filled information according to the established layout rules, thereby forming a complete structured schema data.

[0026] At the same time that the process of creating structured schema data is started, the built-in llms.txt generator of the system is activated. This tool accesses the structured reading results corresponding to each piece of reading content data one by one according to the website's full domain data traversal rules. It extracts the site semantic description text, content coverage information and verified and confirmed credible content paths from each result, and summarizes all the extracted information into the designated storage area to complete centralized storage.

[0027] The llms.txt generator reads all the summarized information and organizes various information items in the file according to the preset content arrangement order. Based on the website page access permission division standards, it forms a list of website content scope, a list of key page URLs, a set of trusted content paths, and a set of prohibited reading paths. At the same time, it organizes the periodic content change records to form content update description text and integrates the original site semantic description text. All information is combined and arranged according to a fixed format, and finally a complete site-level llms.txt file is generated.

[0028] Locate the root directory storage location preset by the website server, and transfer and save the completed site-level llms.txt file to that directory according to the file deployment specifications. After the deployment operation is completed, associate and bind the structured schema data stored in the corresponding position of the page with the llms.txt file in the root directory. The two types of files are combined to form a machine-readable dataset specifically for use by large models. The entire dataset can be directly read and called by large models and related crawling programs.

[0029] The beneficial effects include content type identification based on semantic tags and content hierarchy, accurate matching of corresponding vocabulary templates to produce standardized structured data, and complete retention of core information and relationships within the content. Simultaneously, relevant information from the entire site is collected and integrated to generate a dedicated site description document, clearly defining various access paths and content descriptions. Deploying these two types of documents together forms a standardized, machine-readable dataset, adaptable to the reading and usage needs of large models, ensuring the orderly conduct of data reading and parsing.

[0030] Step 3: Add the structured schema data to the header or body area of ​​the website's front-end page as embedded code or related references. Based on the association index between the readable semantic description and the reading content data in the machine reading model, establish a source management record for the reading content data. In this embodiment of the invention, adding structured schema data to the header or body area of ​​the website front-end page via embedded code or related references, and based on the association index between readable semantic descriptions and reading content data in the machine reading model, includes: Convert structured schema data into embedded code blocks in JSON-LD format, and insert the embedded code blocks into the script tags in the HTML header or body area of ​​the website front-end page, so that the website front-end page directly exposes machine-readable semantic descriptions when accessed by large models or artificial intelligence crawling programs. In the backend content library of the website content management system, create associated index records for the reading content data. The associated index records contain the unique identifier of the reading content data, the storage path of the structured schema data, and the file address of the readable semantic description output by the machine reading model.

[0031] The establishment of information source management records for reading content data includes: Based on the associated index records, a source management record is generated for each piece of reading content data. The source management record includes the source URL, data type, credibility level, original collection time, and last verification time of the cited materials. The source description records output by the machine reading model are compared one by one with the source URLs of the cited materials in the source management records; When an external reference URL in the source description record does not appear in the source management record, the external reference URL will be automatically added as a new source to the source management record, and the trust level of the new source will be temporarily set to pending verification. When the trust level in the information source management record is lower than the preset verification trigger level, an information source verification prompt is generated. The information source verification prompt is used to notify website administrators or automatic verification programs to re-verify the information source.

[0032] The generated structured schema data is rewritten according to the unified syntax specification corresponding to the JSON-LD format. The rewriting process strictly follows the text organization form specified by the format to arrange the original data information, and completely preserves all the content items and information logic relationships contained in the structured schema data, and finally forms a formatted embedded code block.

[0033] Refer to the website front-end page coding standards to determine the fixed positions of script tags in the HTML header and body areas. Enter the completed embedded code blocks into the selected script tags according to the page code embedding rules. The code entry process will not change the original code structure and display content of the page. After the front-end page completes the code embedding, the page will be fully equipped with semantic content that can be recognized by external programs.

[0034] After the external large model and AI crawling program initiate a page access request and complete the page resource loading, it will traverse all code content on the page according to the established content crawling rules. Embedded code blocks in script tags will be normally retrieved and identified, and the machine-readable semantic description content carried in the code block will be directly presented to the outside world, satisfying the usage needs of external programs to obtain semantic information.

[0035] Access the backend content library of the website content management system. The backend content library has a dedicated independent storage partition for storing index records. Create a new record entry in this partition according to the system's preset record creation rules. This entry serves as a unique associated index record for the corresponding reading content data.

[0036] Retrieve the unique identification information that is uniformly assigned to the corresponding reading content data within the system. This identification information serves as the unique identifier for the reading content data. Fill this identifier completely into the designated field corresponding to the associated index record to ensure that each index record can accurately correspond to a single reading content data.

[0037] Find the complete access path of the current structured schema data within the server storage space, and fill this precise storage path into the corresponding field of the associated index record to complete the registration of the structured schema data storage location information.

[0038] The system locates the specific storage address of the readable semantic description file output by the machine reading model in the system, enters the complete file address into the corresponding field of the associated index record, and performs the record saving operation after completing the filling of all specified information, thus forming an associated index record with complete information and standardized entries.

[0039] Retrieve the associated index records that have already been created in the backend content library, and use a single associated index record as the matching basis for the corresponding reading content data. Open a new blank entry in the backend's dedicated source data storage area. This blank entry is the source management record to be improved.

[0040] Extract the source URLs corresponding to the internal references of the reading content data from the original collection archives, and fill the complete and unmodified URL content into the corresponding preset columns of the information source management record to ensure that the URL characters are completely consistent with the original reference information.

[0041] The current referenced material is determined according to the pre-defined data classification standards of the website. The classification standards are based on the content, purpose, and presentation of the material. The name of the determined data type is then entered into the designated column of the information source management record.

[0042] According to the initial system configuration rules, the newly created information source management records are marked with the corresponding trust level. The system sets a fixed level identifier for newly entered regular information sources, and this identifier will be completely stored in the corresponding column of the information source management record.

[0043] When reading the collected content data, the system automatically records the time information and fills it into the column corresponding to the original collection time in the information source management record. This time information is accurate to the time recording granularity uniformly specified by the system.

[0044] The newly created source management record has no historical verification behavior. According to the system default rules, the last verification time field is marked with the blank mark specified by the system. After completing the filling of all field information, the save operation is performed to complete the creation of a single source management record.

[0045] Retrieve the complete source description records output after the machine reading model runs, and simultaneously retrieve the saved source management records one by one. Extract the external reference URLs within each individual source description record, and then extract the source URLs of the referenced materials within each individual source management record to complete a full comparison of the two sets of URL content.

[0046] The system has a complete set of pre-set URL retrieval and matching rules. These rules use exact character matching as the sole criterion for a successful match, and iteratively compares all external URLs with the URLs of the sources cited in the referenced materials.

[0047] If, during the traversal, an external reference URL is found in the source description record that is not present in the existing source management record, the system will open a new information entry position in the corresponding source management record and enter the external reference URL completely into the corresponding position.

[0048] The website backend system sets a fixed initial trust level identifier for newly added information sources. This identifier is always in a pending verification state. This status identifier will be synchronously filled into the trust level field corresponding to the newly added URL, completing the entry and configuration of the new information source.

[0049] The system pre-sets a unified verification trigger level, which serves as a fixed criterion for initiating the re-verification process of the information source. The background will continuously read the trust level information filled in each information source management record and compare the read content with the preset verification trigger level.

[0050] The trust level information registered in the information source management record corresponds to the preset verification trigger level. When the trust level information meets the judgment condition of being lower than the verification trigger level, the system compiles a standardized information source verification prompt text according to the preset prompt generation rules.

[0051] Once the source verification prompts are completed, they will be distributed according to the system's preset push path. The push path is divided into a backend message channel for website administrators and an instruction transmission channel for automatic verification programs. Both types of channels will receive the corresponding source verification prompts simultaneously.

[0052] After logging into the backend, website administrators can view the received source verification prompts in a dedicated message interface and initiate a manual verification process based on the prompts. Upon receiving the source verification prompts, the automatic verification program will initiate an automated source re-verification process according to the established operating logic.

[0053] The beneficial effects include converting structured data into standard embedded code and embedding it in a specified location on the page, allowing external programs to smoothly read machine-readable semantic content and ensuring the normal external display of semantic information. Based on backend partitions, a linked index record is built, fully registering data identifiers, storage paths, and file addresses, achieving unified linked management of various types of data information. Based on the index records, standardized source management records are generated, fully registering key information such as source, type, time, and level, achieving systematic retention of basic source information. By comparing URLs one by one, source omissions are checked and supplementary configurations are completed, automatically standardizing the status labeling of newly added sources, and triggering verification prompts based on fixed-level standards, connecting to both manual and automated verification channels to comprehensively ensure the compliance and reliability of sources cited by the website.

[0054] Step 4: Compare and correlate the source management records with the source description records in the reading results. Based on the comparison results, determine the credibility of the description of the reading content data and the data version status. Based on the request path parsed from the reading request, complete the corresponding operation of the readable data in the machine reading model. In this embodiment of the invention, the step of associating and comparing the source management record with the source description record in the reading result, determining the credibility of the description and the data version status of the reading content data based on the comparison result, and completing the corresponding operation of the readable data in the machine reading model according to the request path parsed from the reading request includes: Each source URL in the source management record is matched with the external reference URL in the source description record. If a match is successful and the credibility level in the source management record is verified, the credibility of the description of the reading content data is marked as high credibility. If the match fails or the credibility level is pending verification, the credibility of the reading content data will be marked as pending review. Based on the original collection time and the last verification time in the information source management record, determine whether the reading content data has undergone content changes. If it has undergone changes, increment the version number in its data version status and retain the content snapshot of the previous version. If it has not undergone changes, maintain the original version status. When a read request is received from a large model or AI crawling program, the request path is parsed from the read request and matched against the set of prohibited read paths in the llms.txt site description file. If the match is successful, an access denied response is returned. If the requested path matches the trusted content path or key page list in the llms.txt site description file, the reading content data corresponding to the requested path, along with the associated description credibility and data version status, are obtained and returned to the machine reading model along with the structured schema data. This allows the machine reading model to perform readable data referencing or filtering operations based on the description credibility and data version status.

[0055] Retrieve the complete source management records and source description records attached to the reading results from the background storage area. Extract all referenced source URLs within the source management records according to the system's set check-up rules. At the same time, extract all external referenced URLs within the source description records. Use complete consistency of character content as the criterion for a successful match to complete the item-by-item check of the two sets of URLs.

[0056] When a single URL is matched and the credibility level marked in the source management record is verified, a high credibility identifier is filled in the dedicated information field of the corresponding reading content data according to the system's preset marking rules, thereby completing the marking operation of the credibility of the content data.

[0057] If two sets of URLs fail to meet the matching standard of completely identical characters, or if the credibility level in the source management record is found to be pending verification after matching, the system will fill in the pending verification mark in the dedicated information field of the corresponding reading content data according to the system's preset marking rules, thereby completing the marking operation of the credibility of the content data description.

[0058] The system extracts the original collection time and the last verification time registered in the source management record. The system uses the existence of a time difference between the two time records as a fixed basis for judging whether the content has changed. If there is a difference between the time records, it is determined that the reading content data has changed.

[0059] Once it is determined that the reading content data has been changed, the version number in the data version status corresponding to the data is updated progressively according to the system's established version management rules. At the same time, all the content before the change is completely saved in the dedicated version snapshot storage area in the background, forming a traceable snapshot of the previous version content.

[0060] When the original collection time and the last verification time are consistent, the system determines that the reading content data has not changed, and the version number and related auxiliary information in the data version status remain unchanged.

[0061] The system monitors the network interface status in real time. When the interface receives a read request from a large model or artificial intelligence crawling program, it disassembles the request message according to a unified path parsing specification and extracts the complete request path information from the fixed fields of the message.

[0062] Retrieve the llms.txt site description file located in the website's root directory, extract the set of prohibited paths compiled within the file, and perform a full character comparison between the parsed request path and each path in the set, using a complete character match as the criterion for a valid match.

[0063] When the requested path matches any entry in the set of prohibited read paths, the system prepares a denial response according to a preset response message format and sends the response back to the program that initiated the read request through the data transmission channel.

[0064] If the requested path does not match the set of prohibited reading paths, and the path matches the trusted content path in the llms.txt site description file or the item in the list of key pages, the system will locate the corresponding reading content data based on the requested path.

[0065] After locating the target reading content data, the system synchronously retrieves the credibility of the data that has been marked, the current effective data version status, and the corresponding generated structured schema data, and integrates all of the above information into a complete data package.

[0066] The integrated data packet is transmitted to the requesting machine reading model through a dedicated data channel. The machine reading model retrieves the credibility of the description and the data version status information in the data packet, and uses this as the basis for determining whether to perform data referencing or data filtering behavior, and completes the corresponding data processing work.

[0067] The beneficial effects include: standardizing content credibility labeling through two-way comparison of URLs and verification of source levels; determining content changes based on time information and maintaining versions and historical snapshots, thus achieving refined control over content quality and versioning; parsing external read request paths and performing permission verification in conjunction with site documentation; differentiating different paths and executing corresponding response actions; and targeted delivery of complete content and supplementary information. This supports machine reading models in processing data according to established standards, ensuring the standardization and security of website content services.

[0068] Step 5: Repeat steps 1 to 4 within a fixed management cycle to generate a new site-level llms.txt file and update the structured schema data for each changed content data. Based on the reading feedback of website content and content citation statistics, dynamically adjust the condition parameters within the management cycle.

[0069] In this embodiment of the invention, the step of repeatedly executing steps 1 to 4 at a fixed management cycle to generate a new site-level llms.txt file and update the structured schema data of each changed content data includes: At the end of each management cycle, steps 1 to 4 are executed again to obtain the changed reading content data for the current cycle from the website content management terminal. The changed reading content data includes content addition, content modification, or content deletion. For changes to the reading content data, the predefined machine reading model is called again to generate new structured reading results, and the corresponding structured schema data is regenerated based on the new structured reading results. The original structured schema data in the backend content library is replaced with the regenerated structured schema data. Collect the site semantic description, content scope, and trusted content path from the latest structured reading results of all reading content data on the website within the current management cycle. Call the llms.txt generator to regenerate a site-level llms.txt file and overwrite the old version file in the website root directory with the newly generated llms.txt file.

[0070] The method of dynamically adjusting condition parameters within the management cycle based on reading feedback and content citation statistics of the reading website includes: Based on the statistical results of the frequency of content requests and citations on the website, dynamically adjust at least one of the following conditional parameters when executing steps 1 to 4 within the current management cycle: The verification trigger condition for the trust level in the source management record marks the source corresponding to the reading content data with zero citations in the previous period as a low-activity source, thereby reducing the automatic decay rate of the trust level. The sensitivity of content change detection is improved. When reading content data that was requested more frequently than the average in the previous period changes, the update response time of its structured schema data is shortened, and it is given priority to be regenerated in the current period. The rules for compiling the list of key pages in the llms.txt file are as follows: reading content data that ranks highly in both request frequency and citation count in the previous period is automatically added to the list of key pages, and content paths that have not been requested for a long time are removed from the list. The dynamically adjusted condition parameters will be used as the basis for repeatedly executing steps 1 to 4 within the current management cycle, and will be persistently stored in the configuration library of the website content management terminal.

[0071] The system continuously keeps time according to preset time rules. When the time result reaches the judgment standard of the predetermined management cycle, it automatically starts the previously set full-process execution logic and runs all operation links from content semantic parsing to source record establishment.

[0072] The system connects to the data interaction interface of the website content management terminal. Based on the content operation logs recorded inside the terminal, it filters out the reading content data that has generated operation behavior within the current period. New content corresponds to newly entered content items in the terminal, content modification corresponds to content items whose original text and attribute information have been changed, and content deletion corresponds to content items marked as removed in the terminal. After filtering, they are all collected into changed reading content data.

[0073] For the collected changed reading content data, the system retrieves the machine reading model that has been pre-deployed on the platform, inputs each piece of changed reading content data into the model in sequence, and processes the data according to the established semantic parsing and structural sorting rules, outputting a brand-new structured reading result.

[0074] Based on each newly generated structured reading result, and referring to the established template specifications and information filling rules, we extract various core information and related content from the results item by item, thereby creating new structured schema data for the corresponding items.

[0075] Locate the corresponding storage location of the original structured schema data in the backend content library, and move the newly generated structured schema data completely into that storage location according to the file replacement execution specifications. Remove the old structured schema data under the same entry to complete the overall data version replacement.

[0076] It iterates through all the structured reading results corresponding to the reading content data within the website, extracts three types of information from each result: site semantic description, content scope, and credible content path, and then aggregates all the extracted information into a temporary storage area designated by the system for centralized storage.

[0077] The system's built-in llms.txt generator is invoked to read all the information that has been summarized in the temporary storage area. Following the fixed format and content organization rules of the file, various information items are integrated to create a brand new site-level llms.txt file.

[0078] Locate the original llms.txt file in the website's root directory, transfer the newly generated site-level llms.txt file to that location according to the file deployment and overwrite rules, remove the old version file remaining in the root directory, and complete the update and replacement of the site description file.

[0079] The system retrieves complete access logs and content citation logs from the website backend within the current management period. Using each piece of read content as a statistical unit, it calculates the frequency of content requests and the number of times content is cited for each piece of read content, forming complete statistical results for subsequent parameter adjustments.

[0080] The system retrieves the verification trigger condition configuration items corresponding to the source management record, and, referring to the citation count statistics retained in the previous period, marks the associated sources of reading content data that have not generated any citation behavior as low-activity sources according to the system's preset identification rules.

[0081] For entries that have completed the low-activity source marking, modify the automatic decay related configurations corresponding to the trust level in the source management record, and lengthen the interval between changes in trust level status according to the system's established rate adjustment rules, thereby completing the adjustment operation of the automatic decay rate of the trust level.

[0082] The system retrieves the configuration entries corresponding to the content change detection. The backend calculates a unified average request frequency benchmark across the entire platform based on the request frequency statistics of the full reading content data. This benchmark is used as a dividing standard to distinguish content data with different popularity.

[0083] For reading content data whose request frequency reaches or exceeds the average request frequency benchmark, the system modifies the update response time configuration specific to this type of content, shortening the interval from detecting content changes to initiating the structured schema data update operation.

[0084] The system adjusts the task sorting rules within the task execution queue during the current cycle, placing update tasks corresponding to reading content data with request frequencies higher than the average request frequency baseline and with content changes at the front of the queue, thereby prioritizing the regeneration of structured schema data.

[0085] The system retrieves the composition rule configuration corresponding to the list of key pages in the llms.txt file, and sorts the full reading content data according to the request frequency statistics and citation statistics retained from the previous period, following a unified sorting rule.

[0086] Access paths for reading content that simultaneously meet both the requirements of ranking high in request frequency and ranking high in citation count will be added to the list of key pages in the llms.txt file according to the list entry rules.

[0087] The system reads all existing content paths in the list of key pages, checks the access records corresponding to each path against the full-cycle access log, and removes content paths that have not generated any access requests during the complete management cycle from the list of key pages according to the list cleanup rules.

[0088] After adjusting any one or more of the following: source verification trigger conditions, content change detection sensitivity, and key page list composition rules, all new configuration content will be determined as the basis for executing the corresponding process within the current management cycle.

[0089] The website content management terminal has a built-in independent configuration library partition. The system organizes all the adjusted condition parameters into standard configuration entries according to the configuration file storage specifications, and writes them completely into the storage location corresponding to the configuration library to achieve persistent data storage. When executing related processes in subsequent cycles, the configuration content in this partition is directly read.

[0090] The beneficial effects include automatically executing the entire content processing workflow at fixed intervals, accurately filtering various changes and iteratively updating corresponding structured data, and synchronously updating site documentation to ensure that all machine-readable information displayed on the site remains up-to-date. It leverages access and reference data to dynamically adapt multiple operating parameters, setting differentiated operating rules for information sources with varying activity levels and content with different popularity, optimizing data update efficiency and page path management logic. The adjusted parameters are permanently saved, ensuring that subsequent processes operate stably according to the adapted rules. This allows the entire content management system to autonomously optimize its operating mode based on actual usage, improving the overall rationality and efficiency of operations and maintenance.

[0091] Example 2 like Figure 2 As shown in the figure, this embodiment also provides a functional module diagram of a website content management system based on large-scale machine reading.

[0092] The website content management system 100 based on large-scale machine reading described in this embodiment can be installed in an electronic device. Depending on the functions implemented, the website content management system 100 based on large-scale machine reading may include a semantic parsing module 101, a file generation module 102, a data deployment and source archiving module 103, a credibility verification and data response module 104, and a periodic iterative optimization module 105. The modules described in this invention can also be called units, referring to a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0093] In this embodiment, the functions of each module / unit are as follows: The semantic parsing module 101 is used to perform structured parsing of the semantics of the collected reading content data and content-related metadata through the website backend interface, so as to obtain the reading result of the reading content data; The file generation module 102 is used to automatically generate structured schema data and llms.txt site description file that match the content type of the reading results based on the semantic tag set and content hierarchy map in the reading results using a predefined machine reading model. The data deployment and information source archiving module 103 is used to add structured schema data to the header or main area of ​​the website front-end page in the form of embedded code or related references, and to establish information source management records of reading content data based on the association index between readable semantic description and reading content data in the machine reading model. The credibility verification and data response module 104 is used to associate and compare the source management record with the source description record in the reading result, determine the credibility of the description and the data version status of the reading content data based on the comparison result, and complete the corresponding operation of the readable data in the machine reading model according to the request path parsed in the reading request. The periodic iteration optimization module 105 is used to repeatedly execute steps 1 to 4 within a fixed management cycle to generate a new site-level llms.txt file and update the structured schema data of each changed content data. Based on the reading feedback of the website content and the content reference statistics, the condition parameters within the management cycle are dynamically adjusted.

[0094] In detail, each module in the website content management system 100 based on large-scale machine reading described in the embodiments of the present invention adopts the same technical means as the website content management method based on large-scale machine reading described in Embodiments 1 and 2, and can produce the same technical effect, which will not be repeated here.

[0095] Example 3 like Figure 3 As shown, this embodiment also provides a computer electronic device, which may include a processor 10, a memory 11, a communication bus 12 and a communication interface 13, and may also include a computer program stored in the memory 11 and capable of running on the processor 10, such as a website content management program based on large model machine reading.

[0096] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the memory 11 (e.g., executing a website content management program based on a large-scale machine learning model), and calls data stored in the memory 11 to perform various functions of the electronic device and process data.

[0097] The memory 11 includes at least one type of medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of an electronic device, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device, such as a plug-in portable hard drive, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. Furthermore, the memory 11 can include both internal and external storage units of the electronic device. The memory 11 can be used not only to store application software and various types of data installed on the electronic device, such as the code of a website content management program based on large-scale machine reading, but also to temporarily store data that has been output or will be output.

[0098] The communication bus 12 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0099] The communication interface 13 is used for communication between the aforementioned electronic device and other electronic devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, Bluetooth interface, etc.), typically used to establish communication connections between the electronic device and other electronic devices. The user interface may be a display, an input unit (such as a keyboard), or optionally, a standard wired or wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device and to display a visual user interface.

[0100] The figure only shows an electronic device with components. Those skilled in the art will understand that the structure shown in the figure does not constitute a limitation on the electronic device and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0101] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0102] The memory 11 in the electronic device stores a website content management program based on large-scale machine reading, which is a combination of multiple instructions. When run in the processor 10, it can achieve the following: Step 1: Through the website backend interface, perform structured parsing of the semantics of the collected reading content data and the metadata associated with the content to obtain the reading results of the reading content data; Step 2: Using a predefined machine reading model, based on the semantic tag set and content hierarchy map in the reading results, automatically generate structured schema data and llms.txt site description file that match the content type of the reading results; Step 3: Add the structured schema data to the header or body area of ​​the website's front-end page as embedded code or related references. Based on the association index between the readable semantic description in the machine reading model and the reading content data, establish a source management record for the reading content data. Step 4: Compare and correlate the source management records with the source description records in the reading results. Based on the comparison results, determine the credibility of the description of the reading content data and the data version status. Based on the request path parsed from the reading request, complete the corresponding operation of the readable data in the machine reading model. Step 5: Repeat steps 1 to 4 within a fixed management cycle to generate a new site-level llms.txt file and update the structured schema data for each changed content data. Based on the reading feedback of website content and content citation statistics, dynamically adjust the condition parameters within the management cycle.

[0103] Specifically, the specific implementation method of the processor 10 for the above instructions can be referred to the description of the relevant steps in the corresponding embodiment of the accompanying drawings, and will not be repeated here.

[0104] Furthermore, if the modules / units integrated into the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a medium. The medium can be volatile or non-volatile. For example, the medium may include: any entity or system capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0105] In the several embodiments provided by this invention, it should be understood that the disclosed electronic devices, systems, and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0106] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0107] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0108] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0109] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A website content management method based on large-scale machine reading, characterized in that, The method includes: Step 1: Through the website backend interface, perform structured parsing of the semantics of the collected reading content data and the metadata associated with the content to obtain the reading results of the reading content data; Step 2: Using a predefined machine reading model, based on the semantic tag set and content hierarchy map in the reading results, automatically generate structured schema data and llms.txt site description file that match the content type of the reading results; Step 3: Add the structured schema data to the header or body area of ​​the website's front-end page as embedded code or related references. Based on the association index between the readable semantic description and the reading content data in the machine reading model, establish a source management record for the reading content data. Step 4: Compare and correlate the source management records with the source description records in the reading results. Based on the comparison results, determine the credibility of the description of the reading content data and the data version status. Based on the request path parsed from the reading request, complete the corresponding operation of the readable data in the machine reading model. Step 5: Repeat steps 1 to 4 within a fixed management cycle to generate a new site-level llms.txt file and update the structured schema data for each changed content data. Based on the reading feedback of website content and content citation statistics, dynamically adjust the condition parameters within the management cycle.

2. The website content management method based on large-scale machine reading as described in claim 1, characterized in that, The process of automatically generating structured schema data and an llms.txt site description file that match the content type of the reading results, based on the semantic tag set and content hierarchy map in the reading results, includes: Based on the semantic tag set and content hierarchy graph in the reading results, the content type in the reading content data is identified. The content types include articles, product descriptions, case descriptions, service introductions, or press releases. Based on the content type, the system calls the pre-built Schema.org vocabulary template to generate structured schema data in the corresponding format. The structured schema data includes the title, text summary, publication time, author information, entity attributes, and the relationships between them and the content data. While generating structured schema data, an llms.txt generator is invoked to collect site semantic descriptions, content scope, and reliable content paths from the structured reading results corresponding to all reading content data of the current website; The llms.txt generator generates a site-level llms.txt file based on the collected information. The llms.txt file contains at least the following: a list of website content scope, a list of key page URLs, a set of trusted content paths, a set of prohibited paths, content update description text, and site semantic description text. The generated llms.txt file is deployed to the website's root directory, where it, together with the structured schema data, constitutes a machine-readable dataset for large models.

3. The website content management method based on large-scale machine reading as described in claim 1, characterized in that, The step of adding structured schema data to the header or body area of ​​the website's front-end page via embedded code or related references, and based on the association index between the readable semantic description and the readable content data in the machine reading model, includes: Convert structured schema data into embedded code blocks in JSON-LD format, and insert the embedded code blocks into the script tags in the HTML header or body area of ​​the website front-end page, so that the website front-end page directly exposes machine-readable semantic descriptions when accessed by large models or artificial intelligence crawling programs. In the backend content library of the website content management system, create associated index records for the reading content data. The associated index records contain the unique identifier of the reading content data, the storage path of the structured schema data, and the file address of the readable semantic description output by the machine reading model.

4. The website content management method based on large-scale machine reading as described in claim 3, characterized in that, The establishment of information source management records for reading content data includes: Based on the associated index records, a source management record is generated for each piece of reading content data. The source management record includes the source URL, data type, credibility level, original collection time, and last verification time of the cited materials. The source description records output by the machine reading model are compared one by one with the source URLs of the cited materials in the source management records; When an external reference URL in the source description record does not appear in the source management record, the external reference URL will be automatically added as a new source to the source management record, and the trust level of the new source will be temporarily set to pending verification. When the trust level in the information source management record is lower than the preset verification trigger level, an information source verification prompt is generated. The information source verification prompt is used to notify website administrators or automatic verification programs to re-verify the information source.

5. The website content management method based on large-scale machine reading as described in claim 1, characterized in that, The process of associating and comparing the source management records with the source description records in the reading results, determining the credibility of the descriptions and the data version status of the reading content data based on the comparison results, and completing the corresponding operations for readable data in the machine reading model according to the request path parsed from the reading request includes: Each source URL in the source management record is matched with the external reference URL in the source description record. If a match is successful and the credibility level in the source management record is verified, the credibility of the description of the reading content data is marked as high credibility. If the match fails or the credibility level is pending verification, the credibility of the reading content data will be marked as pending review. Based on the original collection time and the last verification time in the information source management record, determine whether the reading content data has undergone content changes. If it has undergone changes, increment the version number in its data version status and retain the content snapshot of the previous version. If it has not undergone changes, maintain the original version status. When a read request is received from a large model or AI crawling program, the request path is parsed from the read request and matched against the set of prohibited read paths in the llms.txt site description file. If the match is successful, an access denied response is returned. If the requested path matches the trusted content path or key page list in the llms.txt site description file, the reading content data corresponding to the requested path, along with the associated description credibility and data version status, are obtained and returned to the machine reading model along with the structured schema data. This allows the machine reading model to perform readable data referencing or filtering operations based on the description credibility and data version status.

6. The website content management method based on large-scale machine reading as described in claim 1, characterized in that, The process involves repeating steps 1 through 4 within a fixed management cycle to generate a new site-level llms.txt file and update the structured schema data for each changed content data, including: At the end of each management cycle, steps 1 to 4 are executed again to obtain the changed reading content data for the current cycle from the website content management terminal. The changed reading content data includes content addition, content modification, or content deletion. For changes to the reading content data, the predefined machine reading model is called again to generate new structured reading results, and the corresponding structured schema data is regenerated based on the new structured reading results. The original structured schema data in the backend content library is replaced with the regenerated structured schema data. Collect the site semantic description, content scope, and trusted content path from the latest structured reading results of all reading content data on the website within the current management cycle. Call the llms.txt generator to regenerate a site-level llms.txt file and overwrite the old version file in the website root directory with the newly generated llms.txt file.

7. The website content management method based on large-scale machine reading as described in claim 6, characterized in that, The method of dynamically adjusting condition parameters within the management cycle based on reading feedback and content citation statistics of the reading website includes: Based on the statistical results of the frequency of content requests and citations on the website, dynamically adjust at least one of the following conditional parameters when executing steps 1 to 4 within the current management cycle: The verification trigger condition for the trust level in the source management record marks the source corresponding to the reading content data with zero citations in the previous period as a low-activity source, thereby reducing the automatic decay rate of the trust level. The sensitivity of content change detection is improved. When reading content data that was requested more frequently than the average in the previous period changes, the update response time of its structured schema data is shortened, and it is given priority to be regenerated in the current period. The rules for compiling the list of key pages in the llms.txt file are as follows: reading content data that ranks highly in both request frequency and citation count in the previous period is automatically added to the list of key pages, and content paths that have not been requested for a long time are removed from the list. The dynamically adjusted condition parameters will be used as the basis for repeatedly executing steps 1 to 4 within the current management cycle, and will be persistently stored in the configuration library of the website content management terminal.

8. A website content management system based on large-scale machine reading, characterized in that, include: The semantic parsing module is used to perform structured parsing of the semantics of the collected reading content data and the metadata associated with the content through the website backend interface, so as to obtain the reading results of the reading content data; The file generation module is used to automatically generate structured schema data and llms.txt site description files that match the content type of the reading results, based on the semantic tag set and content hierarchy graph in the reading results using a predefined machine reading model. The data deployment and information source archiving module is used to add structured schema data to the header or main area of ​​the website front-end page in the form of embedded code or related references, and to establish information source management records of reading content data based on the association index between readable semantic descriptions and reading content data in the machine reading model. The credibility verification and data response module is used to compare and associate the source management records with the source description records in the reading results. Based on the comparison results, it determines the credibility of the description and the data version status of the reading content data. Based on the request path parsed from the reading request, it completes the corresponding operation of the readable data in the machine reading model. The periodic iteration optimization module is used to repeatedly execute steps 1 to 4 within a fixed management cycle to generate a new site-level llms.txt file and update the structured schema data of each changed content data. Based on the reading feedback of website content and content citation statistics, the module dynamically adjusts the condition parameters within the management cycle.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 7.

Images