Heterogeneous multi-mode resource arrangement and decision control method and system

Abstract

The invention provides a heterogeneous multi-modal resource arrangement and decision control method and system, and the method comprises the steps: obtaining a knowledge graph element change feature of a resource library according to a resource updating log of the heterogeneous multi-modal resource library, thereby calibrating a resource block needing to be changed of the resource library; monitoring resource sending samples of all resource providing ends in butt joint with a resource library, so as to determine matching attributes of all the resource providing ends and resource blocks needing to be changed, and implementing differentiated acquisition and arrangement on heterogeneous multi-modal resources from all the resource providing ends to obtain a plurality of heterogeneous multi-modal resource sets; accurate resource docking and arrangement of different resource blocks are realized; according to the method, the editing decision of the heterogeneous multi-mode resource set is determined according to the historical access characteristics of the resource block needing to be changed in the resource library, and the heterogeneous multi-mode resource set for implementing the editing decision is configured to the resource library, so that the updating requirements of different resource blocks are met, and the updating efficiency and reliability of the resource library are improved.

Description

Technical Field

[0001] This invention relates to the field of multimodal data processing, and more particularly to methods and systems for orchestration and decision control of heterogeneous multimodal resources. Background Technology

[0002] Heterogeneous multimodal resources comprise various data resources from different data sources, differing in content, structure, and format. These data resources collectively form a data resource set. The inherent heterogeneity of these resources makes conventional database storage methods inadequate for their systematic and ordered storage. Knowledge graphs, with their triple structure of entities, relations, and attributes, have been widely applied to the representation of heterogeneous multimodal resources. Currently, the use of knowledge graphs for multi-dimensional storage of heterogeneous multimodal resources is widely adopted. Heterogeneous multimodal resource repositories require continuous resource data updates, especially for resource blocks with errors, necessitating the acquisition of new resources from the resource provider for overall orchestration and deployment. Considering the large number of resource blocks within the heterogeneous multimodal resource repository, and the varying resource errors and update requirements of each block, precise resource orchestration and editing decisions for different resource blocks are crucial for the comprehensive and efficient updating of the heterogeneous multimodal resource repository. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for orchestration and decision control of heterogeneous multimodal resources. Based on the resource update logs of the heterogeneous multimodal resource library, the change characteristics of the knowledge graph elements in the resource library are obtained, thereby identifying the resource blocks in the resource library that need modification. Resource transmission samples from all resource providers connected to the resource library are monitored to determine the matching attributes of each resource provider with the resource blocks that need modification. This allows for the differentiated collection and orchestration of heterogeneous multimodal resources from all resource providers, resulting in several heterogeneous multimodal resource sets. This enables precise resource docking and orchestration of different resource blocks. Based on the historical access characteristics of the resource blocks that need modification in the resource library, an editing decision is determined for the heterogeneous multimodal resource sets, and the heterogeneous multimodal resource sets with the editing decision are configured into the resource library to meet the update needs of different resource blocks and improve the update efficiency and reliability of the resource library.

[0004] This invention is achieved through the following technical solution:

[0005] Heterogeneous multimodal resource orchestration and decision control methods include:

[0006] Based on the resource update logs of the heterogeneous multimodal resource library, the change characteristics of the knowledge graph elements of the resource library are obtained; based on the change characteristics of the knowledge graph elements, the resource blocks of the resource library that need to be changed are identified.

[0007] Listen to the resource sending samples of all resource providers connected to the resource library to determine the matching attributes of each resource provider with the resource block to be changed; based on the matching attributes, perform differentiated collection and orchestration of heterogeneous multimodal resources from all resource providers to obtain several heterogeneous multimodal resource sets;

[0008] Based on the historical access characteristics of the resource blocks to be modified in the resource library, an editing decision is determined for the heterogeneous multimodal resource set; based on the real-time access status of the resource library, the heterogeneous multimodal resource set for which the editing decision is implemented is configured into the resource library.

[0009] Optionally, based on the resource update logs of the heterogeneous multimodal resource library, the knowledge graph element change characteristics of the resource library are obtained; based on the knowledge graph element change characteristics, the resource blocks of the resource library that need to be changed are identified, including:

[0010] The resource update logs of the heterogeneous multimodal resource library are parsed to determine the resource update attributes of each resource block within the resource library. The resource update attributes refer to the resource change location, resource change information type, and resource change information volume of the resource block. Based on the resource update attributes, the entity and relation change features of the knowledge graph corresponding to each resource block are determined.

[0011] The number of abnormal entity changes and the number of abnormal relationship changes that occurred in the knowledge graph are extracted from the entity and relationship change features. This is used to determine whether the knowledge graph has a semantic error, and the resource blocks corresponding to the knowledge graph with semantic errors are marked as resource blocks that need to be changed.

[0012] Optionally, resource transmission samples from all resource providers connected to the resource library are monitored to determine the matching attributes between each resource provider and the resource block to be changed; based on the matching attributes, heterogeneous multimodal resources from all resource providers are selectively collected and orchestrated to obtain several heterogeneous multimodal resource sets, including:

[0013] Based on the resource transmission traffic of each resource provider connected to the resource library, resource transmission samples are periodically sampled from each resource provider, and the semantic information of the resources provided by each resource provider is determined based on the resource transmission samples. The semantic information of the resources provided by each resource provider is compared with the original semantic information of the resource block to be changed to determine the matching attribute between each resource provider and the resource block to be changed. The matching attribute refers to the degree of similarity between the semantic information of the resources provided by each resource provider and the original semantic information of the resource block to be changed.

[0014] Based on the matching attributes, a priority level is determined for the differentiated acquisition and orchestration of heterogeneous multimodal resources from all resource providers; based on the priority level, the heterogeneous multimodal resources from all resource providers are individually integrated to obtain several heterogeneous multimodal resource sets.

[0015] Optionally, based on the historical access characteristics of the resource block to be changed in the resource repository, an editing decision for the heterogeneous multimodal resource set is determined; based on the real-time access status of the resource repository, the heterogeneous multimodal resource set for which the editing decision is implemented is configured into the resource repository, including:

[0016] Extract the historical access characteristics of the resource block to be modified from the access log of the resource repository; wherein, the historical access characteristics include the historical access path and historical location keywords of the resource block to be modified; based on the historical access characteristics, determine the editing decision for the heterogeneous multimodal resource set; wherein, the editing decision includes editing decisions for at least one of resource content, resource structure, and resource format;

[0017] Based on the real-time access request volume of the resource library, the available resource configuration bandwidth of the resource library is determined, thereby converting the heterogeneous multimodal resource set implementing the editing decision into a resource data stream with corresponding traffic and configuring it into the resource library.

[0018] Heterogeneous multimodal resource orchestration and decision control system, including:

[0019] The resource library change determination module is used to obtain the knowledge graph element change characteristics of the resource library based on the resource update log of the heterogeneous multimodal resource library;

[0020] The resource block identification module is used to identify the resource blocks in the resource library that need to be changed based on the change characteristics of the knowledge graph elements.

[0021] The resource matching module is used to monitor the resource sending samples of all resource providers connected to the resource library, so as to determine the matching attributes of each resource provider with the resource block to be changed;

[0022] The resource orchestration module is used to differentiate, collect, and orchestrate heterogeneous multimodal resources from all resource providers based on the matching attributes, thereby obtaining several heterogeneous multimodal resource sets.

[0023] The resource editing decision module is used to determine the editing decision for the heterogeneous multimodal resource set based on the historical access characteristics of the resource block to be changed in the resource library.

[0024] The resource configuration module is used to configure the heterogeneous multimodal resource set that implements the editing decision into the resource library based on the real-time access status of the resource library.

[0025] Optionally, the resource library change determination module is used to obtain the knowledge graph element change characteristics of the resource library based on the resource update log of the heterogeneous multimodal resource library, including:

[0026] The resource update logs of the heterogeneous multimodal resource library are parsed to determine the resource update attributes of each resource block within the resource library. The resource update attributes refer to the resource change location, resource change information type, and resource change information volume of the resource block. Based on the resource update attributes, the entity and relation change features of the knowledge graph corresponding to each resource block are determined.

[0027] The modified resource block identification module is used to identify the resource blocks in the resource library that need to be modified based on the modification characteristics of the knowledge graph elements, including:

[0028] The number of abnormal entity changes and the number of abnormal relationship changes that occurred in the knowledge graph are extracted from the entity and relationship change features. This is used to determine whether the knowledge graph has a semantic error, and the resource blocks corresponding to the knowledge graph with semantic errors are marked as resource blocks that need to be changed.

[0029] Optionally, the resource matching module is used to monitor resource sending samples from all resource providers connected to the resource library, thereby determining the matching attributes of each resource provider with the resource block to be changed, including:

[0030] Based on the resource transmission traffic of each resource provider connected to the resource library, resource transmission samples are periodically sampled from each resource provider, and the semantic information of the resources provided by each resource provider is determined based on the resource transmission samples. The semantic information of the resources provided by each resource provider is compared with the original semantic information of the resource block to be changed to determine the matching attribute between each resource provider and the resource block to be changed. The matching attribute refers to the degree of similarity between the semantic information of the resources provided by each resource provider and the original semantic information of the resource block to be changed.

[0031] The resource orchestration module is used to differentiate, collect, and orchestrate heterogeneous multimodal resources from all resource providers based on the matching attributes, resulting in several heterogeneous multimodal resource sets, including:

[0032] Based on the matching attributes, a priority level is determined for the differentiated acquisition and orchestration of heterogeneous multimodal resources from all resource providers; based on the priority level, the heterogeneous multimodal resources from all resource providers are individually integrated to obtain several heterogeneous multimodal resource sets.

[0033] Optionally, the resource editing decision module is used to determine the editing decision for the heterogeneous multimodal resource set based on the historical access characteristics of the resource block to be changed in the resource library, including:

[0034] Extract the historical access characteristics of the resource block to be modified from the access log of the resource repository; wherein, the historical access characteristics include the historical access path and historical location keywords of the resource block to be modified; based on the historical access characteristics, determine the editing decision for the heterogeneous multimodal resource set; wherein, the editing decision includes editing decisions for at least one of resource content, resource structure, and resource format;

[0035] The resource configuration module is used to configure the heterogeneous multimodal resource set that implements the editing decision to the resource library according to the real-time access status of the resource library, including:

[0036] Based on the real-time access request volume of the resource library, the available resource configuration bandwidth of the resource library is determined, thereby converting the heterogeneous multimodal resource set implementing the editing decision into a resource data stream with corresponding traffic and configuring it into the resource library.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] The heterogeneous multimodal resource orchestration and decision control method and system provided in this application obtains the knowledge graph element change characteristics of the resource library based on the resource update logs of the heterogeneous multimodal resource library, thereby identifying the resource blocks in the resource library that need to be changed; it monitors the resource sending samples of all resource providers connected to the resource library to determine the matching attributes of each resource provider with the resource blocks that need to be changed, thereby distinguishing, collecting and orchestrating heterogeneous multimodal resources from all resource providers to obtain several heterogeneous multimodal resource sets, realizing accurate resource docking and orchestration of different resource blocks; based on the historical access characteristics of the resource blocks that need to be changed in the resource library, it determines the editing decision for the heterogeneous multimodal resource sets, and configures the heterogeneous multimodal resource sets with the editing decision to the resource library, meeting the update needs of different resource blocks and improving the update efficiency and reliability of the resource library. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0040] Figure 1 This is a flowchart illustrating the heterogeneous multimodal resource orchestration and decision control method provided by the present invention.

[0041] Figure 2 This is a schematic diagram of the heterogeneous multimodal resource orchestration and decision control system provided by the present invention. Detailed Implementation

[0042] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, it should be noted that, for ease of description, only the parts relevant to this application are shown in the accompanying drawings, not the entire structure. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this application.

[0043] The terms “comprising” and “having”, and any variations thereof, used in this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0044] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0045] Please see Figure 1 As shown, an embodiment of this application provides a heterogeneous multimodal resource orchestration and decision control method. This heterogeneous multimodal resource orchestration and decision control method includes:

[0046] Based on the resource update logs of the heterogeneous multimodal resource library, the change characteristics of the knowledge graph elements of the resource library are obtained; based on the change characteristics of the knowledge graph elements, the resource blocks of the resource library that need to be changed are identified.

[0047] Listen to the resource sending samples of all resource providers connected to the resource library to determine the matching attributes of each resource provider with the resource block to be changed; based on the matching attributes, collect and arrange the heterogeneous multimodal resources from all resource providers to obtain several heterogeneous multimodal resource sets.

[0048] Based on the historical access characteristics of the resource blocks to be changed in the resource repository, the editing decision for the heterogeneous multimodal resource set is determined; based on the real-time access status of the resource repository, the heterogeneous multimodal resource set to be edited is configured into the resource repository.

[0049] The beneficial effects of the above embodiments are as follows: the heterogeneous multimodal resource orchestration and decision control method obtains the knowledge graph element change characteristics of the resource library based on the resource update log of the heterogeneous multimodal resource library, thereby identifying the resource blocks in the resource library that need to be changed; it monitors the resource sending samples of all resource providers connected to the resource library to determine the matching attributes of each resource provider with the resource blocks that need to be changed, thereby distinguishing, collecting and orchestrating heterogeneous multimodal resources from all resource providers to obtain several heterogeneous multimodal resource sets, realizing accurate resource docking and orchestration of different resource blocks; based on the historical access characteristics of the resource blocks that need to be changed in the resource library, it determines the editing decision for the heterogeneous multimodal resource sets, and configures the heterogeneous multimodal resource sets with the editing decision to the resource library, meeting the update needs of different resource blocks and improving the update efficiency and reliability of the resource library.

[0050] In another embodiment, the knowledge graph element change characteristics of the resource library are obtained based on the resource update logs of the heterogeneous multimodal resource library; based on the knowledge graph element change characteristics, the resource blocks in the resource library that need to be changed are identified, including:

[0051] The resource update logs of the heterogeneous multimodal resource library are analyzed to determine the resource update attributes of each resource block within the library. The resource update attributes refer to the location of resource changes, the type of resource change information, and the amount of resource change information in the resource block. Based on the resource update attributes, the entity and relation change characteristics of the corresponding knowledge graph of each resource block are determined.

[0052] The number of abnormal entity changes and abnormal relationship changes in the knowledge graph are extracted from the entity and relation change features to determine whether the knowledge graph has semantic errors. The resource blocks corresponding to the knowledge graph with semantic errors are then marked as resource blocks that need to be changed.

[0053] The beneficial effects of the above embodiments are that the heterogeneous multimodal resource library includes multiple resource blocks, each of which stores heterogeneous multimodal resources formed in different time periods or from different resource providers. When any resource block in the resource library undergoes a heterogeneous multimodal resource change, the resource update log will generate corresponding log information. Simultaneously, the knowledge graph corresponding to the resource block will also experience changes in at least one element among entities, relationships, and attributes during the heterogeneous multimodal resource change process. When the updated heterogeneous multimodal resources received by a resource block contain errors and / or interference, the corresponding knowledge graph will have corresponding errors at the element level. These errors are more obvious at the knowledge graph level. When the knowledge graph shows errors after semantic recognition, it can be determined that the corresponding resource block has resource change errors, etc. In this case, the aforementioned resource block should be identified as a resource block that needs to be changed due to heterogeneous multimodal resource errors. Specifically, the resource update logs of the heterogeneous multimodal resource library are first analyzed to determine the location, type, and amount of resource changes for each resource block within the library. A correspondence exists between the data resources within each resource block and the elements of its corresponding knowledge graph. For example, when the content, structure, format, or data volume of a resource at a certain location within a resource block changes, the corresponding knowledge graph for that resource block will change at certain entity and relation levels. Therefore, based on the resource update status of the resource blocks and the correspondence between the resource blocks and the knowledge graph, the entity and relation change characteristics of the knowledge graph corresponding to the resource block are determined. These entity and relation change characteristics may include, but are not limited to, changes in entity names within the knowledge graph and changes in the relationship types between different entities. When an entity name in the knowledge graph becomes blank or an incorrect name, or when existing relationships between different entities in the knowledge graph disappear or relationships are formed between entities that did not originally have any relationship, these situations can be considered as abnormal entity or relation changes in the knowledge graph. The number of abnormal entity changes and abnormal relationship changes in the knowledge graph are extracted from the entity and relationship change features. If the number of abnormal entity changes or abnormal relationship changes exceeds a preset threshold, it is determined that the knowledge graph has a semantic error; otherwise, it is determined that the knowledge graph has no semantic error, and the resource block corresponding to the knowledge graph with the semantic error is marked as a resource block that needs to be changed. This facilitates subsequent heterogeneous multimodal resource change processing for the resource block that needs to be changed, reducing the workload of changing the resource library.

[0054] In another embodiment, resource transmission samples from all resource providers connected to the resource repository are monitored to determine the matching attributes between each resource provider and the resource block to be modified. Based on the matching attributes, heterogeneous multimodal resources from all resource providers are selectively collected and orchestrated to obtain several heterogeneous multimodal resource sets, including:

[0055] Based on the resource transmission traffic of all resource providers connected to the resource repository, resource transmission samples are periodically sampled from each resource provider, and the semantic information of the resources provided by each resource provider is determined based on the resource transmission samples. The semantic information of the resources provided by each resource provider is compared with the original semantic information of the resource block to be modified to determine the matching attributes between each resource provider and the resource block to be modified. The matching attribute refers to the degree of similarity between the semantic information of the resources provided by each resource provider and the original semantic information of the resource block to be modified.

[0056] Based on the matching attributes, the priority level for differentiated collection and orchestration of heterogeneous multimodal resources from all resource providers is determined; based on the priority level, the heterogeneous multimodal resources from all resource providers are individually integrated to obtain several heterogeneous multimodal resource sets.

[0057] The beneficial effects of the above embodiments are that the heterogeneous multimodal resource library obtains heterogeneous multimodal resources from different resource providers. These resource providers can be, but are not limited to, smart terminals such as smartphones and drones. During their operation, these providers acquire heterogeneous multimodal resources through methods such as photography, user interaction, and internet data mining. The way each resource provider sends heterogeneous multimodal resources to the resource library differs, mainly reflected in the different resource transmission volumes and content (i.e., different amounts of resource data sent per unit time and different content of the sent resource data). Different resource blocks within the resource library correspond to different stored resource attributes (such as resource content, structure type, and format type). To ensure that the resources sent by the resource providers are accurately stored in the corresponding resource blocks, it is necessary to calibrate the resources provided by each resource provider. Specifically, based on the resource transmission traffic of each resource provider connected to the resource repository, resource transmission samples are periodically sampled from each resource provider, and the semantic information of the resources provided by the resource providers is determined based on the resource transmission samples. Then, by comparing the semantic information of the resources provided by each resource provider with the original semantic information of the resource block to be changed, the similarity between the semantic information of the resources provided by each resource provider and the original semantic information of the resource block to be changed is determined. Generally speaking, the higher the similarity, the higher the priority of the heterogeneous multimodal resources sent by the corresponding resource provider in collection and orchestration. Based on the above priority, the heterogeneous multimodal resources from all resource providers are individually integrated to obtain several heterogeneous multimodal resource sets, realizing targeted orchestration processing of heterogeneous multimodal resources from different resource providers, ensuring that subsequent heterogeneous multimodal resources can be accurately configured into the corresponding resource blocks in the resource repository.

[0058] In another embodiment, an editing decision for the heterogeneous multimodal resource set is determined based on the historical access characteristics of the resource block to be modified in the resource repository; the heterogeneous multimodal resource set to be edited is configured into the resource repository according to the real-time access status of the resource repository, including:

[0059] Extract the historical access characteristics of the resource blocks to be modified from the resource repository's access logs; the historical access characteristics include the historical access paths and keywords of the resource blocks to be modified; based on the historical access characteristics, determine the editing decision for the heterogeneous multimodal resource set; the editing decision includes editing decisions for at least one of the resource content, resource structure, and resource format;

[0060] Based on the real-time access request volume of the resource repository, the available resource configuration bandwidth of the resource repository is determined, thereby converting the heterogeneous multimodal resource set that implements editing decisions into resource data streams with corresponding traffic and configuring them into the resource repository.

[0061] The beneficial effects of the above embodiments are that the resource data format of the resource blocks that need to be changed within the heterogeneous multimodal resource library is specific. In order to ensure that the heterogeneous multimodal resource set from the resource provider is compatible with the resource blocks that need to be changed in terms of resource data format, it is necessary to edit the heterogeneous multimodal resource set. Specifically, historical access features such as historical access paths and historical location keywords of the resource blocks that need to be changed in the resource library are extracted from the access log of the resource library. The resource data format corresponding to the resource blocks that need to be changed is determined through the above historical access features, thereby determining the editing decision for at least one of the resource content, resource structure, and resource format of the heterogeneous multimodal resource set, thereby improving the data format compatibility between the heterogeneous multimodal resource set and the resource blocks that need to be changed. Furthermore, based on the real-time access request volume of the resource repository, the available resource configuration bandwidth of the resource repository is determined. This allows the heterogeneous multimodal resource set implementing editing decisions to be converted into a resource data stream with corresponding traffic and configured into the resource repository, enabling resource changes to resource blocks within the resource repository that need to be modified. The traffic of the aforementioned resource data stream is matched with the available resource configuration bandwidth of the resource repository, ensuring that the resource data stream is transmitted quickly and stably to the resource blocks that need to be modified, thereby improving the efficiency and reliability of resource repository updates.

[0062] In one embodiment, determining the editing decision for the heterogeneous multimodal resource set based on historical access characteristics can be specifically implemented as follows:

[0063] For each type of editing decision, a personalized weight is configured. The personalized weight includes access frequency weight, keyword relevance weight, and update cycle weight, and the sum of these three weights is 1. A first ratio is determined between the historical access frequency of the resource block to be changed and the average access frequency of all resource blocks in the resource library. The relevance between the historical access keyword set of the resource block to be changed and the semantic keyword set to be changed is determined. A second ratio is determined between the last update cycle of the resource block to be changed and the maximum allowed update cycle of the resource library. The weight value corresponding to each type of editing decision is determined based on the first ratio, relevance, second ratio, and the personalized weight.

[0064] Based on the weight values ​​corresponding to the three types of editing decisions and the preset decision rules, the editing decision for the heterogeneous multimodal resource set is determined. The preset rules include: taking the editing decision corresponding to the weight value exceeding the preset threshold as the editing decision for the heterogeneous multimodal resource set, or taking the editing decision with the largest weight value as the editing decision for the heterogeneous multimodal resource set, or taking the editing decision with the weight value exceeding the preset threshold and ranking in the top N as the editing decision for the heterogeneous multimodal resource set, where N is 1 or 2.

[0065] The calculation method for the weight value corresponding to each type of editing decision is expressed by formula (1) as follows:

[0066]

[0067] in, This indicates the weight of the j-th resource block requiring modification for the k-th type of editing decision, where k=1 represents content editing, k=2 represents structure editing, and k=3 represents format editing. , , Let the access frequency weight, keyword relevance weight, and update cycle weight be the weights for the k-th type of editing decision. Each weight's value is greater than 0 and less than 1, and satisfies the following condition: It can be flexibly configured according to resource update needs; The average access frequency of all resource blocks in the resource library; Rel(,) is the relevance between the historical access keyword set and the current semantic keyword set that needs to be changed, which can be calculated based on cosine similarity and takes a value of [0,1]. , These are the historical access keyword set and the current semantic keyword set that need to be changed for the j-th resource block, respectively. For the j-th resource block that needs to be changed, the last update period is... The maximum allowed update cycle for the resource repository is configured based on the timeliness requirements of the resources; the current set of semantic keywords to be changed is obtained by comparing the original semantic information of the resource block to be changed with the knowledge graph element change features.

[0068] The above embodiments integrate three core features—historical access frequency of resource blocks requiring modification, relevance between historical access keywords and semantic keywords requiring modification, and update cycle ratio—and combine these with personalized weight parameters to quantify the priority of three types of editing decisions: resource content, structure, and format. This achieves precise anchoring of editing needs, relying on historical access features to align with user habits and focusing on semantic gaps in blocks based on semantic keywords requiring modification (derived from original semantic information and knowledge graph element change features). This ensures that editing decisions directly address the core need for data format compatibility, avoiding wasted computing power due to blind editing. Simultaneously, the quantifiable nature of its weight values ​​allows for accurate control of editing execution intensity; higher weights correspond to more in-depth editing operations. This ensures the editing quality of high-frequency access and high-importance blocks while adapting to the differentiated update needs of different resource blocks, effectively improving the compatibility of heterogeneous multimodal resource sets and resource blocks requiring modification, laying the foundation for efficient resource library updates.

[0069] In one embodiment, the available resource configuration bandwidth of the resource repository is determined based on the real-time access request volume of the resource repository. This bandwidth is then used to convert the heterogeneous multimodal resource set implementing editing decisions into a resource data stream of corresponding traffic and configure it into the resource repository. Specifically, this can be implemented as follows:

[0070] Determine the available resource bandwidth for the resource repository based on the real-time access request volume.

[0071] Based on the available resource configuration bandwidth of the resource pool, the weight value corresponding to each resource block that needs to be changed, and the data importance coefficient of each resource block that needs to be changed, determine the resource configuration traffic that needs to be configured for each resource block that needs to be changed;

[0072] Based on the resource configuration traffic of each resource block that needs to be changed, the heterogeneous multimodal resource set after the implementation of the editing decision is converted into a resource data stream with corresponding traffic, and then transmitted and configured to the corresponding resource block that needs to be changed.

[0073] In one embodiment, the resource configuration traffic for each resource block that needs to be changed can be calculated according to the following formula (2):

[0074] (2)

[0075] in, Configure bandwidth for available resources in the resource library, calculated based on the real-time access request volume; The importance coefficient of the j-th resource block that needs to be changed can be [1,5], and is preset based on the importance of the data in the block; p is the total number of resource blocks that need to be changed in the resource library; This represents the data importance coefficient for the m-th resource block that needs to be modified.

[0076] The above embodiments, based on the available bandwidth of the resource repository, combine editing decision weights (reflecting the intensity of editing needs) and data importance coefficients (reflecting the core nature of the data) to allocate traffic ratios, achieving optimized bandwidth resource allocation. This ensures that the total traffic of all resource blocks requiring modification perfectly matches the available bandwidth, avoiding bandwidth congestion or idleness and guaranteeing fast and stable transmission of resource data streams. Furthermore, by comprehensively prioritizing data, blocks with urgent editing needs and high data importance receive more bandwidth support and are prioritized for resource allocation. This approach aligns with the dynamic changes in the real-time access status of the resource repository and meets the differentiated update priority requirements of different blocks, improving the efficiency and reliability of resource repository updates and increasing the speed at which the edited heterogeneous multimodal resource set is adapted to the resource blocks requiring modification.

[0077] Please see Figure 2 As shown, an embodiment of this application provides a heterogeneous multimodal resource orchestration and decision control system. This heterogeneous multimodal resource orchestration and decision control system includes:

[0078] The resource library change determination module is used to obtain the change characteristics of knowledge graph elements of the resource library based on the resource update log of the heterogeneous multimodal resource library;

[0079] The resource block identification module is used to identify the resource blocks in the resource library that need to be changed based on the change characteristics of knowledge graph elements.

[0080] The resource matching module is used to monitor the resource sending samples of all resource providers connected to the resource library, so as to determine the matching attributes of each resource provider with the resource block that needs to be changed;

[0081] The resource orchestration module is used to differentiate and orchestrate heterogeneous multimodal resources from all resource providers based on matching attributes, resulting in several heterogeneous multimodal resource sets.

[0082] The resource editing decision module is used to determine the editing decision for the heterogeneous multimodal resource set based on the historical access characteristics of the resource blocks to be changed in the resource library.

[0083] The resource configuration module is used to configure the heterogeneous multimodal resource set that will be used for editing decisions to the resource library based on the real-time access status of the resource library.

[0084] The beneficial effects of the above embodiments are as follows: the heterogeneous multimodal resource orchestration and decision control system obtains the knowledge graph element change characteristics of the resource library based on the resource update log of the heterogeneous multimodal resource library, thereby identifying the resource blocks in the resource library that need to be changed; it monitors the resource sending samples of all resource providers connected to the resource library to determine the matching attributes of each resource provider with the resource blocks that need to be changed, thereby distinguishing, collecting and orchestrating heterogeneous multimodal resources from all resource providers to obtain several heterogeneous multimodal resource sets, realizing accurate resource docking and orchestration of different resource blocks; based on the historical access characteristics of the resource blocks that need to be changed in the resource library, it determines the editing decision for the heterogeneous multimodal resource sets, and configures the heterogeneous multimodal resource sets with the editing decision to the resource library, meeting the update needs of different resource blocks and improving the update efficiency and reliability of the resource library.

[0085] In another embodiment, the resource repository change determination module is used to obtain the knowledge graph element change characteristics of the resource repository based on the resource update log of the heterogeneous multimodal resource repository, including:

[0086] The resource update logs of the heterogeneous multimodal resource library are analyzed to determine the resource update attributes of each resource block within the library. The resource update attributes refer to the location of resource changes, the type of resource change information, and the amount of resource change information in the resource block. Based on the resource update attributes, the entity and relation change characteristics of the corresponding knowledge graph of each resource block are determined.

[0087] The resource block identification module is used to identify resource blocks in the resource library that need to be changed based on the change characteristics of knowledge graph elements, including:

[0088] The number of abnormal entity changes and abnormal relationship changes in the knowledge graph are extracted from the entity and relation change features to determine whether the knowledge graph has semantic errors. The resource blocks corresponding to the knowledge graph with semantic errors are then marked as resource blocks that need to be changed.

[0089] The beneficial effects of the above embodiments are that the heterogeneous multimodal resource library includes multiple resource blocks, each of which stores heterogeneous multimodal resources formed in different time periods or from different resource providers. When any resource block in the resource library undergoes a heterogeneous multimodal resource change, the resource update log will generate corresponding log information. Simultaneously, the knowledge graph corresponding to the resource block will also experience changes in at least one element among entities, relationships, and attributes during the heterogeneous multimodal resource change process. When the updated heterogeneous multimodal resources received by a resource block contain errors and / or interference, the corresponding knowledge graph will have corresponding errors at the element level. These errors are more obvious at the knowledge graph level. When the knowledge graph shows errors after semantic recognition, it can be determined that the corresponding resource block has resource change errors, etc. In this case, the aforementioned resource block should be identified as a resource block that needs to be changed due to heterogeneous multimodal resource errors. Specifically, the resource update logs of the heterogeneous multimodal resource library are first analyzed to determine the location, type, and amount of resource changes for each resource block within the library. A correspondence exists between the data resources within each resource block and the elements of its corresponding knowledge graph. For example, when the content, structure, format, or data volume of a resource at a certain location within a resource block changes, the corresponding knowledge graph for that resource block will change at certain entity and relation levels. Therefore, based on the resource update status of the resource blocks and the correspondence between the resource blocks and the knowledge graph, the entity and relation change characteristics of the knowledge graph corresponding to the resource block are determined. These entity and relation change characteristics may include, but are not limited to, changes in entity names within the knowledge graph and changes in the relationship types between different entities. When an entity name in the knowledge graph becomes blank or an incorrect name, or when existing relationships between different entities in the knowledge graph disappear or relationships are formed between entities that did not originally have any relationship, these situations can be considered as abnormal entity or relation changes in the knowledge graph. The number of abnormal entity changes and abnormal relationship changes in the knowledge graph are extracted from the entity and relationship change features. If the number of abnormal entity changes or abnormal relationship changes exceeds a preset threshold, it is determined that the knowledge graph has a semantic error; otherwise, it is determined that the knowledge graph has no semantic error, and the resource block corresponding to the knowledge graph with the semantic error is marked as a resource block that needs to be changed. This facilitates subsequent heterogeneous multimodal resource change processing for the resource block that needs to be changed, reducing the workload of changing the resource library.

[0090] In another embodiment, the resource matching module is used to monitor resource sending samples from all resource providers connected to the resource library, thereby determining the matching attributes of each resource provider with the resource block to be changed, including:

[0091] Based on the resource transmission traffic of all resource providers connected to the resource repository, resource transmission samples are periodically sampled from each resource provider, and the semantic information of the resources provided by each resource provider is determined based on the resource transmission samples. The semantic information of the resources provided by each resource provider is compared with the original semantic information of the resource block to be modified to determine the matching attributes between each resource provider and the resource block to be modified. The matching attribute refers to the degree of similarity between the semantic information of the resources provided by each resource provider and the original semantic information of the resource block to be modified.

[0092] The resource orchestration module is used to differentiate and orchestrate heterogeneous multimodal resources from all resource providers based on matching attributes, resulting in several heterogeneous multimodal resource sets, including:

[0093] Based on the matching attributes, the priority level for differentiated collection and orchestration of heterogeneous multimodal resources from all resource providers is determined; based on the priority level, the heterogeneous multimodal resources from all resource providers are individually integrated to obtain several heterogeneous multimodal resource sets.

[0094] The beneficial effects of the above embodiments are that the heterogeneous multimodal resource library obtains heterogeneous multimodal resources from different resource providers. These resource providers can be, but are not limited to, smart terminals such as smartphones and drones. During their operation, these providers acquire heterogeneous multimodal resources through methods such as photography, user interaction, and internet data mining. The way each resource provider sends heterogeneous multimodal resources to the resource library differs, mainly reflected in the different resource transmission volumes and content (i.e., different amounts of resource data sent per unit time and different content of the sent resource data). Different resource blocks within the resource library correspond to different stored resource attributes (such as resource content, structure type, and format type). To ensure that the resources sent by the resource providers are accurately stored in the corresponding resource blocks, it is necessary to calibrate the resources provided by each resource provider. Specifically, based on the resource transmission traffic of each resource provider connected to the resource repository, resource transmission samples are periodically sampled from each resource provider, and the semantic information of the resources provided by the resource providers is determined based on the resource transmission samples. Then, by comparing the semantic information of the resources provided by each resource provider with the original semantic information of the resource block to be changed, the similarity between the semantic information of the resources provided by each resource provider and the original semantic information of the resource block to be changed is determined. Generally speaking, the higher the similarity, the higher the priority of the heterogeneous multimodal resources sent by the corresponding resource provider in collection and orchestration. Based on the above priority, the heterogeneous multimodal resources from all resource providers are individually integrated to obtain several heterogeneous multimodal resource sets, realizing targeted orchestration processing of heterogeneous multimodal resources from different resource providers, ensuring that subsequent heterogeneous multimodal resources can be accurately configured into the corresponding resource blocks in the resource repository.

[0095] In another embodiment, the resource editing decision module is used to determine the editing decision for the heterogeneous multimodal resource set based on the historical access characteristics of the resource block to be changed in the resource library, including:

[0096] Extract the historical access characteristics of the resource blocks to be modified from the resource repository's access logs; the historical access characteristics include the historical access paths and historical location keywords of the resource blocks to be modified; based on the historical access characteristics, determine the editing decision for the heterogeneous multimodal resource set; the editing decision includes editing decisions for at least one of the resource content, resource structure, and resource format;

[0097] The resource configuration module is used to configure heterogeneous multimodal resource sets that will be subject to editing decisions to the resource repository based on the real-time access status of the repository, including:

[0098] Based on the real-time access request volume of the resource repository, the available resource configuration bandwidth of the resource repository is determined, thereby converting the heterogeneous multimodal resource set that implements editing decisions into resource data streams with corresponding traffic and configuring them into the resource repository.

[0099] The beneficial effects of the above embodiments are that the resource data format of the resource blocks that need to be changed within the heterogeneous multimodal resource library is specific. In order to ensure that the heterogeneous multimodal resource set from the resource provider is compatible with the resource blocks that need to be changed in terms of resource data format, it is necessary to edit the heterogeneous multimodal resource set. Specifically, historical access features such as historical access paths and historical location keywords of the resource blocks that need to be changed in the resource library are extracted from the access log of the resource library. The resource data format corresponding to the resource blocks that need to be changed is determined through the above historical access features, thereby determining the editing decision for at least one of the resource content, resource structure, and resource format of the heterogeneous multimodal resource set, thereby improving the data format compatibility between the heterogeneous multimodal resource set and the resource blocks that need to be changed. Furthermore, based on the real-time access request volume of the resource repository, the available resource configuration bandwidth of the resource repository is determined. This allows the heterogeneous multimodal resource set implementing editing decisions to be converted into a resource data stream with corresponding traffic and configured into the resource repository, enabling resource changes to resource blocks within the resource repository that need to be modified. The traffic of the aforementioned resource data stream is matched with the available resource configuration bandwidth of the resource repository, ensuring that the resource data stream is transmitted quickly and stably to the resource blocks that need to be modified, thereby improving the efficiency and reliability of resource repository updates.

[0100] In summary, this heterogeneous multimodal resource orchestration and decision control method and system obtains the knowledge graph element change characteristics of the resource library based on the resource update logs of the heterogeneous multimodal resource library, thereby identifying the resource blocks in the resource library that need to be changed; it monitors the resource sending samples of all resource providers connected to the resource library to determine the matching attributes of each resource provider with the resource blocks that need to be changed, thereby distinguishing and orchestrating the heterogeneous multimodal resources from all resource providers to obtain several heterogeneous multimodal resource sets, achieving accurate resource docking and orchestration for different resource blocks; based on the historical access characteristics of the resource blocks that need to be changed in the resource library, it determines the editing decision for the heterogeneous multimodal resource sets, and configures the heterogeneous multimodal resource sets with the editing decision to the resource library, meeting the update needs of different resource blocks and improving the update efficiency and reliability of the resource library.

[0101] The above is only one specific embodiment of the present invention, and any improvements made based on the concept of the present invention shall be considered within the scope of protection of the present invention.

Claims

1. A heterogeneous multimodal resource orchestration and decision control method, characterized in that, include: Based on the resource update logs of the heterogeneous multimodal resource library, the change characteristics of the knowledge graph elements of the resource library are obtained; based on the change characteristics of the knowledge graph elements, the resource blocks of the resource library that need to be changed are identified. Listen to the resource sending samples of all resource providers connected to the resource library to determine the matching attributes of each resource provider with the resource block to be changed; based on the matching attributes, perform differentiated collection and orchestration of heterogeneous multimodal resources from all resource providers to obtain several heterogeneous multimodal resource sets; Based on the historical access characteristics of the resource blocks to be modified in the resource library, an editing decision is determined for the heterogeneous multimodal resource set; based on the real-time access status of the resource library, the heterogeneous multimodal resource set for which the editing decision is implemented is configured into the resource library.

2. The heterogeneous multimodal resource orchestration and decision control method as described in claim 1, characterized in that: Based on the resource update logs of the heterogeneous multimodal resource library, the change characteristics of the knowledge graph elements of the resource library are obtained; based on the change characteristics of the knowledge graph elements, the resource blocks of the resource library that need to be changed are identified, including: The resource update logs of the heterogeneous multimodal resource library are parsed to determine the resource update attributes of each resource block within the resource library. The resource update attributes refer to the resource change location, resource change information type, and resource change information volume of the resource block. Based on the resource update attributes, the entity and relation change features of the knowledge graph corresponding to each resource block are determined. The number of abnormal entity changes and the number of abnormal relationship changes that occurred in the knowledge graph are extracted from the entity and relationship change features. This is used to determine whether the knowledge graph has a semantic error, and the resource blocks corresponding to the knowledge graph with semantic errors are marked as resource blocks that need to be changed.

3. The heterogeneous multimodal resource orchestration and decision control method as described in claim 1, characterized in that: Listen to the resource sending samples of all resource providers connected to the resource library to determine the matching attributes of each resource provider with the resource block to be changed; Based on the matching attributes, heterogeneous multimodal resources from all resource providers are selectively collected and orchestrated to obtain several heterogeneous multimodal resource sets, including: Based on the resource transmission traffic of each resource provider connected to the resource library, resource transmission samples are periodically sampled from each resource provider, and the semantic information of the resources provided by each resource provider is determined based on the resource transmission samples. The semantic information of the resources provided by each resource provider is compared with the original semantic information of the resource block to be changed to determine the matching attribute between each resource provider and the resource block to be changed. The matching attribute refers to the degree of similarity between the semantic information of the resources provided by each resource provider and the original semantic information of the resource block to be changed. Based on the matching attributes, a priority level is determined for the differentiated acquisition and orchestration of heterogeneous multimodal resources from all resource providers; based on the priority level, the heterogeneous multimodal resources from all resource providers are individually integrated to obtain several heterogeneous multimodal resource sets.

4. The heterogeneous multimodal resource orchestration and decision control method as described in claim 1, characterized in that: Based on the historical access characteristics of the resource blocks to be changed in the resource library, an editing decision is made for the heterogeneous multimodal resource set; Based on the real-time access status of the resource repository, the heterogeneous multimodal resource set for implementing the editing decision is configured into the resource repository, including: Extract the historical access characteristics of the resource block to be modified from the access log of the resource repository; wherein, the historical access characteristics include the historical access path and historical location keywords of the resource block to be modified; based on the historical access characteristics, determine the editing decision for the heterogeneous multimodal resource set; wherein, the editing decision includes editing decisions for at least one of resource content, resource structure, and resource format; Based on the real-time access request volume of the resource library, the available resource configuration bandwidth of the resource library is determined, thereby converting the heterogeneous multimodal resource set implementing the editing decision into a resource data stream with corresponding traffic and configuring it into the resource library.

5. A heterogeneous multimodal resource orchestration and decision-making control system, characterized in that, include: The resource library change determination module is used to obtain the knowledge graph element change characteristics of the resource library based on the resource update log of the heterogeneous multimodal resource library; The resource block identification module is used to identify the resource blocks in the resource library that need to be changed based on the change characteristics of the knowledge graph elements. The resource matching module is used to monitor the resource sending samples of all resource providers connected to the resource library, so as to determine the matching attributes of each resource provider with the resource block to be changed; The resource orchestration module is used to differentiate, collect, and orchestrate heterogeneous multimodal resources from all resource providers based on the matching attributes, thereby obtaining several heterogeneous multimodal resource sets. The resource editing decision module is used to determine the editing decision for the heterogeneous multimodal resource set based on the historical access characteristics of the resource block to be changed in the resource library. The resource configuration module is used to configure the heterogeneous multimodal resource set that implements the editing decision into the resource library based on the real-time access status of the resource library.

6. The heterogeneous multimodal resource orchestration and decision control system as described in claim 5, characterized in that: The resource library change determination module is used to obtain the knowledge graph element change characteristics of the resource library based on the resource update logs of the heterogeneous multimodal resource library, including: The resource update logs of the heterogeneous multimodal resource library are parsed to determine the resource update attributes of each resource block within the resource library. The resource update attributes refer to the resource change location, resource change information type, and resource change information volume of the resource block. Based on the resource update attributes, the entity and relation change features of the knowledge graph corresponding to each resource block are determined. The modified resource block identification module is used to identify the resource blocks in the resource library that need to be modified based on the modification characteristics of the knowledge graph elements, including: The number of abnormal entity changes and the number of abnormal relationship changes that occurred in the knowledge graph are extracted from the entity and relationship change features. This is used to determine whether the knowledge graph has a semantic error, and the resource blocks corresponding to the knowledge graph with semantic errors are marked as resource blocks that need to be changed.

7. The heterogeneous multimodal resource orchestration and decision control system as described in claim 5, characterized in that: The resource matching module is used to monitor resource sending samples from all resource providers connected to the resource library, thereby determining the matching attributes of each resource provider with the resource block to be changed, including: Based on the resource transmission traffic of each resource provider connected to the resource library, resource transmission samples are periodically sampled from each resource provider, and the semantic information of the resources provided by each resource provider is determined based on the resource transmission samples. The semantic information of the resources provided by each resource provider is compared with the original semantic information of the resource block to be changed to determine the matching attribute between each resource provider and the resource block to be changed. The matching attribute refers to the degree of similarity between the semantic information of the resources provided by each resource provider and the original semantic information of the resource block to be changed. The resource orchestration module is used to differentiate, collect, and orchestrate heterogeneous multimodal resources from all resource providers based on the matching attributes, resulting in several heterogeneous multimodal resource sets, including: Based on the matching attributes, a priority level is determined for the differentiated acquisition and orchestration of heterogeneous multimodal resources from all resource providers; based on the priority level, the heterogeneous multimodal resources from all resource providers are individually integrated to obtain several heterogeneous multimodal resource sets.

8. The heterogeneous multimodal resource orchestration and decision control system as described in claim 5, characterized in that: The resource editing decision module is used to determine the editing decision for the heterogeneous multimodal resource set based on the historical access characteristics of the resource block to be changed in the resource library, including: Extract the historical access characteristics of the resource block to be modified from the access log of the resource repository; wherein, the historical access characteristics include the historical access path and historical location keywords of the resource block to be modified; based on the historical access characteristics, determine the editing decision for the heterogeneous multimodal resource set; wherein, the editing decision includes editing decisions for at least one of resource content, resource structure, and resource format; The resource configuration module is used to configure the heterogeneous multimodal resource set that implements the editing decision to the resource library according to the real-time access status of the resource library, including: Based on the real-time access request volume of the resource library, the available resource configuration bandwidth of the resource library is determined, thereby converting the heterogeneous multimodal resource set implementing the editing decision into a resource data stream with corresponding traffic and configuring it into the resource library.

Images