A community college multi-column resource hierarchical authorization publishing method and system

Abstract

The application discloses a community college multi-column resource hierarchical authorization publishing method and system, and particularly relates to the technical field of resource hierarchical authorization publishing, and comprises the following steps: obtaining text data, title data, source data, target community data, submission subject data, candidate column data and display terminal data of a resource to be published; performing community entity extraction, column semantic extraction, resource type extraction, cooperation relationship extraction and right ownership constraint extraction; and outputting a resource feature set; performing feature extraction on the resource to be published and mapping to a community college publishing knowledge graph; based on a candidate hypergraph, carrying out multi-path competitive reasoning, constraint backtracking resolution, hierarchical auditing and cross-terminal mirror image distribution to jointly determine a target column, an auditing link, a publishing range and a terminal display boundary of the resource.

Description

Technical Field

[0001] This invention relates to the field of resource hierarchical authorization and publishing technology, and more specifically, to a method and system for hierarchical authorization and publishing of multi-section resources in a community college. Background Technology

[0002] In community college resource management technology, the mainstream practice in the industry mainly addresses the issue of publishing content such as news announcements, course resources, event records, and special resources to the corresponding columns and completing basic review. This is usually achieved by setting fixed permissions according to account roles, configuring whitelists for publishing according to columns, or having the submitter manually select the community, target column, and scope of publication before entering the review process. For example, in scenarios where schools collaborate with entities such as prisons, street communities, and townships to jointly build multiple community colleges and simultaneously conduct course development, activity promotion, and resource display based on online learning platforms, mobile learning systems, smart portals, and independent sub-portals of each community, the platform must simultaneously meet hard constraints such as multiple columns coexisting, multi-terminal display, cross-column resource reuse, some resources being limited to display on community sub-portals, some resources being able to be simultaneously accessed on the school's main portal, and self-built course resources having clear ownership boundaries, all without interrupting service. Under this constraint, mainstream practices consistently reveal an observable and verifiable flaw: even though the same resource has passed the review, it will still repeatedly be placed in mismatched sections, spread on terminals or communities that should not be displayed, and the display boundaries of the same resource are inconsistent on different sections and terminals. The root cause is that the existing solution can only control the publishing action based on account permissions or manual options, and cannot automatically solve the correct adaptation relationship between resources and sections based on resource semantics, community academy relationship, section function and ownership constraints. The technical problem this application aims to solve is: how to automatically determine the target column, review process, publishing scope, and terminal display boundaries of resources in a community college multi-column resource publishing scenario, based on resource semantics, community college relationships, column attributes, and ownership constraints. Summary of the Invention

[0003] To overcome the aforementioned deficiencies in the prior art, embodiments of the present invention provide a method and system for hierarchical authorization and publishing of multi-section resources in community colleges. By extracting features of the resources to be published and mapping them to a knowledge graph for publishing in the community college, and conducting multi-path competitive reasoning, constraint backtracking and resolution, hierarchical review and cross-terminal mirror distribution based on the candidate hypergraph, the target section, review link, publishing scope and terminal display boundary of the resources are jointly determined to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for hierarchical authorization and publishing of multi-section resources in community colleges, comprising: S1. Obtain the main text data, title data, source data, target community data, submission entity data, candidate column data, and display terminal data of the resource to be published, and perform community entity extraction, column semantic extraction, resource type extraction, cooperation relationship extraction, and ownership constraint extraction, and output the resource feature set; S2. Obtain the knowledge graph published by the community college, map the resource feature set into resource nodes, community nodes, column nodes, cooperation entity nodes, ownership nodes and terminal nodes, and construct candidate hypergraphs based on column adaptation relationship, community affiliation relationship, cooperation constraint relationship, ownership restriction relationship and terminal display relationship; S3. Perform multi-path competition reasoning on the candidate hypergraph, calculate the column adaptation strength, community consistency strength, ownership satisfaction strength and terminal matching strength of each associated path, and output the candidate publishing path set and the candidate review path set. S4. Perform constraint backtracking and resolution on the candidate publishing path set and the candidate review path set. Perform priority suppression, edge weight decay and illegal path elimination on the super-edges corresponding to conflicting paths. Output the target publishing column set, the target review link set and the target display boundary set. S5. Perform hierarchical review according to the target review link set, and after the review is passed, perform targeted release, cross-terminal mirror distribution and release status write-back according to the target release column set and target display boundary set, and output hierarchical authorization release results.

[0005] In a preferred embodiment, S1 includes: S1-1. Perform field normalization and semantic segmentation on the text data, title data, source data, target community data, submission subject data, candidate column data, and display terminal data of the resources to be published, extract community name segment, column function segment, resource category segment, cooperation direction segment, and ownership limitation segment, and output the initial feature fragment set. S1-2. Perform cross-field co-occurrence association and position constraint aggregation on the initial feature fragment set, construct candidate semantic chains from feature fragments that point to the same community, the same column purpose, the same resource category, the same cooperation relationship, and the same ownership limit, and output the candidate semantic chain set. S1-3. Perform conflict fragment suppression and main semantic chain determination on the candidate semantic chain set, and determine the candidate semantic chains that are jointly verified by source data, target community data and candidate column data as the resource feature set.

[0006] In a preferred embodiment, S2 includes: S2-1. Input the resource feature set into the knowledge graph published by the community college, and perform node retrieval, alias merging and field alignment on the community features, column features, cooperation features, ownership features and terminal features in the resource feature set respectively, and output a candidate node set composed of resource nodes, community nodes, column nodes, cooperation entity nodes, ownership nodes and terminal nodes. S2-2. Perform relation expansion and coreference verification on the candidate node set in the knowledge graph published by the community college. Construct the associated node group that simultaneously has column adaptation relationship, community affiliation relationship, cooperation constraint relationship, ownership restriction relationship and terminal display relationship into candidate relation clusters, and output the candidate relation cluster set. S2-3. Perform hyperedge encapsulation and conflict edge suppression on the candidate relation cluster set, and construct the candidate relation clusters that meet the preset conditions for association strength and do not have node semantic conflicts into candidate hypergraphs.

[0007] In a preferred embodiment, S3 includes: S3-1. Perform path expansion on all reachable paths in the candidate hypergraph that start from resource nodes and end at column nodes, retain the associated paths that pass through community nodes, ownership nodes and terminal nodes and whose connection directions are consistent, and output the initial set of associated paths. S3-2. Perform itemized calculations on each associated path in the initial associated path set according to path segments. Determine the corresponding result from the resource node to the column node as the column adaptation strength, determine the ownership closure result from the resource node through the community node to the column node as the community consistency strength, determine the permission closure result from the resource node through the ownership node to the column node as the ownership satisfaction strength, and determine the display closure result from the resource node through the terminal node to the column node as the terminal matching strength. Output the associated path strength set. S3-3. Perform publishing path division and review path division on the set of associated path strengths. Write the associated paths that simultaneously close the column adaptation relationship, community affiliation relationship, ownership restriction relationship and terminal display relationship into the candidate publishing path set. Write the associated paths that simultaneously close the resource source relationship, cooperation constraint relationship and ownership confirmation relationship into the candidate review path set. Output the first round of candidate publishing path set and the first round of candidate review path set.

[0008] In a preferred embodiment, S3 further includes: S3-4. Perform consistency checks on the first round of candidate release path set and the first round of candidate review path set. Identify the associated paths in the candidate release path set that lack the support of the corresponding candidate review path as release mismatch paths, and identify the associated paths in the candidate review path set that cannot support the candidate release path to reach the column node and terminal node as review redundant paths. Then, perform backtracking correction on the associated path strength set based on the release mismatch path and the review redundant path, and output the corrected associated path strength set. S3-5. Repeatedly execute the release path division, review path division, and consistency check on the modified associated path strength set until the candidate release path set and candidate review path set output in the previous and next rounds are consistent. Then, output the candidate release path set and candidate review path set.

[0009] In a preferred embodiment, S4 includes: S4-1. Perform pairwise alignment between each candidate publishing path in the candidate publishing path set and each candidate review path in the candidate review path set. Compare the consistency results of community nodes, column nodes, ownership nodes, and terminal nodes segment by segment. Record the path segments with inconsistent nodes, inconsistent relationship directions, inconsistent relationship start and end points, and inconsistent path ends into the community conflict table, column conflict table, ownership conflict table, and terminal conflict table, respectively. Output the conflict path set and conflict superedge set. S4-2. Perform backtracking on each conflict hyperedge in the conflict hyperedge set, and backtrack along each conflict path to form the starting hyperedge, propagation hyperedge, and ending hyperedge of the conflict path. Construct a conflict cost sequence based on the number of conflict types of the starting hyperedge, the number of times the propagation hyperedge is reused, and the deviation of the endpoint of the ending hyperedge. Prioritize each conflict hyperedge in the order of number of conflict types first, then deviation of endpoint, and then number of reuses, and output the conflict hyperedge priority sequence. S4-3. Perform constraint backtracking and resolution on each conflicting superedge according to the priority sequence of conflicting superedges. For low-priority conflicting superedges in the same conflicting path, deduct one corresponding relationship count. For conflicting superedges that no longer simultaneously connect resource nodes, community nodes, column nodes, ownership nodes, and terminal nodes after deduction, perform illegal path elimination. Write the remaining candidate review paths back to the candidate publishing path set to correct the edge weights of the corresponding paths. Output the corrected publishing path set and the corrected review path set.

[0010] In a preferred embodiment, S4 further includes: S4-4. Repeatedly execute conflict table generation, priority sorting and constraint backtracking resolution on the revised release path set and the revised review path set until only path groups with the same endpoint and consistent nodes are retained in the conflict path set. Then, determine the column nodes corresponding to each path group at the time of stopping as the target release column set, determine the corresponding candidate review path as the target review link set, and determine the common reachability of the terminal nodes in each path group as the target display boundary set. S4-5. Perform closure verification on the target publication column set, target review link set, and target display boundary set. The result that each target publication column corresponds to at least one target review link and the terminal node corresponding to each target review link is included in the target display boundary set is determined as the constraint backtracking resolution result.

[0011] In a preferred embodiment, S5 includes: S5-1. Write the resources to be published, the target review link set, the target publishing column set, and the target display boundary set into the review execution sequence. Load the corresponding review fields, previous review results, and current fields to be reviewed one by one according to the connection order of each review node in the target review link set, and output the hierarchical review task sequence. S5-2. Perform closed-loop auditing on the hierarchical audit task sequence. At each audit node, read the current field to be audited and the previous audit results, and verify the consistency of resource source, column ownership, ownership direction and display boundary item by item. Write the audit nodes with consistent verification into the pass link and write the audit nodes with inconsistent verification into the return link. Then write the inconsistent fields in the return link back to the previous audit node to form the error correction result, and output the set of audit pass links and the set of audit return links. S5-3. Perform a publish reachability calculation on the approved link set. Align each approved link with the target publishing category set and the target display boundary set one by one. Keep the links that simultaneously satisfy the category reach closure, terminal reach closure, and boundary containment closure as valid publishing links. Move the links that cannot be closed at the same time into the review return link set and perform link closure review again until the valid publishing links obtained in two consecutive rounds are consistent. Output the target publishing link set.

[0012] In a preferred embodiment, S5 further includes: S5-4. Perform targeted publishing and cross-terminal mirror distribution according to the target publishing link set. Write the resources to be published into the column storage location corresponding to the target publishing column set. Generate a mirror distribution table according to the correspondence between each terminal node and column node in the target display boundary set. Then copy the column index, boundary marker and review marker to the corresponding terminal one by one according to the mirror distribution table. Output the cross-terminal publishing result set. S5-5. Perform a write-back of the publishing status for the cross-terminal publishing result set. Summarize the write results, column results and boundary results of each terminal into a publishing status record and write it back to the resource node, column node and terminal node corresponding to the resource to be published. At the same time, write the published relationship between the resource node, column node and terminal node into the community college publishing knowledge graph and output the hierarchical authorization publishing results.

[0013] A community college multi-section resource hierarchical authorization publishing system, the system comprising a feature extraction module, a hypergraph construction module, an inference module, a conflict resolution module, and an authorization publishing module: The feature extraction module is used to obtain the text data, title data, source data, target community data, submission entity data, candidate column data, and display terminal data of the resources to be published, and to perform community entity extraction, column semantic extraction, resource type extraction, cooperation relationship extraction, and ownership constraint extraction, and output a set of resource features; The hypergraph construction module is used to obtain the knowledge graph published by the community college, map the resource feature set into resource nodes, community nodes, column nodes, cooperation entity nodes, ownership nodes and terminal nodes, and construct candidate hypergraphs based on column adaptation relationship, community affiliation relationship, cooperation constraint relationship, ownership restriction relationship and terminal display relationship; The reasoning module is used to perform multi-path competition reasoning on candidate hypergraphs, calculate the column adaptation strength, community consistency strength, ownership satisfaction strength and terminal matching strength of each associated path, and output the candidate publishing path set and the candidate review path set; The conflict resolution module is used to perform constraint backtracking resolution on the candidate publishing path set and the candidate review path set, and to perform priority suppression, edge weight decay and illegal path elimination on the super edges corresponding to conflict paths. It outputs the target publishing column set, the target review link set and the target display boundary set. The authorization and release module is used to perform hierarchical review according to the target review link set, and after the review is passed, it performs targeted release, cross-terminal mirror distribution and release status write-back according to the target release column set and target display boundary set, and outputs hierarchical authorization and release results.

[0014] The technical effects and advantages of this invention are as follows: This solution extracts features from resources to be published, maps knowledge graphs, performs multi-path reasoning, and resolves constraints by backtracking and eliminating constraints. It can jointly transform resource semantics, community relationships, column attributes, and ownership constraints into target publishing columns, review links, publishing scope, and terminal display boundaries, thereby relatively suppressing the problems of misdirected resources, cross-scope diffusion, and inconsistent display boundaries across multiple terminals. By unifying heterogeneous information from titles, text, sources, target communities, candidate columns, and display terminals into a resource feature set, and completing semantic convergence through cross-field co-occurrence association and main semantic chain determination, the consistency and computability of resource semantic expression can be improved, thereby providing a more stable input basis for subsequent node mapping and relationship expansion. By mapping the set of resource features to resource nodes, community nodes, column nodes, ownership nodes and terminal nodes, and constructing candidate hypergraphs based on column adaptation relationships, community affiliation relationships, ownership restriction relationships and terminal display relationships, the originally scattered publishing constraints can be uniformly incorporated into the same graph structure for processing, thereby relatively improving the relationship organization capabilities in the parallel scenarios of multiple columns, multiple communities and multiple terminals. Perform multi-path competitive reasoning on candidate hypergraphs and calculate column adaptation strength, community consistency strength, ownership satisfaction strength and terminal matching strength respectively. This can screen out candidate paths that simultaneously meet the publishing and review conditions from multiple reachable paths, thereby relatively reducing the path deviation caused by relying on a single field or a single rule for judgment. By performing conflict table generation, priority sorting, and constraint backtracking on the candidate release path set and candidate review path set, community conflicts, column conflicts, ownership conflicts, and terminal conflicts can be located and reduced segment by segment and round by round, thereby relatively improving the convergence and consistency of the target release column set, target review link set, and target display boundary set. Attached Figure Description

[0015] Fig. 1 This is a flowchart outlining the method steps of the present invention; Fig. 2 This is a schematic diagram of the system module structure of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Refer to the instruction manual appendix Figs. 1-2 The present invention provides a method for hierarchical authorization and publishing of multi-section resources in community colleges, comprising: S1. Obtain the main text data, title data, source data, target community data, submission entity data, candidate column data, and display terminal data of the resource to be published, and perform community entity extraction, column semantic extraction, resource type extraction, cooperation relationship extraction, and ownership constraint extraction, and output the resource feature set; In this implementation, the purpose of S1 is to organize the title information, body information, source information, community targeting information, submission entity information, column targeting information, and terminal targeting information of the resources to be published into a resource feature set that can be directly used for subsequent graph mapping. The processing approach is as follows: first, unify the definitions of each field, then extract semantic fragments with clear targeting, then form candidate semantic chains through cross-field association, and finally eliminate conflicts and determine the main semantic chain. This implementation process includes the following steps: The purpose of S1-1 is to eliminate differences in field representation and form a calculable basic fragment. The input consists of the main text data, title data, source data, target community data, submission body data, candidate column data, and display terminal data of the resource to be published, among which title data, main text data, target community data, and submission body data are mandatory fields. The processing actions are as follows: For the main text data and title data, character normalization is performed, converting full-width characters to half-width characters, unifying simplified and traditional characters, unifying dates to year-month-day format, and unifying punctuation to standard separators; For the source data, target community data, submission body data, candidate column data, and display terminal data, encoding normalization is performed, replacing the community name with the standard community name in the community master file, replacing the column name with the column identifier in the column master file, and replacing the terminal name with the terminal category code in the terminal master file. Then, perform semantic segmentation based on field boundaries and sentence boundaries to extract community name segments, column function segments, resource category segments, cooperation guidance segments, and ownership restriction segments; output an initial feature fragment set, and write the fragment type, source field, start and end position, and standard code into the fragment cache table for the next step; when the source data is empty, it is filled back by the institution affiliation field corresponding to the submitted main data; when the candidate column data is empty, the column function segments in the title data and body data are retained to proceed to the next step; when the display terminal data is empty, the default terminal category code corresponding to the candidate column is read and added to the fragment cache table; when no community name segment is extracted from the body data, the standard community name corresponding to the target community data is written into the fragment cache table. The purpose of S1-2 is to combine interrelated feature fragments into candidate semantic chains. The input consists of an initial set of feature fragments and a fragment cache table. The processing steps are as follows: using the community name fragment corresponding to the target community data as the main anchor point, the column function fragments and resource category fragments that appear together with the main anchor point in the title data are preferentially associated. Then, the cooperation pointing fragments and ownership limiting fragments that are in the same sentence or two adjacent sentences of the main anchor point in the body data are added to the same association group. For each association group, position constraint aggregation is performed in the order of community name fragments, column function fragments, resource category fragments, cooperation pointing fragments, and ownership limiting fragments. When there are multiple fragments of the same type, the title field fragments are retained first, followed by the source field and target community field fragments, and then the fragments in the body text that are closest to the main anchor point are retained. Association groups containing community name segments, column function segments, and resource category segments are constructed as candidate semantic chains. A set of candidate semantic chains is output, and the chain identifier, main community identifier, column purpose identifier, resource category identifier, cooperation relationship identifier, and ownership restriction identifier are written to the semantic chain cache table for subsequent reading. When multiple community name segments appear for the same resource, the community name segment corresponding to the target community data is used as the main community segment, and the remaining community name segments are attached and saved as associated community segments. When a cooperation pointing segment is missing, the cooperation relationship identifier is supplemented from the institutional relationship table corresponding to the source data and the submitting entity data. When an ownership restriction segment is missing, the ownership restriction identifier is supplemented from the resource ownership table. Association groups that simultaneously lack both column function segments and candidate column data are not included in the candidate semantic chain set. The purpose of S1-3 is to determine a unique set of usable resource features from the candidate semantic chain set. The inputs are the candidate semantic chain set, the semantic chain cache table, the source data, the target community data, and the candidate column data. The processing actions are as follows: compare the main community identifier, column purpose identifier, resource category identifier, cooperation relationship identifier, and ownership restriction identifier of each candidate semantic chain. If the main community identifier is inconsistent with the target community data, the corresponding main community identifier is downgraded to the associated community identifier. If the column purpose identifier is inconsistent with the candidate column data but the resource category identifier is consistent, retain the resource category identifier and record the column conflict. If the ownership restriction identifier is inconsistent with the resource ownership table corresponding to the source data, replace it with the ownership restriction identifier corresponding to the source data. After suppressing conflicting segments, candidate semantic chains are filtered in the order of source data verification, target community data verification, and candidate column data verification. Candidate semantic chains that pass all three verifications are retained as the main semantic chain, and the main community identifier, column purpose identifier, resource category identifier, cooperation relationship identifier, ownership restriction identifier, and terminal category identifier are combined into a resource feature set. The resource feature set is output and written into the resource feature table for S2 to call. When there are multiple candidate semantic chains that pass the verifications, the candidate semantic chain with the most segments in the title field is retained first. If the number is the same, the candidate semantic chain whose column purpose segment in the text is closest to the main community segment is retained. If they still cannot be distinguished, they are written into the manual review queue. When no candidate semantic chain passes all three verifications, the candidate semantic chain that passes both source data verification and target community data verification is determined as the temporary main semantic chain, and a column review mark is written into the resource feature table. Through the above processing, the original resources to be published can be organized into a set of resource features with unified field definitions, clear community orientation, clear column purpose, clear resource category, clear cooperation relationship, and clear ownership, providing stable input for subsequent node mapping and candidate hypergraph construction; In practical applications: When a resource title is "Reading Activity of a Community College", the text states "Initiated by the Community College in conjunction with the Street Office", the source field is the school's continuing education department, and the candidate column field and display terminal field are empty, the system will unify "Community College" into the standard community name in the community master file, extract "Reading Activity" as an activity-type resource category segment, extract "Initiated by the Street Office in conjunction with the Street Office" as a cooperation-oriented segment, extract the resource ownership information corresponding to the source field as an ownership restriction identifier, and then combine it with the target community data to complete the correction, ultimately forming a set of resource features that can be directly used in subsequent steps.

[0018] S2. Obtain the knowledge graph published by the community college, map the resource feature set into resource nodes, community nodes, column nodes, cooperation entity nodes, ownership nodes and terminal nodes, and construct candidate hypergraphs based on column adaptation relationship, community affiliation relationship, cooperation constraint relationship, ownership restriction relationship and terminal display relationship; In this implementation, the purpose of S2 is to map the resource feature set output by S1 to standard nodes in the community college publishing knowledge graph, and to expand the relationships directly related to the publishing determination between the nodes, further forming a candidate hypergraph that can be called for subsequent path reasoning. The processing idea is as follows: first, based on the resource feature set, node retrieval, alias merging, and field alignment are completed in the community college publishing knowledge graph; then, relationship expansion and coreference verification are performed on the retrieved candidate node set; finally, the candidate relationship clusters that meet the publishing association requirements are encapsulated into a candidate hypergraph. This implementation process includes the following steps: The purpose of S2-1 is to stably map various features in the resource feature set to standard nodes in the community college publishing knowledge graph. The inputs are the resource feature set and the community college publishing knowledge graph, in which the community college publishing knowledge graph pre-stores resource nodes, community nodes, column nodes, cooperative entity nodes, ownership nodes, and terminal nodes, as well as column adaptation relationships, community affiliation relationships, cooperative constraint relationships, ownership restriction relationships, and terminal display relationships. The processing actions are as follows: first, read the main community identifier, column purpose identifier, resource category identifier, cooperative relationship identifier, ownership restriction identifier, and terminal category identifier from the resource feature set; then, perform node retrieval in the corresponding master index for the main community identifier, column purpose identifier, cooperative relationship identifier, ownership restriction identifier, and terminal category identifier, with the retrieval order being standard name exact matching, alias table matching, and field code matching. When multiple nodes with the same name are found, the target community identifier, source identifier, and column purpose identifier in the resource feature set are matched item by item, and nodes with consistent field matching are retained. Then, the resource to be published itself is registered as a resource node, and the community node, column node, cooperative entity node, ownership node, and terminal node determined by the search are written together with the resource node into the candidate node cache table. The candidate node set consisting of resource node, community node, column node, cooperative entity node, ownership node, and terminal node is output for S2-2 to read. The abnormal or missing handling is as follows: when no corresponding node is found for a certain feature, the alias merging table of the feature is read first, the alias is replaced, and the search is performed again. If no node is found, the feature is written into the node to be supplemented table and the subsequent relationship expansion corresponding to the feature is stopped. When multiple nodes with consistent field matching are found for the same feature, the node with the same main community identifier in the resource feature set is retained as the main node, and the remaining nodes are written into the candidate node table. The purpose of S2-2 is to filter out the related node groups that can jointly support the release judgment from the candidate node set. The inputs are the candidate node set, the candidate node cache table, and the relationship table in the community college release knowledge graph. The processing actions are as follows: starting from the resource node, perform relationship expansion one by one for the community affiliation relationship between the resource node and the community node, the column adaptation relationship between the resource node and the column node, the cooperation constraint relationship between the resource node and the cooperating entity node, the ownership restriction relationship between the resource node and the ownership node, and the terminal display relationship between the resource node and the terminal node, and record the starting node, ending node, relationship type, and relationship source of each relationship. After the relationship expansion is completed, perform common reference verification. The verification rules are: the community node and the main community identifier point to the same community master file, the column node and the column purpose identifier point to the same column configuration item, the cooperating entity node and the cooperation relationship identifier point to the same cooperating entity record, the ownership node and the ownership restriction identifier point to the same resource affiliation record, and the terminal node and the terminal category identifier point to the same terminal configuration item. Only the associated node groups that have all five types of relationships and whose nodes have all passed the co-reference check are retained as candidate relation clusters; the candidate relation cluster set is output, and the cluster identifier, node composition, relation composition and co-reference check result of each candidate relation cluster are written into the relation cluster cache table for S2-3 to read; the abnormal or missing handling is as follows: when a certain associated node group is missing a cooperating subject node, the cooperating subject node is added from the cooperation ledger corresponding to the source identifier and then re-verified; when a certain associated node group is missing a terminal node, the terminal node is added from the default terminal configuration corresponding to the column node and then re-verified; if the associated node group still cannot satisfy the existence of all five types of relationships after the addition, it will not enter the candidate relation cluster set. The purpose of S2-3 is to transform candidate relation clusters into candidate hypergraphs with a unified structure and exclude relation clusters with semantic conflicts. The inputs are the set of candidate relation clusters, the relation cluster cache table, and the node semantic table in the knowledge graph published by the community college. The processing actions are as follows: perform hyperedge encapsulation on each candidate relation cluster, encapsulating the resource nodes, community nodes, column nodes, cooperative entity nodes, ownership nodes, and terminal nodes in the same candidate relation cluster, as well as the corresponding column adaptation relationship, community affiliation relationship, cooperative constraint relationship, ownership restriction relationship, and terminal display relationship into a hyperedge; then perform conflict edge suppression, comparing the community nodes, column nodes, ownership nodes, and terminal nodes in each candidate relation cluster to see if there are semantic conflicts. The semantic conflict judgment rules are: the community to which the community node belongs is inconsistent with the main community identifier; the column group to which the column node belongs is inconsistent with the column purpose identifier; the allowed scope recorded by the ownership node is inconsistent with the usage scope of the column node or terminal node; and the terminal node does not belong to the set of displayable terminals corresponding to the column node. For candidate relation clusters that do not have semantic conflicts and simultaneously contain resource nodes, community nodes, column nodes, ownership nodes, and terminal nodes, the hyperedge encapsulation results are retained. For candidate relation clusters with semantic conflicts, conflict edges are suppressed and written to the conflict relation record table. The candidate hypergraph is output, and the retained hyperedges are written to the hypergraph cache table for S3 to read. The abnormal or missing handling is as follows: when the same resource corresponds to multiple non-conflicting candidate relation clusters, all of them are retained and together form a candidate hypergraph. If all candidate relation clusters have semantic conflicts, the candidate relation cluster with the most consistent items with the main community identifier, column purpose identifier, and ownership restriction identifier is temporarily stored as a relation cluster to be reviewed and written to the manual review queue. Through the above processing, the resource feature set can be stably mapped to the standard nodes in the knowledge graph published by the community college, and a candidate hypergraph with complete relation types, clear node pointers and can be directly used for subsequent path reasoning can be formed, avoiding distortion of subsequent path development due to mixed use of node aliases, missing relations or semantic conflicts of nodes. In practical applications: When a resource feature set contains the main community identifier of "a certain community college", the purpose identifier of the "activity promotion" column, the resource category identifier of "activity-type resources", the cooperative relationship identifier of "street co-construction", the ownership restriction identifier of "school publishing authority", and the terminal category identifier of "mobile terminal", the system first retrieves the corresponding nodes in the community master file, column master file, cooperation ledger, resource ownership table and terminal configuration table respectively. Then, it completes the field alignment with the main community identifier and source identifier. Subsequently, it expands the five types of relationships between the resource node and the community node, column node, cooperative entity node, ownership node and terminal node, and retains only the associated node group that has all five types of relationships and does not have inconsistencies in community, column group, ownership scope and terminal scope. Finally, it forms the candidate hypergraph corresponding to the resource, which can be directly called in subsequent multi-path competitive reasoning.

[0019] S3. Perform multi-path competition reasoning on the candidate hypergraph, calculate the column adaptation strength, community consistency strength, ownership satisfaction strength and terminal matching strength of each associated path, and output the candidate publishing path set and the candidate review path set. In this implementation, the purpose of S3 is to extract effective paths that can truly support resource publishing and review from the candidate hypergraph formed in S2. The processing approach is as follows: First, expand the reachable paths in the candidate hypergraph, starting from resource nodes and ending at category nodes. Then, calculate the category adaptation strength, community consistency strength, ownership satisfaction strength, and terminal matching strength for each reachable path. Based on this, divide the candidate publishing paths and candidate review paths, and eliminate the mismatch between publishing paths and review paths through consistency checks and backtracking corrections. Finally, output a stable set of candidate publishing paths and a set of candidate review paths. This implementation process includes the following steps: The purpose of S3-1 is to filter out the associated paths that meet the requirements of the publishing decision structure from the candidate hypergraph. The input is the candidate hypergraph and the hyperedge records in the hypergraph cache table. The processing action is as follows: taking the resource node as a fixed starting point and the column node as a fixed ending point, perform path expansion on each hyperedge in the candidate hypergraph, and connect the relationship starting point and relationship ending point in the hyperedge segment by segment to form a reachable path. When expanding the path, the same node in the same path will not be entered twice. The path must pass through the community node, ownership node and terminal node at the same time, and the direction of each relationship connection should be converged from the resource node to the column node. When a path is entered again... When a node has been visited, the path expansion stops. When a path reaches a category node and already contains a community node, ownership node, and terminal node, the path extension stops. The initial set of associated paths is output, and the path identifier, node sequence, relationship sequence, and endpoint category identifier of each associated path are written into the path cache table for the next step. The exception or missing handling is as follows: if a path is missing a community node, ownership node, or terminal node, it is directly removed from the initial set of associated paths. If there is no reachable path in the candidate hypergraph that meets the above structural requirements, the corresponding resource is written into the path missing record table and the subsequent S3 processing stops. The purpose of S3-2 is to form a unified strength determination result for each associated path in the initial associated path set. The inputs are the initial associated path set, the path cache table, the resource feature set, and the relationship records in the knowledge graph published by the community college. The processing actions are as follows: for each associated path, perform itemized calculations by path segment. The relationship segment from the resource node to the column node is used to calculate the column adaptation strength. The calculation rule is that when the resource category identifier and the column purpose identifier are consistent and the scope of use of the column node does not exclude the resource category, it is determined to be completely closed. When the resource category identifier and the column purpose identifier belong to the same column group but are not completely consistent, it is determined to be partially closed. Otherwise, it is determined to be unclosed. The relationship segment from the resource node through the community node to the column node is used to calculate the community consistency strength. The calculation rule is that when the main community identifier in the resource feature set, the community node identifier in the path, and the community scope corresponding to the column node are consistent, it is determined to be completely closed. When the community scope corresponding to the column node includes the main community identifier, it is determined to be partially closed. Otherwise, it is determined to be unclosed. The relationship segment from the resource node to the ownership node and then to the category node is used to calculate the ownership satisfaction strength. The calculation rule is that if the allowed publishing range recorded by the ownership node covers both the current category node and the current resource category, it is considered a complete closure; if it covers only one of them, it is considered a partial closure; otherwise, it is considered an open closure. The relationship segment from the resource node to the terminal node and then to the category node is used to calculate the terminal matching strength. The calculation rule is that if the terminal node belongs to the set of displayable terminals corresponding to the category node and the ownership node does not exclude the terminal, it is considered a complete closure; if it meets only one of the conditions, it is considered a partial closure; otherwise, it is considered an open closure. The set of associated path strengths is output, and the four strength results corresponding to each associated path are written into the path strength table for the next step of reading. The handling of anomalies or missing fields is as follows: if a relationship segment of a certain path lacks a field used for judgment, it is first supplemented from the resource feature set and the corresponding master file of the knowledge graph. If it still cannot be judged after supplementation, the strength of that item is directly recorded as an open closure. The purpose of S3-3 is to classify reachable paths into two categories—publishing and review—based on the set of associated path strengths. The inputs are the set of associated path strengths, the path strength table, and the resource source relationships, cooperation constraint relationships, and ownership confirmation relationships in the community college's publishing knowledge graph. The processing involves: performing publishing path segmentation and review path segmentation for each associated path; adding associated paths with completely closed categories in terms of column adaptation strength, community consistency strength, ownership satisfaction strength, and terminal matching strength to the candidate publishing path set; simultaneously checking whether the same associated path possesses resource source relationships between resource nodes and source records, cooperation constraint relationships between resource nodes and cooperating entity nodes, and ownership confirmation relationships between resource nodes and ownership nodes. If all three types of relationships exist simultaneously, the associated path is added to the candidate review path set. If the same associated path simultaneously meets the conditions for writing both the candidate release path and the candidate review path, then the same path identifier is retained in both the candidate release path set and the candidate review path set, and a corresponding relationship record is established. The first round of candidate release path set and the first round of candidate review path set are output, and the path identifier correspondence is written into the path correspondence table for the next step of reading. The handling of anomalies or missing items is as follows: if a path has some closures but no open closures among the four strengths, it is not written into the first round of candidate release path set but into the backup path table. If a path lacks any of the relationships among resource source relationship, cooperation constraint relationship, or ownership confirmation relationship, it is not written into the first round of candidate review path set. The purpose of S3-4 is to correct the support deviation between the first round of candidate release path set and the first round of candidate review path set. The input quantities are the first round of candidate release path set, the first round of candidate review path set, the path correspondence table, and the associated path strength set. The processing actions are as follows: read the first round of candidate release paths one by one, check whether there is a candidate review path with the same resource node, the same community node, and the same ownership node corresponding to it. If not, the candidate release path is identified as a release mismatch path. Read the first round of candidate review paths one by one, check whether it can support the corresponding candidate release path to reach the same column node and terminal node. If it cannot support it, the candidate review path is identified as a review redundancy path. After marking is completed, backtracking correction is performed. For mismatched release paths, the path sequence is backtracked to the relationship segment between resource nodes and ownership nodes. If there is an approval relationship segment in the first round of candidate review paths that has the same resource node, ownership node, and community node, then the corresponding relationship segment in the mismatched release path is replaced with this approval relationship segment. After replacement, the ownership satisfaction strength and community consistency strength of the path are recalculated. For redundant review paths, the relationship segments that cannot support the arrival of column nodes and terminal nodes are deleted, and the path is re-checked to see if it still has resource source relationship, cooperation constraint relationship, and ownership confirmation relationship. If it has them, it is retained; if not, it is removed. The set of corrected associated path strengths is output, and the corrected path strength results are written to the path strength table for the next step. Abnormal or missing handling is as follows: if the mismatched release path still cannot form a completely closed four-item strength result after replacement, it is deleted from the candidate release path set. If the redundant review path no longer has the review support structure after the relationship segment is deleted, it is deleted from the candidate review path set. The purpose of S3-5 is to converge the candidate release path set and candidate review path set to a stable result through repeated partitioning and verification. The inputs are the corrected associated path strength set, the path strength table, and the corrected path correspondence table. The processing actions are: repeatedly performing the release path partitioning and review path partitioning in S3-3 on the corrected associated path strength set, and repeatedly performing the consistency verification and backtracking correction in S3-4 on the newly obtained candidate release path set and candidate review path set. After each round of processing, the candidate release path set and candidate review path set output in this round are compared with the corresponding sets in the previous round. The process involves comparing the path identifiers, corresponding column nodes, and corresponding terminal nodes in two rounds. If the path identifiers, corresponding column nodes, and corresponding terminal nodes are completely identical, the outputs of the two rounds are considered to be consistent, and processing is stopped. If they are inconsistent, the result of this round is used as the input for the next round. The process outputs a set of candidate publishing paths and a set of candidate review paths, and writes the paths that are ultimately retained into the candidate path table for S4 to read. The process handles anomalies or missing paths as follows: if there are still new or deleted paths after three consecutive rounds of processing, the paths that appear in the last two rounds are retained as the set of candidate publishing paths and the set of candidate review paths, and the remaining paths are written into the path table to be conflicted and resolved for S4 to process first. Through the above processing, candidate publishing paths that simultaneously meet the requirements of column purpose, community scope, ownership scope and terminal scope can be solved from the candidate hypergraph, and candidate review paths that can support the publishing judgment can be solved at the same time. This ensures that the subsequent constraint backtracking and resolution is based on the path results that are structurally complete, have closed relationships and are mutually verified, avoiding mis-publishing of columns, unauthorized publishing and terminal boundary mismatch caused by judging only a single path or a single relationship. In practical applications: When a resource is associated with both "Community Activity Promotion Section" and "Public Information Section" reachable paths, the system first expands both paths and checks whether they both pass through community nodes, ownership nodes, and terminal nodes. Then, it calculates the section adaptation strength between the activity resource and the two types of sections, the community consistency strength between the target community and the section scope, the ownership satisfaction strength between the school's publishing permissions and the section's usage scope, and the terminal matching strength between the mobile terminal and the section's terminal configuration. If all four strengths of the path corresponding to "Community Activity Promotion Section" are completely closed, while the path corresponding to "Public Information Section" only has a partially closed section adaptation strength, then the former enters the candidate publishing path set. At the same time, the system checks whether the former has resource source relationships, street co-construction cooperation constraints, and ownership confirmation relationships corresponding to the school's publishing permissions. If they are, it is written into the candidate review path set. If the candidate review path cannot support the path to reach a certain terminal, the corresponding terminal relationship segment is deleted and the terminal matching strength is recalculated until the candidate publishing path set and the candidate review path set are consistent, and then a stable result is output.

[0020] S4. Perform constraint backtracking and resolution on the candidate publishing path set and the candidate review path set. Perform priority suppression, edge weight decay and illegal path elimination on the super-edges corresponding to conflicting paths. Output the target publishing column set, the target review link set and the target display boundary set. In this implementation, the purpose of S4 is to perform conflict localization, priority ranking, and backtracking resolution on the candidate publishing path set and candidate review path set output by S3. This aims to retain a stable path group that simultaneously meets the requirements of community scope, column scope, ownership scope, and terminal scope when multiple feasible paths coexist. Based on this, the target publishing column set, target review link set, and target display boundary set are determined. The processing approach is as follows: first, the candidate publishing path and candidate review path are aligned segment by segment to generate a conflict path set and a conflict super-edge set; then, backtracking and priority ranking are performed on the conflict super-edges; subsequently, constraint backtracking resolution is performed according to priority order; finally, a unique and valid resolution result is output through multiple rounds of repeated resolution and closure verification. This implementation process includes the following steps: The purpose of S4-1 is to accurately locate inconsistent path segments in the candidate publishing path set and the candidate review path set as processable conflict objects. The inputs are the candidate publishing path set, the candidate review path set, the candidate path table, and the hyperedge records in the hypergraph cache table. The processing actions are as follows: each candidate publishing path and each candidate review path are paired up according to the same conditions of resource nodes, and then the consistency results of community nodes, column nodes, ownership nodes, and terminal nodes are compared segment by segment according to the path segment order. At the same time, the connection direction, relationship start point, relationship end point, and path end point of the corresponding relationship are checked. During the comparison process, path segments with inconsistent community nodes are recorded in the community conflict table, path segments with inconsistent column nodes or different column paths at the path end point are recorded in the column conflict table, path segments with inconsistent ownership nodes are recorded in the ownership conflict table, path segments with inconsistent terminal nodes are recorded in the terminal conflict table, and path segments with inconsistent relationship directions and inconsistent relationship start and end points are written into the conflict table to which the corresponding node belongs. Then, the path that hits at least one type of conflict table is identified as a conflict path, and the corresponding superedge in the conflict path is extracted as a conflict superedge; the set of conflict paths and the set of conflict superedges are output, and the conflict path identifier, conflict superedge identifier, conflict type, conflict position, corresponding column node and corresponding terminal node are written into the conflict record table for S4-2 to read; the abnormal or missing handling is as follows: when a candidate publishing path does not find a candidate review path with the same resource node, the candidate publishing path is directly recorded into the column conflict table and the ownership conflict table; when a candidate review path does not find a candidate publishing path with the same resource node, the candidate review path is directly recorded into the terminal conflict table and the column conflict table. The purpose of S4-2 is to establish an executable resolution sequence for each conflicting superedge. The inputs are a set of conflicting superedges, a set of conflicting paths, a conflict record table, and a candidate path table. The processing steps are as follows: for each conflicting superedge, perform reverse positioning along the corresponding conflicting path; determine the superedge corresponding to the location where the first conflict occurs as the starting superedge; determine the superedge located between the starting and ending superedges and participating in conflict propagation as the propagation superedge; and determine the last superedge that causes the path endpoint to deviate from the target section or target terminal as the ending superedge. Then, calculate the number of conflict types and propagation... The number of times a superedge is reused and the deviation amount from the endpoint of the terminated superedge are determined by the number of conflict types that the starting superedge matches in the community conflict table, column conflict table, ownership conflict table, and terminal conflict table. The number of reuses is the number of paths in the current candidate publishing path set and candidate review path set that contain the same propagation superedge. The deviation amount from the endpoint is determined by the level of deviation between the endpoint of the terminated superedge and the target column target identification. The same column is recorded as a level 0 deviation, different columns in the same column group are recorded as a level 1 deviation, and different column groups are recorded as a level 2 deviation. If both exceed the target community range, an additional level of deviation is added. After obtaining the above results, prioritize each conflicting hyperedge in the following order: number of conflict types, followed by endpoint deviation, and then reuse count. A higher number of conflict types results in higher priority. If the number of conflict types is the same, a higher endpoint deviation results in higher priority. If the endpoint deviation is the same, fewer reuse counts result in higher priority. Output the conflicting hyperedge priority sequence and write the priority number to the conflict record table for S4-3 to read. For exceptions or missing values: if a conflicting hyperedge cannot be located as a termination hyperedge, the last hyperedge in the path containing that conflicting hyperedge is taken as the termination hyperedge. If a propagation hyperedge is not reused in other paths, its reuse count is recorded as 1. The purpose of S4-3 is to gradually reduce low-priority conflict relationships and correct path results according to the conflict superedge priority sequence. The inputs are the conflict superedge priority sequence, the conflict superedge set, the candidate publishing path set, the candidate review path set, and the conflict record table. The processing actions are as follows: read the conflict superedges one by one according to the conflict superedge priority sequence, deduct one corresponding relationship count for the lower priority conflict superedge in the same conflict path, where the initial corresponding relationship count for each relationship segment in each conflict superedge is 1, and the relationship segment that is reduced to 0 after deduction is determined as the invalid relationship segment; then check whether the conflict superedge can still connect resource nodes, community nodes, column nodes, ownership nodes, and terminal nodes simultaneously after deduction. If it cannot be connected simultaneously, the path corresponding to the conflict superedge is removed from the candidate publishing path set or the candidate review path set and recorded as an illegal path removal result. For the retained candidate review paths, the review relationship segments corresponding to their community nodes, ownership nodes, and terminal nodes are written back to the candidate release path set to correct the edge weights of the corresponding candidate release paths. The correction rule is as follows: when the relationship segment in the candidate release path is consistent with the relationship segment of the retained candidate review path, the edge weight of the relationship segment is increased by 1; when they are inconsistent, the original edge weight is maintained; when there is no corresponding relationship segment, a new relationship segment is added and an initial edge weight of 1 is assigned. The corrected release path set and the corrected review path set are output, and the corrected paths and edge weight results are written to the corrected path table for S4-4 to read. The abnormal or missing handling is as follows: when all conflicting super edges in the same conflicting path are judged to be of low priority, the conflicting super edges that are consistent with the candidate column field of the endpoint column node are retained first, and the corresponding relationship count is deducted for the rest. If the candidate review path is written back and duplicate relationship segments appear in a candidate release path, only the relationship segments with larger edge weights are retained. The purpose of S4-4 is to converge the path results into a stable path group with consistent nodes and endpoints through multiple rounds of conflict backtracking and resolution. The inputs are the set of revised published paths, the set of revised reviewed paths, the revised path table, and the conflict record table. The processing actions are as follows: repeatedly execute the conflict table generation in S4-1, the priority sorting in S4-2, and the constraint backtracking and resolution in S4-3 on the set of revised published paths and the set of revised reviewed paths, and re-count the path endpoints and node composition in the conflict path set after each round. When only path groups with the same endpoint and consistent nodes are retained in the conflict path set, the repeated processing stops. The same endpoint means that the column nodes of each path in the path group are the same, and consistent nodes mean that the community nodes, ownership nodes, and terminal nodes of each path in the path group are the same. After stopping, the column nodes corresponding to each path group are determined as the target publishing column set, the candidate review paths retained in each path group are determined as the target review link set, and the intersection of all terminal node sets in each path group is determined as the target display boundary set. If the terminal node intersection is empty, the intersection of the default terminal set corresponding to the target publishing column set and the terminal set allowed by the ownership node is read as the alternative result. The target publishing column set, target review link set, and target display boundary set are output, and the results are written to the target result table for S4-5 to read. The abnormal or missing handling is as follows: if there are still path groups with the same endpoint but inconsistent nodes after two consecutive rounds of processing, the path groups with community nodes and main community identifiers are retained to enter the next round, and the remaining path groups are deleted. If it is still impossible to form a path group with the same endpoint and consistent nodes after three consecutive rounds of processing, the current result is written to the queue for manual review. The purpose of S4-5 is to perform final closure confirmation on the target publishing category set, target review link set, and target display boundary set. The inputs are the target publishing category set, target review link set, target display boundary set, and target result table. The processing actions are as follows: For each target publishing category, read it one by one and check if there is at least one target review link ending with that target publishing category. If not, record that target publishing category as an unclosed category. Then, read the terminal node corresponding to each target review link one by one and check if that terminal node is included in the target display boundary set. If not included, the target audit link is marked as an unclosed link; after the check is completed, only the result that has no unclosed column target record and no unclosed link mark is determined as the constraint backtracking resolution result, and the constraint backtracking resolution result is written into the resolution result table for S5 to read; the abnormal or missing handling is as follows: when there is an unclosed column, the path group corresponding to the unclosed column is rolled back to S4-1 to regenerate the conflict table; when there is an unclosed link, the path group corresponding to the unclosed link is rolled back to S4-3 to re-execute the constraint backtracking resolution; the path group that still cannot be closed after rollback is not included in the constraint backtracking resolution result; Through the above processing, conflicting paths in the candidate publishing path set and candidate review path set can be located segment by segment, sorted hierarchically, and backtracked to reduce them. Finally, stable path groups that simultaneously meet the requirements of consistent column, consistent community, consistent ownership, and consistent terminal are retained. Based on this, the target publishing column set, the target review link set, and the target display boundary set are formed, providing a unique and verifiable input result for subsequent hierarchical review and targeted publishing. In practical applications: When the same resource corresponds to two candidate publishing paths simultaneously, namely "Community Activity Promotion Section" and "Public Information Section," but the candidate review path can only support the former for publishing on mobile and portal terminals, the system first aligns the two candidate publishing paths with their corresponding candidate review paths segment by segment. Inconsistencies in section nodes and terminal nodes appearing in the Public Information Section path are recorded in the section conflict table and terminal conflict table, respectively. Then, the starting superedge, propagation superedge, and ending superedge in the conflict path are located sequentially, and a priority sequence is formed based on the number of conflict types, the amount of endpoint deviation, and the number of reuses. Then, for low-priority conflicting edges, the corresponding relationship count is reduced and illegal paths that no longer connect community nodes, column nodes, ownership nodes, and terminal nodes are removed. At the same time, the remaining candidate review paths are written back to the candidate publishing path set to correct the edge weights. After repeated processing, only the path group with the destination "Community Activity Promotion Column" and the same community node, ownership node, and terminal node is retained. The column node corresponding to this path group is determined as the target publishing column set, its review path is determined as the target review link set, and the common reachable terminal range of mobile terminal and portal terminal is determined as the target display boundary set.

[0021] S5. Perform hierarchical review according to the target review link set, and after the review is passed, perform targeted release, cross-terminal mirror distribution and release status write-back according to the target release column set and target display boundary set, and output hierarchical authorization release results; In this implementation, the purpose of S5 is to complete hierarchical review, release reachability determination, targeted release, cross-terminal mirror distribution, and release status write-back after determining the target release category set, target review link set, and target display boundary set in S4, thereby forming a traceable hierarchical authorized release result. The processing logic is as follows: first, generate a review execution sequence according to the target review link set; then, complete the link closure review for each node in the review execution sequence; and perform release reachability calculation for the approved links. Subsequently, complete the category writing and terminal mirror distribution according to the target release link set, and finally write the release result back to the community college's release knowledge graph. This implementation process includes the following steps: The purpose of S5-1 is to convert resources to be published into hierarchical review tasks that can be executed node by node. The inputs are the resources to be published, the target review link set, the target publication category set, and the target display boundary set. The processing actions are: writing the resource identifier, target publication category identifier, target display boundary identifier, and each target review link into the review execution table, and generating a review execution sequence according to the sequential connection order of each review node in the target review link set; loading the corresponding review fields, previous review results, and currently pending review fields for each review node, where the corresponding review fields include the resource source field, category ownership field, and ownership pointer. The fields and display boundary fields, the previous review result is the review status and error correction record output by the previous review node, and the current field to be reviewed is the field value corresponding to the current review node; output the hierarchical review task sequence, and write the review node identifier, node order, corresponding review field, previous review result and current field to be reviewed into the review task table for S5-2 to read; the abnormal or missing handling is as follows: when a certain review node in the target review link set is missing the corresponding review field, it is supplemented from the resource feature table and resolution result table corresponding to the resource to be released. If it is still missing after supplementation, the review node is directly written into the return warning table and the link will not enter the subsequent execution; The purpose of S5-2 is to confirm, node by node, whether the review process forms a closed loop with consistent fields and continuous results. The input consists of a hierarchical review task sequence, a review task table, and a resource feature table corresponding to the resource to be published. The processing steps are as follows: Read the current field to be reviewed and the previous review results one by one according to the review node order, and verify the consistency of resource source, category affiliation, ownership direction, and display boundary. Consistent resource source means that the source field in the current review node is the same as the source field of the resource to be published; consistent category affiliation means that the category node in the current review node belongs to the target published category set; consistent ownership direction means that the ownership node in the current review node is the same as the ownership node in the target review link; and consistent display boundary means that the terminal node in the current review node is included in the target display boundary set. When all four results are consistent, the review node is written into the approved link. If any item is inconsistent, the audit node is written into the fallback link, and the inconsistent field identifier, the current field value, and the corresponding field value in the previous audit result are written back to the previous audit node to form an error correction record; the audit pass link set and the audit fallback link set are output, and the pass links and fallback links are written into the audit result table for S5-3 to read; the exception or missing handling is as follows: if an audit node is missing the previous audit result, its previous audit result is set to empty and directly compared with the original field of the resource to be published. If a consistent result still cannot be formed, it is written into the audit fallback link set; The purpose of S5-3 is to filter out the truly valid publishing links from the approved link set that can actually reach the target publishing section and the target display boundary. The input quantities are the approved link set, the rejected link set, the target publishing section set, and the target display boundary set. The processing actions are as follows: for each approved link, match it one by one with the target publishing section set and the target display boundary set, and perform a publishing reachability solution. Here, section reach closure means that the end section node of the approved link belongs to the target publishing section set, terminal reach closure means that the terminal node in the approved link belongs to the target display boundary set, and boundary containment closure means that the target display boundary set contains all the terminal nodes corresponding to the approved link. When all three conditions are met, the approved link is retained as a valid publishing link; if any condition is not met, the link is moved to the rejected link set, and the link closure review in S5-2 is performed again. After each round of processing, the valid links in this round are compared with the valid links in the previous round one by one. Processing stops when the link identifier, endpoint column node, and terminal node are consistent in two consecutive rounds. The target release link set is output and written to the release link table for S5-4 to read. The abnormal or missing handling is as follows: if no valid release link can be formed in two consecutive rounds, the current resource is written to the release table pending manual review and no subsequent release is executed. The purpose of S5-4 is to complete targeted publishing and cross-terminal mirror distribution based on the target publishing link set; the inputs are the target publishing link set, the target publishing category set, the target display boundary set, and the content of the resource to be published; the processing action is: read the corresponding category node one by one according to the target publishing link set, and write the resource to be published into the category storage location corresponding to the target publishing category set. The category storage location includes the resource identifier field, category index field, boundary marker field, and approval marker field in the category resource table; Subsequently, a mirror distribution table is generated based on the correspondence between each terminal node and the column node in the target display boundary set. The mirror distribution table includes at least the resource identifier, source column identifier, target terminal identifier, column index, boundary marker, review marker, and distribution status. Then, the column index, boundary marker, and review marker are copied to the corresponding terminal one by one according to the mirror distribution table. A mirror record is generated when the terminal node is included in the target display boundary set, and no mirror record is generated when the terminal node is not included in the target display boundary set. The cross-terminal publishing result set is output, and the distribution status of each terminal is written to the distribution table result area for S5-5 to read. The exception or missing handling is as follows: if a mirror record of the same resource already exists on the same terminal, a second record is not generated repeatedly. Only the boundary marker and review marker are updated. If the writing fails on a certain terminal, the corresponding distribution status of the terminal is recorded as failed, and the successful results of the other terminals are retained. The purpose of S5-5 is to perform a unified write-back on the cross-terminal publishing result set and form the final hierarchical authorization publishing result. The inputs are the cross-terminal publishing result set, the publishing link table, the distribution publishing result area, and the community college publishing knowledge graph. The processing actions are as follows: summarize the write results, column results, and boundary results of each terminal, generate a publishing status record, which includes at least the resource identifier, publishing column identifier, publishing terminal identifier, display boundary identifier, review link identifier, distribution result, write-back time, and current publishing status; then write the publishing status record back to the resource node, column node, and terminal node corresponding to the resource to be published, and write the published relationship between the resource node, column node, and terminal node in the community college publishing knowledge graph; if a corresponding published relationship already exists, update its boundary marker, review marker, and publishing time; output the hierarchical authorization publishing result and write the result to the publishing result table for subsequent calls; the exception or missing handling is as follows: when a terminal distribution fails, the published relationship of that terminal is not written, only the failure record is written; when a column write fails, delete the terminal mirror record corresponding to that column and mark the result corresponding to that column as incomplete publishing. Through the above processing, the target review link set, target publication column set, and target display boundary set can be converted into executable review results and publication results. This ensures that the resources to be published can be published and distributed across terminals only under the conditions of closed review links, closed column arrivals, closed terminal arrivals, and closed boundaries. The publication status is also synchronously written back to the community college's publication knowledge graph to ensure that the publication results are traceable, verifiable, and reusable. In practical applications: When the target publication category set corresponding to a community activity resource is the community activity promotion category, and the target review link set includes the community affiliation review node, the ownership confirmation review node, and the display boundary review node, and the target display boundary set includes the portal and mobile terminals, the system first generates a hierarchical review task sequence according to the above review node order, and verifies the resource source, category affiliation, ownership direction, and display boundary node node by node; after the review is passed, it checks whether the review link reaches the community activity promotion category at the same time and whether its terminal nodes fall into the target display boundary set of the portal and mobile terminals; when the conditions are met, the resource is written to the category storage location of the community activity promotion category, and a mirror distribution table for the portal and mobile terminals is generated, and the category index, boundary marker, and review marker are copied to the corresponding terminals; finally, the writing results, category results, and boundary results of the portal and mobile terminals are summarized into a publication status record, written back to the corresponding resource node, category node, and terminal node, and a published relationship is formed in the community college publication knowledge graph, thereby outputting the hierarchical authorization publication result of the resource.

[0022] Furthermore, the present invention also includes a community college multi-section resource hierarchical authorization publishing system, the system comprising a feature extraction module, a hypergraph construction module, an inference module, a conflict resolution module, and an authorization publishing module: The feature extraction module is used to obtain the text data, title data, source data, target community data, submission entity data, candidate column data, and display terminal data of the resources to be published, and to perform community entity extraction, column semantic extraction, resource type extraction, cooperation relationship extraction, and ownership constraint extraction, and output a set of resource features; The hypergraph construction module is used to obtain the knowledge graph published by the community college, map the resource feature set into resource nodes, community nodes, column nodes, cooperation entity nodes, ownership nodes and terminal nodes, and construct candidate hypergraphs based on column adaptation relationship, community affiliation relationship, cooperation constraint relationship, ownership restriction relationship and terminal display relationship; The reasoning module is used to perform multi-path competition reasoning on candidate hypergraphs, calculate the column adaptation strength, community consistency strength, ownership satisfaction strength and terminal matching strength of each associated path, and output the candidate publishing path set and the candidate review path set; The conflict resolution module is used to perform constraint backtracking resolution on the candidate publishing path set and the candidate review path set, and to perform priority suppression, edge weight decay and illegal path elimination on the super edges corresponding to conflict paths. It outputs the target publishing column set, the target review link set and the target display boundary set. The authorization and release module is used to perform hierarchical review according to the target review link set, and after the review is passed, it performs targeted release, cross-terminal mirror distribution and release status write-back according to the target release column set and target display boundary set, and outputs hierarchical authorization and release results.

[0023] The working principle of this solution is as follows: First, the information in the resources to be published, such as title, body text, source, target community, candidate column, and display terminal, is organized into unified resource features. Then, these resource features are mapped to resource nodes, community nodes, column nodes, ownership nodes, and terminal nodes in the community college's publishing knowledge graph to form a candidate hypergraph. Subsequently, the reachable paths from resource nodes to column nodes are expanded in the candidate hypergraph, and the validity of each path in terms of column purpose, community affiliation, ownership scope, and terminal scope is determined to obtain candidate publishing paths and candidate review paths. Finally, the two types of paths are compared segment by segment to locate community conflicts and column conflicts. Conflicts in ownership and terminals are identified and eliminated by backtracking, leaving stable path groups that can simultaneously support the publishing section, review process, and display boundaries. Finally, hierarchical review tasks are generated according to the stable path groups. Once approved, resources are directed to the corresponding section and synchronized to the corresponding terminals according to the display boundaries. At the same time, the publishing status is written back to the knowledge graph, forming a traceable hierarchical authorized publishing result. In this way, the output of the previous step becomes the basis for the judgment of the next step. The whole process converges layer by layer in the order of "resource content, community relationship, section function, ownership restriction, and terminal scope", rather than relying on manual selection of sections and manual determination of the public scope. For example, in scenarios where schools collaborate with communities, townships, or other partners to establish community colleges, and rely on online learning platforms, mobile learning systems, course resources, and smart portals to conduct news releases, course displays, activity promotions, and community services, if a resource titled "A community college is conducting a reading activity" is submitted, the system will first identify the community name, activity category, partnership, and ownership information from the title, body, and source. Then, it will determine whether the resource is more suitable for inclusion in the activity promotion section, learning section, or general information section of the community college portal. Simultaneously, it will verify whether the resource can only be displayed on that community's page. Can it be synchronized to the school's main portal? Can it be distributed to mobile devices? If a path can reach a certain section, but it is inconsistent with the community scope, activity purpose, or terminal scope of the resource, the system will remove the path during the conflict resolution phase. Only paths that conform to community affiliation, section purpose, ownership restrictions, and terminal boundaries will be retained. Then, the system will complete the review according to the corresponding review link. Once approved, the content will be published to the correct section and synchronized to the terminals that are allowed to display it. This will prevent content that should be displayed in the community college section from being mistakenly published in the public section, or content that should be subject to scope restrictions from being disseminated beyond the scope.

[0024] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for hierarchical authorization and publishing of multi-section resources in a community college, characterized in that, include: S1. Obtain the main text data, title data, source data, target community data, submission entity data, candidate column data, and display terminal data of the resource to be published, and perform community entity extraction, column semantic extraction, resource type extraction, cooperation relationship extraction, and ownership constraint extraction, and output the resource feature set; S2. Obtain the knowledge graph published by the community college, map the resource feature set into resource nodes, community nodes, column nodes, cooperation entity nodes, ownership nodes and terminal nodes, and construct candidate hypergraphs based on column adaptation relationship, community affiliation relationship, cooperation constraint relationship, ownership restriction relationship and terminal display relationship; S3. Perform multi-path competition reasoning on the candidate hypergraph, calculate the column adaptation strength, community consistency strength, ownership satisfaction strength and terminal matching strength of each associated path, and output the candidate publishing path set and the candidate review path set. S4. Perform constraint backtracking and resolution on the candidate publishing path set and the candidate review path set. Perform priority suppression, edge weight decay and illegal path elimination on the super-edges corresponding to conflicting paths. Output the target publishing column set, the target review link set and the target display boundary set. S5. Perform hierarchical review according to the target review link set, and after the review is passed, perform targeted release, cross-terminal mirror distribution and release status write-back according to the target release column set and target display boundary set, and output hierarchical authorization release results.

2. The method for hierarchical authorization and publishing of multi-section resources in a community college according to claim 1, characterized in that: S1 includes: S1-1. Perform field normalization and semantic segmentation on the text data, title data, source data, target community data, submission subject data, candidate column data, and display terminal data of the resources to be published, extract community name segment, column function segment, resource category segment, cooperation direction segment, and ownership limitation segment, and output the initial feature fragment set. S1-2. Perform cross-field co-occurrence association and position constraint aggregation on the initial feature fragment set, construct candidate semantic chains from feature fragments that point to the same community, the same column purpose, the same resource category, the same cooperation relationship, and the same ownership limit, and output the candidate semantic chain set. S1-3. Perform conflict fragment suppression and main semantic chain determination on the candidate semantic chain set, and determine the candidate semantic chains that are jointly verified by source data, target community data and candidate column data as the resource feature set.

3. The method for hierarchical authorization and publishing of multi-section resources in a community college according to claim 2, characterized in that: S2 includes: S2-1. Input the resource feature set into the knowledge graph published by the community college, and perform node retrieval, alias merging and field alignment on the community features, column features, cooperation features, ownership features and terminal features in the resource feature set respectively, and output a candidate node set composed of resource nodes, community nodes, column nodes, cooperation entity nodes, ownership nodes and terminal nodes. S2-2. Perform relation expansion and coreference verification on the candidate node set in the knowledge graph published by the community college. Construct the associated node group that simultaneously has column adaptation relationship, community affiliation relationship, cooperation constraint relationship, ownership restriction relationship and terminal display relationship into candidate relation clusters, and output the candidate relation cluster set. S2-3. Perform hyperedge encapsulation and conflict edge suppression on the candidate relation cluster set, and construct the candidate relation clusters that meet the preset conditions for association strength and do not have node semantic conflicts into candidate hypergraphs.

4. The method for hierarchical authorization and publishing of multi-section resources in a community college according to claim 3, characterized in that: S3 includes: S3-1. Perform path expansion on all reachable paths in the candidate hypergraph that start from resource nodes and end at column nodes, retain the associated paths that pass through community nodes, ownership nodes and terminal nodes and whose connection directions are consistent, and output the initial set of associated paths. S3-2. Perform itemized calculations on each associated path in the initial associated path set according to path segments. Determine the corresponding result from the resource node to the column node as the column adaptation strength, determine the ownership closure result from the resource node through the community node to the column node as the community consistency strength, determine the permission closure result from the resource node through the ownership node to the column node as the ownership satisfaction strength, and determine the display closure result from the resource node through the terminal node to the column node as the terminal matching strength. Output the associated path strength set. S3-3. Perform publishing path division and review path division on the set of associated path strengths. Write the associated paths that simultaneously close the column adaptation relationship, community affiliation relationship, ownership restriction relationship and terminal display relationship into the candidate publishing path set. Write the associated paths that simultaneously close the resource source relationship, cooperation constraint relationship and ownership confirmation relationship into the candidate review path set. Output the first round of candidate publishing path set and the first round of candidate review path set.

5. The method for hierarchical authorization and publishing of multi-section resources in a community college according to claim 4, characterized in that: S3 also includes: S3-4. Perform consistency checks on the first round of candidate release path set and the first round of candidate review path set. Identify the associated paths in the candidate release path set that lack the support of the corresponding candidate review path as release mismatch paths, and identify the associated paths in the candidate review path set that cannot support the candidate release path to reach the column node and terminal node as review redundant paths. Then, perform backtracking correction on the associated path strength set based on the release mismatch path and the review redundant path, and output the corrected associated path strength set. S3-5. Repeatedly execute the release path division, review path division, and consistency check on the modified associated path strength set until the candidate release path set and candidate review path set output in the previous and next rounds are consistent. Then, output the candidate release path set and candidate review path set.

6. The method for hierarchical authorization and publishing of multi-section resources in a community college according to claim 5, characterized in that: S4 includes: S4-1. Perform pairwise alignment between each candidate publishing path in the candidate publishing path set and each candidate review path in the candidate review path set. Compare the consistency results of community nodes, column nodes, ownership nodes, and terminal nodes segment by segment. Record the path segments with inconsistent nodes, inconsistent relationship directions, inconsistent relationship start and end points, and inconsistent path ends into the community conflict table, column conflict table, ownership conflict table, and terminal conflict table, respectively. Output the conflict path set and conflict superedge set. S4-2. Perform backtracking on each conflict hyperedge in the conflict hyperedge set, and backtrack along each conflict path to form the starting hyperedge, propagation hyperedge, and ending hyperedge of the conflict path. Construct a conflict cost sequence based on the number of conflict types of the starting hyperedge, the number of times the propagation hyperedge is reused, and the deviation of the endpoint of the ending hyperedge. Prioritize each conflict hyperedge in the order of number of conflict types first, then deviation of endpoint, and then number of reuses, and output the conflict hyperedge priority sequence. S4-3. Perform constraint backtracking and resolution on each conflicting superedge according to the priority sequence of conflicting superedges. For low-priority conflicting superedges in the same conflicting path, deduct one corresponding relationship count. For conflicting superedges that no longer simultaneously connect resource nodes, community nodes, column nodes, ownership nodes, and terminal nodes after deduction, perform illegal path elimination. Write the remaining candidate review paths back to the candidate publishing path set to correct the edge weights of the corresponding paths. Output the corrected publishing path set and the corrected review path set.

7. The method for hierarchical authorization and publishing of multi-section resources in a community college according to claim 6, characterized in that: S4 also includes: S4-4. Repeatedly execute conflict table generation, priority sorting and constraint backtracking resolution on the revised release path set and the revised review path set until only path groups with the same endpoint and consistent nodes are retained in the conflict path set. Then, determine the column nodes corresponding to each path group at the time of stopping as the target release column set, determine the corresponding candidate review path as the target review link set, and determine the common reachability of the terminal nodes in each path group as the target display boundary set. S4-5. Perform closure verification on the target publication column set, target review link set, and target display boundary set. The result that each target publication column corresponds to at least one target review link and the terminal node corresponding to each target review link is included in the target display boundary set is determined as the constraint backtracking resolution result.

8. The method for hierarchical authorization and publishing of multi-section resources in a community college according to claim 7, characterized in that: S5 includes: S5-1. Write the resources to be published, the target review link set, the target publishing column set, and the target display boundary set into the review execution sequence. Load the corresponding review fields, previous review results, and current fields to be reviewed one by one according to the connection order of each review node in the target review link set, and output the hierarchical review task sequence. S5-2. Perform closed-loop auditing on the hierarchical audit task sequence. At each audit node, read the current field to be audited and the previous audit results, and verify the consistency of resource source, column ownership, ownership direction and display boundary item by item. Write the audit nodes with consistent verification into the pass link and write the audit nodes with inconsistent verification into the return link. Then write the inconsistent fields in the return link back to the previous audit node to form the error correction result, and output the set of audit pass links and the set of audit return links. S5-3. Perform a publish reachability calculation on the approved link set. Align each approved link with the target publishing category set and the target display boundary set one by one. Keep the links that simultaneously satisfy the category reach closure, terminal reach closure, and boundary containment closure as valid publishing links. Move the links that cannot be closed at the same time into the review return link set and perform link closure review again until the valid publishing links obtained in two consecutive rounds are consistent. Output the target publishing link set.

9. A method for hierarchical authorization and publishing of multi-section resources in a community college according to claim 8, characterized in that: S5 also includes: S5-4. Perform targeted publishing and cross-terminal mirror distribution according to the target publishing link set. Write the resources to be published into the column storage location corresponding to the target publishing column set. Generate a mirror distribution table according to the correspondence between each terminal node and column node in the target display boundary set. Then copy the column index, boundary marker and review marker to the corresponding terminal one by one according to the mirror distribution table. Output the cross-terminal publishing result set. S5-5. Perform a write-back of the publishing status for the cross-terminal publishing result set. Summarize the write results, column results and boundary results of each terminal into a publishing status record and write it back to the resource node, column node and terminal node corresponding to the resource to be published. At the same time, write the published relationship between the resource node, column node and terminal node into the community college publishing knowledge graph and output the hierarchical authorization publishing results.

10. A community college multi-section resource hierarchical authorization publishing system, used to implement the community college multi-section resource hierarchical authorization publishing method according to any one of claims 1-9, the system comprising a feature extraction module, a hypergraph construction module, an inference module, a conflict resolution module, and an authorization publishing module, characterized in that: The feature extraction module is used to obtain the text data, title data, source data, target community data, submission entity data, candidate column data, and display terminal data of the resources to be published, and to perform community entity extraction, column semantic extraction, resource type extraction, cooperation relationship extraction, and ownership constraint extraction, and output a set of resource features; The hypergraph construction module is used to obtain the knowledge graph published by the community college, map the resource feature set into resource nodes, community nodes, column nodes, cooperation entity nodes, ownership nodes and terminal nodes, and construct candidate hypergraphs based on column adaptation relationship, community affiliation relationship, cooperation constraint relationship, ownership restriction relationship and terminal display relationship; The reasoning module is used to perform multi-path competition reasoning on candidate hypergraphs, calculate the column adaptation strength, community consistency strength, ownership satisfaction strength and terminal matching strength of each associated path, and output the candidate publishing path set and the candidate review path set; The conflict resolution module is used to perform constraint backtracking resolution on the candidate publishing path set and the candidate review path set, and to perform priority suppression, edge weight decay and illegal path elimination on the super edges corresponding to conflict paths. It outputs the target publishing column set, the target review link set and the target display boundary set. The authorization and release module is used to perform hierarchical review according to the target review link set, and after the review is passed, it performs targeted release, cross-terminal mirror distribution and release status write-back according to the target release column set and target display boundary set, and outputs hierarchical authorization and release results.

Images