Knowledge graph-based information processing method and system

Abstract

The present application provides a knowledge graph-based information processing method and a system. The method comprises: acquiring a target operation scheduling knowledge graph and loading same to an entity recognition network, and generating a target entity definition domain associated with each target node; and on the basis of the target entity definition domain, recognizing a target entity from the target operation scheduling knowledge graph. Network parameter learning is performed on the basis of a first knowledge reasoning error, such that the entity recognition network can recognize entities more accurately. In this way, by using the knowledge graph and the entity recognition network, the accuracy of entity recognition is improved; then, a new method for calculating a knowledge reasoning error is used, such that the learning process of network parameters is optimized, improving network performance; and finally, a priori operation scheduling knowledge graph and a restructuring operation scheduling knowledge graph are used, such that entity recognition aligns more closely with preset knowledge association parameters, further improving the accuracy and efficiency of entity recognition.

Description

Knowledge Graph-Based Information Processing Methods and Systems Technical Field

[0001] This application relates to the field of artificial intelligence technology, and more specifically, to an information processing method and system based on knowledge graphs. Background Technology

[0002] In today's information age, the volume of data is growing rapidly, and effectively processing this data and extracting useful information from it has become a significant challenge. Knowledge graphs, as a method for representing and using knowledge, have been widely applied in fields such as information retrieval, intelligent recommendation, and semantic search. For example, in the field of computational scheduling, knowledge graphs can describe complex computational tasks, resources, and their dependencies, facilitating efficient resource allocation and task scheduling.

[0003] Traditional knowledge graphs typically rely on predefined rules or patterns for information extraction and entity recognition, but this approach has certain limitations. First, predefined rules or patterns may not cover all entity and relation types, leading to the omission of some important information. Second, the lack of effective mechanisms to handle the uncertainty and ambiguity in knowledge graphs affects the accuracy of the results. Finally, as knowledge graphs expand, processing complexity and computational costs increase, reducing system performance. In other words, traditional solutions often rely on manually defined rules and features, making them poorly adaptable to complex and ever-changing computational scheduling scenarios. Furthermore, these methods face problems such as high computational cost and low recognition accuracy when processing large-scale knowledge graphs. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide an information processing method and system based on knowledge graphs.

[0005] According to a first aspect of this application, a knowledge graph-based information processing method is provided, the method comprising:

[0006] Obtain the target computation scheduling knowledge graph and load the target computation scheduling knowledge graph into the entity recognition network to generate the target entity delimitation domain associated with each target node of the target computation scheduling knowledge graph;

[0007] Based on the target entity definition domain, the target entity is identified from the target computation scheduling knowledge graph. The entity recognition network learns network parameters based on a first knowledge inference error. The first knowledge inference error is calculated based on the first estimated definition domain of each first node in the first paradigm computation scheduling knowledge graph, which is identified by the entity recognition network, and the first labeled entity definition domain of each first node. At least two first paradigm computation scheduling knowledge graphs include a prior computation scheduling knowledge graph and at least two reconstructed computation scheduling knowledge graphs generated by applying at least two differentiated knowledge reconstruction strategies to the first prior entity in the prior computation scheduling knowledge graph. The knowledge association parameters of the first prior entity between at least two first paradigm computation scheduling knowledge graphs meet a first preset requirement.

[0008] In one possible implementation of the first aspect, the step of generating the entity recognition network includes:

[0009] Obtain a first example operation scheduling knowledge graph sequence, the first example operation scheduling knowledge graph sequence includes at least two first example operation scheduling knowledge graphs, the at least two first example operation scheduling knowledge graphs include the prior operation scheduling knowledge graph and at least two reconstructed operation scheduling knowledge graphs generated by applying at least two differentiated knowledge reconstruction strategies to the first prior entity in the prior operation scheduling knowledge graph, the knowledge association parameters of the first prior entity between at least two first example operation scheduling knowledge graphs meet the first preset requirements, and each first example operation scheduling knowledge graph has a first labeled entity delimitation domain associated with each first node;

[0010] Each first example operation scheduling knowledge graph is loaded into the entity recognition network to generate the first estimated bounding domain associated with each first node of the first example operation scheduling knowledge graph.

[0011] Based on the first example, the first estimated bounding domain and the first labeled entity bounding domain associated with each first node of the knowledge graph are calculated and scheduled, and the first knowledge reasoning error is output.

[0012] Based on the first knowledge inference error, network parameters are learned for the entity recognition network.

[0013] In one possible implementation of the first aspect, the step of learning network parameters for the entity recognition network based on the first knowledge inference error includes:

[0014] Obtain a second paradigm operation scheduling knowledge graph sequence, the second paradigm operation scheduling knowledge graph sequence includes at least two second paradigm operation scheduling knowledge graphs parsed from the same operation scheduling knowledge graph, the second paradigm operation scheduling knowledge graph includes a second prior entity, and the second paradigm operation scheduling knowledge graph has a second labeled entity delimitation domain associated with each second node in the second paradigm operation scheduling knowledge graph;

[0015] Each second paradigm operation scheduling knowledge graph is loaded into the entity recognition network to generate a second estimated bounding domain associated with each second node of the second paradigm operation scheduling knowledge graph;

[0016] Based on the second example, the second estimated bounding domain and the second labeled entity bounding domain associated with each second node of the knowledge graph are operated and scheduled, and the second knowledge reasoning error is output.

[0017] Based on the first knowledge inference error and the second knowledge inference error, network parameters are learned for the entity recognition network.

[0018] In one possible implementation of the first aspect, the step of generating the second labeled entity delimitation region includes:

[0019] Each second paradigm operation scheduling knowledge graph is loaded into the prior recognition network of the entity recognition network to generate a third estimated bounding domain associated with each second node of the second paradigm operation scheduling knowledge graph, and the third estimated bounding domain is used as the second labeled entity bounding domain.

[0020] In one possible implementation of the first aspect, the step of generating the prior recognition network includes:

[0021] Obtain a third-paradigm operation scheduling knowledge graph sequence, the third-paradigm operation scheduling knowledge graph sequence including at least two third-paradigm operation scheduling knowledge graphs, the third-paradigm operation scheduling knowledge graphs including third prior entities, the third-paradigm operation scheduling knowledge graphs having third labeled entity delimitations associated with each third node in the third-paradigm operation scheduling knowledge graphs, and the count of the third-paradigm operation scheduling knowledge graphs being greater than the count of the first-paradigm operation scheduling knowledge graphs;

[0022] Each of the third paradigm operation scheduling knowledge graphs is loaded into the prior recognition network to generate a fourth estimated bounding domain associated with each of the third nodes of the third paradigm operation scheduling knowledge graph.

[0023] Based on the fourth estimated bounding domain and the third labeled entity bounding domain associated with each of the third nodes of the third paradigm operation scheduling knowledge graph, the third knowledge reasoning error is output.

[0024] Based on the third knowledge inference error, network parameters are learned for the prior recognition network.

[0025] In one possible implementation of the first aspect, obtaining the first paradigm operation scheduling knowledge graph sequence includes:

[0026] Obtain the prior computation scheduling knowledge graph, which includes the first prior entity and has the first annotation entity delimitation domain associated with each first node;

[0027] The knowledge reconstruction strategy under the same stage is applied to the first prior entity and the first labeled entity bounding domain in the prior operation scheduling knowledge graph to generate the reconstructed operation scheduling knowledge graph and the associated first labeled entity bounding domain. The knowledge association parameters of the first prior entity between the prior operation scheduling knowledge graph and the reconstructed operation scheduling knowledge graph meet the first preset requirements.

[0028] The knowledge reconstruction strategy under the same stage is applied to the first prior entity and the first labeled entity in the reconstructed operation scheduling knowledge graph to generate an iterative reconstructed operation scheduling knowledge graph. The knowledge association parameters of the first prior entity in the associated reconstructed operation scheduling knowledge graphs meet the first preset requirements.

[0029] The steps of applying the same knowledge reconstruction strategy to the first prior entity and the first labeled entity in the reconstructed operation scheduling knowledge graph and generating iterative reconstructed operation scheduling knowledge graphs are executed repeatedly until the count of the reconstructed operation scheduling knowledge graph meets the second preset requirement. Based on the prior operation scheduling knowledge graph and at least two reconstructed operation scheduling knowledge graphs, a first paradigm operation scheduling knowledge graph sequence is determined.

[0030] In one possible implementation of the first aspect, the step of determining whether the knowledge association parameters of the first prior entity meet the first preset requirement among at least two first paradigm operation scheduling knowledge graphs includes:

[0031] For each prior node of the first prior entity in the prior operation scheduling knowledge graph, obtain the link where the prior node is located in at least two first example operation scheduling knowledge graphs.

[0032] If, for each prior node, the deviation between the prior node and the link where the prior node is located in any two first example operation scheduling knowledge graphs is less than a first set deviation, then the output knowledge association parameters of the first prior entity between at least two first example operation scheduling knowledge graphs meet the first preset requirements.

[0033] In one possible implementation of the first aspect, the step of determining whether the knowledge association parameters of the first prior entity meet the first preset requirement among at least two first paradigm operation scheduling knowledge graphs includes:

[0034] Each first example computation scheduling knowledge graph is decomposed into computation scheduling knowledge units, and unit labels are assigned to the computation scheduling knowledge units according to the participating links of the computation scheduling knowledge units.

[0035] If the first prior entity is located in the same computation scheduling knowledge unit associated with the same unit label in each of the first paradigm computation scheduling knowledge graphs, the output of the knowledge association parameters of the first prior entity between at least two first paradigm computation scheduling knowledge graphs shall meet the first preset requirement.

[0036] In one possible implementation of the first aspect, the target computation scheduling knowledge graph includes multiple uninterrupted target computation scheduling knowledge graphs, the target entity delimitation domain includes a target entity delimitation subdomain of each of the target computation scheduling knowledge graphs, and the step of loading the target computation scheduling knowledge graphs into an entity recognition network to generate target entity delimitation domains associated with each target node of the target computation scheduling knowledge graph includes:

[0037] For the target operation scheduling knowledge graph traversed in this round, obtain the target operation scheduling knowledge graph generated in the forward direction and the target entity delimitation subdomain of the target operation scheduling knowledge graph generated in the forward direction;

[0038] The target computation scheduling knowledge graph traversed in this round, the target computation scheduling knowledge graph generated in the forward pass, and the target entity delimitation subdomain of the target computation scheduling knowledge graph generated in the forward pass are loaded into the entity recognition network to generate the target entity delimitation subdomain associated with each target node of the target computation scheduling knowledge graph traversed in this round.

[0039] According to a second aspect of this application, a computational query system is provided, the computational query system including a machine-readable storage medium and a processor, the machine-readable storage medium storing machine-executable instructions, and the computational query system implementing the aforementioned knowledge graph-based information processing method when the processor executes the machine-executable instructions.

[0040] According to a third aspect of this application, a computer-readable storage medium is provided, wherein computer-executable instructions are stored therein, and when the computer-executable instructions are executed, the aforementioned knowledge graph-based information processing method is implemented.

[0041] Based on any of the above aspects, the technical effect of this application is as follows:

[0042] This application achieves accurate entity recognition by acquiring the target computation scheduling knowledge graph and loading it into an entity recognition network, generating target entity delimitations associated with each target node. Furthermore, it identifies target entities from the knowledge graph based on these target entity delimitations, improving the accuracy and efficiency of entity recognition. In addition, the entity recognition network learns network parameters based on a first knowledge inference error, calculated from the difference between the estimated delimitations of nodes in multiple example computation scheduling knowledge graphs and the labeled entity delimitations, thereby enhancing the network's ability to learn entity delimitations. At least two example knowledge graphs include prior knowledge graphs and reconstructed knowledge graphs generated using different knowledge reconstruction strategies, and the knowledge association parameters of prior entities between these example knowledge graphs meet preset requirements, further enriching the diversity and representativeness of the learning samples and improving the generalization ability and robustness of the entity recognition network.

[0043] In other words, this application generates target entity delimitations associated with each target node by acquiring the target computation scheduling knowledge graph and loading it into the entity recognition network. Based on these target entity delimitations, target entities can be identified from the target computation scheduling knowledge graph. By learning network parameters based on the first knowledge inference error, the entity recognition network can more accurately identify entities. Thus, by utilizing the knowledge graph and the entity recognition network, the accuracy of entity recognition is improved. Secondly, a novel method for calculating knowledge inference error is adopted, which helps optimize the network parameter learning process and improves network performance. Finally, the use of a priori computation scheduling knowledge graphs and reconstructed computation scheduling knowledge graphs makes entity recognition more consistent with preset knowledge association parameters, further improving the accuracy and efficiency of entity recognition. Attached Figure Description

[0044] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 is a schematic flowchart of the knowledge graph-based information processing method provided in an embodiment of this application;

[0046] Figure 2 shows a schematic diagram of the component structure of the computational query system for implementing the above-described knowledge graph-based information processing method provided in an embodiment of this application. Detailed Implementation

[0047] The embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the embodiments described below with reference to the accompanying drawings are exemplary descriptions for explaining the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions of the embodiments of this application.

[0048] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the terms “comprising” and “including” as used in embodiments of this application mean that the corresponding feature can be implemented as the presented feature, information, data, step, operation, element, and / or component, but do not exclude implementation as other features, information, data, step, operation, element, component, and / or combinations thereof supported by the art. It should be understood that when an element is said to be “connected” or “coupled” to another element, the element may be directly connected or coupled to the other element, or it may mean that the element and the other element are connected through an intermediate element. Furthermore, “connected” or “coupled” as used herein may include wireless connection or wireless coupling. The term “and / or” as used herein indicates at least one of the items defined by the term; for example, “A and / or B” can be implemented as “A,” or as “B,” or as “A and B.”

[0049] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings. The technical solutions of the embodiments of this application and the technical effects generated by the technical solutions of this application will be explained below through the description of several exemplary embodiments. It should be noted that the following embodiments can be referenced, borrowed from, or combined with each other, and the same terms, similar features, and similar implementation steps in different embodiments will not be repeated.

[0050] Figure 1 shows a flowchart of the knowledge graph-based information processing method and system provided in an embodiment of this application. It should be understood that in other embodiments, the order of some steps in the knowledge graph-based information processing method of this embodiment can be shared according to actual needs, or some steps can be omitted or maintained. The detailed steps of the knowledge graph-based information processing method include:

[0051] Step S110: Obtain the target computation scheduling knowledge graph and load the target computation scheduling knowledge graph into the entity recognition network to generate the target entity delimitation domain associated with each target node of the target computation scheduling knowledge graph.

[0052] In this embodiment, it is assumed that the computation query system is executing a complex computation scheduling task, which involves a large amount of data computation and scheduling decisions. To improve the efficiency and accuracy of computation scheduling, the computation query system needs to utilize knowledge graph technology to assist in decision-making.

[0053] For example, the computation query system first retrieves the data structure of the target computation scheduling knowledge graph from its internal storage or external data source. This target computation scheduling knowledge graph is a complex network composed of nodes and edges, where nodes represent various computation entities (such as data, algorithms, computing resources, etc.) and edges represent the relationships between these entities (such as data dependencies, computation order, etc.).

[0054] Then, the computation query system loads the acquired target computation scheduling knowledge graph into a pre-trained entity recognition network. This entity recognition network is a deep learning model capable of parsing and learning the input knowledge graph to identify various entities within it. Specifically, the entity recognition network generates a target entity delimitation domain for each target node in the target computation scheduling knowledge graph through learning. This target entity delimitation domain is an abstract conceptual space used to describe and define the characteristics and attributes of the entity represented by the target node.

[0055] In detail, the computation query system can retrieve and extract knowledge graphs related to computation scheduling from specified data sources. For example, the computation query system downloads a knowledge graph file named "computation scheduling strategy" from an internal database or external cloud storage. This knowledge graph file contains nodes such as computation tasks, computing resources, and scheduling rules, as well as the relationships between them. The target computation scheduling knowledge graph can refer to the knowledge graph in the computation scheduling domain that the computation query system needs to process and analyze. For example, in the computation scheduling system of the computation query system, there is a knowledge graph that specifically describes how different computation tasks are allocated to various computing nodes, as well as the dependencies and execution order between these tasks. Therefore, the computation query system runs a deep learning framework, such as TensorFlow or PyTorch, loads a pre-trained entity recognition network into it, and passes the target computation scheduling knowledge graph as input data to this entity recognition network. The entity recognition network is a neural network model that can automatically identify and classify entities in the knowledge graph. It can detect nodes in the knowledge graph and identify them as different entity types, such as computation tasks and computing resources, based on their features and attributes. That is, the entity recognition network generates a delineation space or range for each target node in the knowledge graph, describing its features and attributes.

[0056] For a node "matrix multiplication operation" in the target computation scheduling knowledge graph, the entity recognition network generates a delimitation domain. This domain may include feature information such as the dimension of the operation, data type, and required computing resources. The target node can refer to a node in the target computation scheduling knowledge graph that is of interest to a specific operation or analysis. For example, in the computation scheduling knowledge graph, a computation query system may pay particular attention to nodes representing large-scale computational tasks or critical computing resources; these nodes are target nodes. Therefore, the target entity delimitation domain is generated for the target node and is used to describe and define the specific range and feature space of the entity represented by that node. For example, for the target node "high-performance computing computation query system," its target entity delimitation domain may include detailed information such as the model, configuration, computing power, and storage capacity of the computation query system.

[0057] Step S120: Identify the target entity from the target operation scheduling knowledge graph based on the target entity definition domain.

[0058] In this embodiment, the entity recognition network learns network parameters based on a first knowledge inference error. The first knowledge inference error is calculated based on the first estimated bounding domain of each first node in the first paradigm operation scheduling knowledge graph, which is identified by the entity recognition network, and the first labeled entity bounding domain of each first node. The at least two first paradigm operation scheduling knowledge graphs include a prior operation scheduling knowledge graph and at least two reconstructed operation scheduling knowledge graphs generated by applying at least two differentiated knowledge reconstruction strategies to the first prior entity in the prior operation scheduling knowledge graph. The knowledge association parameters of the first prior entity between the at least two first paradigm operation scheduling knowledge graphs meet a first preset requirement.

[0059] The target entity refers to an entity node in the target computation scheduling knowledge graph that is of particular interest or needs to be identified. For example, suppose there is a node in the target computation scheduling knowledge graph that represents a video encoding / decoding task. During the computation scheduling process of the computation query system, this task node may be a target entity that needs special attention.

[0060] The first knowledge inference error refers to the error generated by the entity recognition network when recognizing nodes in the paradigm computation scheduling knowledge graph. It is used to measure the difference between the accuracy of the network's recognition and the actual annotation. For example, the entity recognition network attempts to recognize a node "image rendering task" in the paradigm computation scheduling knowledge graph and generates an estimated bounding domain for it; at the same time, the node has an labeled entity bounding domain as the standard answer. By comparing the difference between these two bounding domains, the first knowledge inference error can be calculated.

[0061] The first example computation scheduling knowledge graph is a standard knowledge graph used to train or validate the performance of entity recognition networks, containing pre-annotated entity delimitation information. For example, a computation query system may possess one or more pre-annotated computation scheduling knowledge graphs, which are used as training data to help the entity recognition network learn how to accurately identify different entities in the graph.

[0062] The first node refers to a node in the first paradigm computation scheduling knowledge graph, which is usually a node that has been labeled with an entity delimiter. For example, in the first paradigm computation scheduling knowledge graph, each node representing a computation task, computing resource, etc., is considered a first node, and they all have corresponding labeled entity delimiters.

[0063] The first estimated entity bounding domain refers to the estimated entity bounding domain generated by the entity recognition network for the first node in the first paradigm operation scheduling knowledge graph. For example, when the entity recognition network processes a node in the first paradigm operation scheduling knowledge graph, it generates an estimated entity bounding domain based on the node's features and attributes. This bounding domain may differ from the labeled entity bounding domain.

[0064] The first labeled entity definition domain refers to the correct entity definition domain pre-labeled for the first node in the first paradigm operation scheduling knowledge graph. For example, in the first paradigm operation scheduling knowledge graph, each node has a corresponding entity definition domain labeled by experts or standard procedures, and this labeled entity definition domain is used as a standard to measure the performance of the entity recognition network.

[0065] The prior computation scheduling knowledge graph refers to an existing computation scheduling knowledge graph that contains prior knowledge, and it is usually used as the basis for generating and reconstructing computation scheduling knowledge graphs. For example, a computation query system may have a computation scheduling knowledge graph built based on historical data and expert experience. This graph contains a large amount of prior knowledge and serves as the basis for building other knowledge graphs.

[0066] The knowledge reconstruction strategy refers to the strategy of transforming or reorganizing entities and relations in the prior computation scheduling knowledge graph to generate a new reconstructed computation scheduling knowledge graph. For example, a computation query system can use different knowledge reconstruction strategies, such as entity replacement and relation adjustment, to modify certain parts of the prior graph, thereby generating multiple different versions of the reconstructed graph.

[0067] The reconstructed computation scheduling knowledge graph refers to a new knowledge graph generated by applying knowledge reconstruction strategies to a prior computation scheduling knowledge graph. For example, a computation query system can generate multiple reconstructed computation scheduling knowledge graphs, each of which is a different transformation or reorganization of the prior graph, used to increase data diversity and the generalization ability of the model.

[0068] The first prior entity refers to an entity that is given special attention or used as the basis for knowledge reconstruction in the prior operation scheduling knowledge graph. For example, in the first prior operation scheduling knowledge graph, certain key operation tasks or computing resource nodes may be selected as the first prior entities, and the operation query system will apply knowledge reconstruction strategies around these entities.

[0069] The knowledge association parameters are parameters that describe the relationships or similarities between entities in the first paradigm computation scheduling knowledge graph. For example, when a computation query system compares similar entities in different knowledge graphs, it may consider parameters such as structural similarity and semantic association between them; these parameters are the knowledge association parameters.

[0070] The first preset requirement is a specific standard or threshold set for the knowledge association parameters, used to determine whether the association between the first prior entity and different knowledge graphs meets the requirements. For example, a computational query system may set a threshold, and only when the knowledge association parameters of similar entities in two knowledge graphs exceed this threshold are they considered to have sufficient association.

[0071] In this embodiment, the computation query system has generated a corresponding target entity delimitation domain for each node in the target computation scheduling knowledge graph. Next, the computation query system needs to use these target entity delimitation domains to identify the target entities in the target computation scheduling knowledge graph.

[0072] For example, the computation query system identifies the corresponding target entity from the target computation scheduling knowledge graph based on the target entity delimitation domain of each node. This process is similar to pattern matching or searching within the target computation scheduling knowledge graph; the computation query system searches for nodes that match the target entity delimitation domain. Before performing entity recognition, the entity recognition network can learn and adjust its network parameters based on the first knowledge inference error metric. This first knowledge inference error is calculated by comparing the entity delimitation domain identified by the entity recognition network (i.e., the first estimated delimitation domain) with the entity delimitation domain labeled in the knowledge graph (i.e., the first labeled entity delimitation domain).

[0073] To improve the accuracy and robustness of entity recognition, the computational query system utilizes at least two first-paradigm computational scheduling knowledge graphs for assisted learning. These first-paradigm computational scheduling knowledge graphs include a prior computational scheduling knowledge graph and multiple reconstructed computational scheduling knowledge graphs generated by applying different knowledge reconstruction strategies to the prior computational scheduling knowledge graph. Entities in these first-paradigm computational scheduling knowledge graphs possess knowledge association parameters that meet a first preset requirement between the graphs, meaning they are similar or consistent in certain aspects (such as structure, semantics, etc.).

[0074] The computation query system compares the difference between the labeled entity bounding domains in the first-paradigm computation scheduling knowledge graph and the estimated bounding domains identified by the entity recognition network, thereby calculating the first knowledge inference error. This first knowledge inference error reflects the accuracy of the entity recognition network in recognizing entities in the first-paradigm computation scheduling knowledge graph and is an important basis for network parameter learning.

[0075] Based on the above steps, this application acquires the target computation scheduling knowledge graph and loads it into the entity recognition network to generate target entity delimitations associated with each target node, thus achieving accurate entity recognition in the knowledge graph. Furthermore, it identifies target entities from the knowledge graph based on the target entity delimitations, improving the accuracy and efficiency of entity recognition. In addition, the entity recognition network learns network parameters based on a first knowledge inference error, calculated based on the difference between the estimated delimitations of nodes in multiple example computation scheduling knowledge graphs and the labeled entity delimitations, thereby enhancing the entity recognition network's ability to learn entity delimitations. At least two example knowledge graphs include prior knowledge graphs and reconstructed knowledge graphs generated using different knowledge reconstruction strategies, and the knowledge association parameters of prior entities between these example knowledge graphs meet preset requirements, further enriching the diversity and representativeness of the learning samples and improving the generalization ability and robustness of the entity recognition network.

[0076] In other words, this application generates target entity delimitations associated with each target node by acquiring the target computation scheduling knowledge graph and loading it into the entity recognition network. Based on these target entity delimitations, target entities can be identified from the target computation scheduling knowledge graph. By learning network parameters based on the first knowledge inference error, the entity recognition network can more accurately identify entities. Thus, by utilizing the knowledge graph and the entity recognition network, the accuracy of entity recognition is improved. Secondly, a novel method for calculating knowledge inference error is adopted, which helps optimize the network parameter learning process and improves network performance. Finally, the use of a priori computation scheduling knowledge graphs and reconstructed computation scheduling knowledge graphs makes entity recognition more consistent with preset knowledge association parameters, further improving the accuracy and efficiency of entity recognition.

[0077] In one possible implementation, the generation step of the entity recognition network includes:

[0078] Step A110: Obtain a first paradigm operation scheduling knowledge graph sequence. The first paradigm operation scheduling knowledge graph sequence includes at least two first paradigm operation scheduling knowledge graphs. The at least two first paradigm operation scheduling knowledge graphs include the prior operation scheduling knowledge graph and at least two reconstructed operation scheduling knowledge graphs generated by applying at least two differentiated knowledge reconstruction strategies to the first prior entity in the prior operation scheduling knowledge graph. The knowledge association parameters of the first prior entity between the at least two first paradigm operation scheduling knowledge graphs meet the first preset requirements. Each first paradigm operation scheduling knowledge graph has a first labeled entity delimitation domain associated with each first node.

[0079] In this embodiment, the computation query system retrieves a series of first-paradigm computation scheduling knowledge graphs from its internal storage or external data sources. These first-paradigm computation scheduling knowledge graphs include a priori computation scheduling knowledge graph and multiple reconstructed computation scheduling knowledge graphs. Each first-paradigm computation scheduling knowledge graph represents a different aspect or scenario in the computation scheduling domain, and they are interconnected through first priori entities. The knowledge association parameters of these first priori entities across different graphs all conform to preset standards, indicating that they have consistency and importance in the computation scheduling context.

[0080] Step A120: Load each of the first paradigm operation scheduling knowledge graphs into the entity recognition network to generate the first estimated bounding domain associated with each of the first nodes of the first paradigm operation scheduling knowledge graph.

[0081] The computation query system loads each first-paradigm computation scheduling knowledge graph it acquires into the entity recognition network. The entity recognition network then scans and analyzes each node in the first-paradigm computation scheduling knowledge graph, attempting to understand the meaning of each node and its position within the entire first-paradigm computation scheduling knowledge graph.

[0082] For each first instance computational scheduling knowledge graph loaded into the entity recognition network, the computational query system uses its built-in algorithm to generate an estimated bounding domain for each first node. For example, the computational query system generates this estimated bounding domain based on the node's attributes, connectivity, and contextual information within the entire knowledge graph.

[0083] Step A130: Based on the first example, calculate the first estimated bounding domain and the first labeled entity bounding domain associated with each first node of the knowledge graph, and output the first knowledge reasoning error.

[0084] In this embodiment, the computational query system compares the estimated bounding domain of each first node with its corresponding labeled entity bounding domain, calculates the difference or error between them, and uses it as the first knowledge reasoning error. This error information is then used to measure the performance of the entity recognition network in recognizing graph nodes.

[0085] Step A140: Based on the first knowledge inference error, perform network parameter learning on the entity recognition network.

[0086] Based on the calculated initial knowledge inference error, the computational query system adjusts and optimizes the parameters of its internal entity recognition network. For example, machine learning techniques such as backpropagation can be used to update the weights and configuration of the entity recognition network to more accurately identify entities in the graph during future computational scheduling tasks.

[0087] In one possible implementation, step A140 may include:

[0088] Step A141: Obtain the second paradigm operation scheduling knowledge graph sequence. The second paradigm operation scheduling knowledge graph sequence includes at least two second paradigm operation scheduling knowledge graphs parsed from the same operation scheduling knowledge graph. The second paradigm operation scheduling knowledge graph includes a second prior entity. The second paradigm operation scheduling knowledge graph has a second labeled entity delimitation domain associated with each second node in the second paradigm operation scheduling knowledge graph.

[0089] In this embodiment, the computation query system retrieves a sequence of second-paradigm computation scheduling knowledge graphs from internal storage or external data sources. These second-paradigm computation scheduling knowledge graphs differ from the previous first-paradigm computation scheduling knowledge graphs, but they also originate from the same computation scheduling domain and are related. Each second-paradigm computation scheduling knowledge graph contains specific second prior entities, which occupy key positions within the second-paradigm computation scheduling knowledge graph, and each second-paradigm computation scheduling knowledge graph provides accurate second-label entity delimitations for these second prior entities.

[0090] Step A142: Load each of the second paradigm operation scheduling knowledge graphs into the entity recognition network to generate a second estimated bounding domain associated with each of the second nodes of the second paradigm operation scheduling knowledge graph.

[0091] In this embodiment, the computation query system sequentially loads each acquired second-paradigm computation scheduling knowledge graph into the pre-trained entity recognition network. This loading process ensures the network is exposed to more computation scheduling scenarios and entity types, thereby enhancing its generalization ability. The entity recognition network analyzes and processes each node (i.e., the second node) in each newly loaded second-paradigm computation scheduling knowledge graph. For each second-paradigm computation scheduling knowledge graph loaded into the entity recognition network, the computation query system uses its built-in algorithm to generate a second estimated bounding domain for each second node. This estimated bounding domain is generated based on the network's current learning state and understanding of node features, and it may differ from the actual labeled bounding domain.

[0092] Step A143: Based on the second example, the second estimated bounding domain and the second labeled entity bounding domain associated with each second node of the knowledge graph are calculated and scheduled, and the second knowledge reasoning error is output.

[0093] The computational query system compares the second estimated bounding domain of each second node with its corresponding second labeled entity bounding domain to calculate the second knowledge inference error. This error reflects the accuracy of the entity recognition network in processing the second paradigm computational scheduling knowledge graph and provides crucial information for subsequent network parameter learning.

[0094] Step A144: Based on the first knowledge inference error and the second knowledge inference error, perform network parameter learning on the entity recognition network.

[0095] By combining the first knowledge inference error calculated on the first paradigm's computational scheduling knowledge graph and the second knowledge inference error calculated on the second paradigm's computational scheduling knowledge graph, the computational query system adjusts and optimizes the entity recognition network's parameters. This process aims to improve the network's overall performance in recognizing entities in the computational scheduling knowledge graph by integrating error information from multiple sources. The computational query system may employ optimization algorithms such as gradient descent to update the network's weights and configuration based on the magnitude and direction of the error, thereby achieving more accurate entity recognition in future computational scheduling tasks.

[0096] In one possible implementation, the step of generating the second labeled entity delimitation domain includes: loading each of the second paradigm operation scheduling knowledge graphs into the prior recognition network of the entity recognition network, generating a third estimated delimitation domain associated with each of the second nodes of the second paradigm operation scheduling knowledge graph, and using the third estimated delimitation domain as the second labeled entity delimitation domain.

[0097] In this embodiment, the computation query system begins by loading each second-paradigm computation scheduling knowledge graph into a specific part of the entity recognition network, called the prior recognition network. This prior recognition network may be a subnetwork of the entity recognition network, which has been pre-trained and is specifically designed to generate labeled entity delimitations.

[0098] Once the second-paradigm computation scheduling knowledge graph is loaded into the prior recognition network, the computation query system initiates the domain generation process. For each second node in the second-paradigm computation scheduling knowledge graph, the prior recognition network uses its internal algorithms and learned knowledge to generate a third estimated domain. Finally, the computation query system directly uses the third estimated domain generated for each second node as the second labeled entity domain for that node. This means that in this scenario, the computation query system does not use externally provided labeled data, but relies on the self-learning and estimation capabilities of the prior recognition network to generate labeled entity domains for each node. This approach is particularly useful when sufficient labeled data is lacking, as it allows the computation query system to automatically label new data using existing knowledge and algorithms.

[0099] In this way, the computation query system can autonomously generate labeled entity delimiters for each node in the second paradigm computation scheduling knowledge graph. These delimiters will be used in subsequent network parameter learning and knowledge reasoning processes.

[0100] In one possible implementation, the generation step of the prior recognition network includes:

[0101] Step B110: Obtain a third paradigm operation scheduling knowledge graph sequence, wherein the third paradigm operation scheduling knowledge graph sequence includes at least two third paradigm operation scheduling knowledge graphs, the third paradigm operation scheduling knowledge graph includes a third prior entity, the third paradigm operation scheduling knowledge graph has a third labeled entity delimitation domain associated with each third node in the third paradigm operation scheduling knowledge graph, and the count of the third paradigm operation scheduling knowledge graph is greater than the count of the first paradigm operation scheduling knowledge graph.

[0102] In this embodiment, the computation query system obtains a sequence of third-paradigm computation scheduling knowledge graphs from an internal database or an external data source. This sequence of third-paradigm computation scheduling knowledge graphs contains at least two third-paradigm computation scheduling knowledge graphs, each of which contains a third prior entity. Unlike previous knowledge graphs, these third-paradigm computation scheduling knowledge graphs have a higher order of magnitude, meaning their count is greater than that of the first-paradigm computation scheduling knowledge graph. Furthermore, each third-paradigm computation scheduling knowledge graph contains a third-annotated entity delimiter associated with each third node; these third-annotated entity delimiters are pre-annotated for use in the subsequent learning process.

[0103] Step B120: Load each of the third paradigm operation scheduling knowledge graphs into the prior recognition network to generate a fourth estimated bounding domain associated with each of the third nodes of the third paradigm operation scheduling knowledge graph.

[0104] The computation query system loads each acquired third-paradigm computation scheduling knowledge graph into the prior recognition network, which is under construction or has been partially constructed. This loading process ensures that every node and edge in the third-paradigm computation scheduling knowledge graph is correctly processed by the network. At this stage, the prior recognition network may be in an initial state or already possess a certain level of learning ability; it will further learn and optimize based on the input third-paradigm computation scheduling knowledge graph.

[0105] Once the third-paradigm computational scheduling knowledge graph is loaded into the prior recognition network, the computational query system initiates a process to generate a fourth estimated bounding domain associated with each third node. This process is implemented through the network's internal algorithm and the current learning state, attempting to generate an accurate bounding domain estimate for each node. These fourth estimated bounding domains may differ from the actual bounding domains of the third-labeled entities, but they represent the best estimates in the current state of the network.

[0106] Step B130: Based on the fourth estimated bounding domain and the third labeled entity bounding domain associated with each of the third nodes of the third paradigm operation scheduling knowledge graph, output the third knowledge reasoning error.

[0107] The computational query system compares the fourth estimated bounding domain of each third node with its corresponding third labeled entity bounding domain, calculating the difference between them, i.e., the third knowledge inference error. This third knowledge inference error reflects the accuracy of the prior recognition network in processing the third-paradigm computational scheduling knowledge graph, and it is an important reference for network parameter learning.

[0108] Step B140: Based on the third knowledge inference error, perform network parameter learning on the prior recognition network.

[0109] Based on the calculated third-knowledge inference error, the computational query system adjusts and optimizes the parameters of the prior recognition network. This process may include updating network weights, adjusting network structure, or optimizing the network's learning algorithm. The goal of parameter learning is to reduce the network's inference error when processing similar graphs and improve its accuracy in estimating entity delimitations. Through this process, the prior recognition network gradually learns effective methods for extracting and delimiting entities from the computational scheduling knowledge graph.

[0110] In one possible implementation, step A110 may include:

[0111] Step A111: Obtain the prior computation scheduling knowledge graph, which includes the first prior entity and has the first labeled entity delimitation domain associated with each first node.

[0112] In this embodiment, the computation query system first retrieves a priori computation scheduling knowledge graph from its internal storage or an external data source. This priori computation scheduling knowledge graph contains rich knowledge in the computation scheduling domain and has pre-annotated first prior entities and their associated first annotated entity delimitations. These first prior entities occupy key positions in the graph, and their delimitations clearly define the scope and meaning of the entities in the computation scheduling context.

[0113] Step A112: Apply the same knowledge reconstruction strategy to the first prior entity and the first labeled entity bounding domain in the prior operation scheduling knowledge graph to generate the reconstructed operation scheduling knowledge graph and the associated first labeled entity bounding domain. The knowledge association parameters of the first prior entity between the prior operation scheduling knowledge graph and the reconstructed operation scheduling knowledge graph meet the first preset requirements.

[0114] The computation query system then executes a knowledge reconstruction strategy on the acquired prior computation scheduling knowledge graph. This strategy is specifically designed for the delimitation domains of the first prior entity and the first labeled entity, aiming to reorganize and express the knowledge in the prior computation scheduling knowledge graph without changing the essential meaning of the entities. By applying this strategy, the computation query system generates a new version of the prior computation scheduling knowledge graph, called the reconstructed computation scheduling knowledge graph, while retaining the first labeled entity delimitation domains associated with the entities in the original prior computation scheduling knowledge graph. In this process, the knowledge association parameters of the first prior entity between the prior computation scheduling knowledge graph and the reconstructed computation scheduling knowledge graph are ensured to meet the first preset requirements, which guarantees the consistency and accuracy of knowledge during the reconstruction process.

[0115] Step A113: Apply the same knowledge reconstruction strategy to the first prior entity and the first labeled entity in the reconstructed operation scheduling knowledge graph to generate an iterative reconstructed operation scheduling knowledge graph. The knowledge association parameters of the first prior entity in the associated reconstructed operation scheduling knowledge graphs meet the first preset requirements.

[0116] The computation query system continues to apply the same knowledge reconstruction strategy to the reconstructed computation scheduling knowledge graph. This time, the strategy is applied to the delimited domains of the first prior entity and the first labeled entity that have already been reconstructed in the computation scheduling knowledge graph. This process iteratively generates a new reconstructed computation scheduling knowledge graph, ensuring in each iteration that the knowledge association parameters of the first prior entity between related reconstructed computation scheduling knowledge graphs meet the first preset requirements. This iterative reconstruction method helps to gradually optimize and refine the knowledge representation in the graph.

[0117] Step A114: Repeat step A113 until the count of the reconstructed computation scheduling knowledge graph meets the second preset requirement, and determine the first paradigm computation scheduling knowledge graph sequence based on the prior computation scheduling knowledge graph and at least two of the reconstructed computation scheduling knowledge graphs.

[0118] The computation query system iteratively applies knowledge reconstruction strategies to the first prior entity and the first labeled entity bounding domain in the reconstructed computation scheduling knowledge graph. This iterative process continues until the number of generated reconstructed computation scheduling knowledge graphs meets a second preset requirement. This second preset requirement may be a specific threshold for the number of graphs, or it may be a reconstruction quality standard determined based on some evaluation metric. Once this requirement is met, the computation query system stops iterating and, based on the initial prior computation scheduling knowledge graph and at least two reconstructed computation scheduling knowledge graphs generated through reconstruction, determines a first-paradigm computation scheduling knowledge graph sequence. This first-paradigm computation scheduling knowledge graph sequence constitutes the foundational dataset for subsequent learning and inference.

[0119] In one possible implementation, the step of determining whether the knowledge association parameters of the first prior entity meet the first preset requirements among at least two first paradigm operation scheduling knowledge graphs includes:

[0120] Step C110: For each prior node of the first prior entity in the prior operation scheduling knowledge graph, obtain the link where the prior node is located in at least two of the first paradigm operation scheduling knowledge graphs.

[0121] In this embodiment, the computation query system first focuses on the first prior entity in the prior computation scheduling knowledge graph. For each prior node in this entity, the computation query system checks its position in at least two first-paradigm computation scheduling knowledge graphs. Specifically, the computation query system determines the stage in which these prior nodes are located in each first-paradigm computation scheduling knowledge graph. These stages may be different phases, tasks, or subtasks in the computation scheduling process, which together constitute the complete computation scheduling flow.

[0122] Step C120: If, for each prior node, the deviation between the prior node and the link where the prior node is located in any two first example operation scheduling knowledge graphs is less than a first set deviation, then output the knowledge association parameters of the first prior entity between at least two first example operation scheduling knowledge graphs that meet the first preset requirements.

[0123] The computation query system then compares the differences in the stages where each prior node is located in any two first-paradigm computation scheduling knowledge graphs. This difference is called the deviation, which measures the degree to which the position or role of the same prior node changes in different graphs. The computation query system calculates the deviation for each prior node and compares it with a preset first deviation.

[0124] For example, if the computation query system finds that for each prior node, its deviation between any two first-paradigm computation scheduling knowledge graphs is less than a first preset deviation, then this means that these first-paradigm computation scheduling knowledge graphs maintain a high degree of consistency in describing the first prior entity. In other words, these first-paradigm computation scheduling knowledge graphs meet the first preset requirement in terms of knowledge association parameters.

[0125] Finally, the computation query system outputs a judgment result indicating whether the knowledge association parameters of the first prior entity between at least two first-paradigm computation scheduling knowledge graphs meet the first preset requirements. This judgment result will serve as the basis for subsequent processing or decision-making, ensuring the accuracy and reliability of the computation scheduling knowledge graph.

[0126] In one possible implementation, the step of determining whether the knowledge association parameters of the first prior entity meet the first preset requirements among at least two first paradigm operation scheduling knowledge graphs includes:

[0127] Step D110: Decompose each of the first example operation scheduling knowledge graphs into operation scheduling knowledge units, and assign unit labels to the operation scheduling knowledge units according to the participating links of the operation scheduling knowledge units.

[0128] In this embodiment, the computation query system begins processing each first-paradigm computation scheduling knowledge graph. Firstly, these complex first-paradigm computation scheduling knowledge graphs are decomposed into smaller, more manageable parts, called computation scheduling knowledge units. These computation scheduling knowledge units may represent specific steps, tasks, or subtasks in the computation scheduling process, and each computation scheduling knowledge unit contains specific knowledge directly related to computation scheduling.

[0129] After decomposing the computation scheduling knowledge graph of the first paradigm, the computation query system assigns a unit label to each computation scheduling knowledge unit. This unit label is determined based on the unit's participation in the computation scheduling process. For example, if a knowledge unit describes the resource allocation process, it might be assigned a resource allocation-related label. These unit labels help the computation query system quickly identify and classify different computation scheduling knowledge units.

[0130] Step D120: If the first prior entity is in the same computation scheduling knowledge unit associated with the same unit label in each of the first paradigm computation scheduling knowledge graphs, output that the knowledge association parameters of the first prior entity between at least two first paradigm computation scheduling knowledge graphs meet the first preset requirements.

[0131] Next, the computation query system examines the position of the first prior entity in each first-paradigm computation scheduling knowledge graph. The system pays particular attention to whether this entity appears in computation scheduling knowledge units with the same unit label. This is because if the first prior entity appears in the same type of knowledge unit in different graphs, it may indicate that these graphs maintain consistency in describing this entity.

[0132] The computation query system compares the position of the first prior entity in each first-paradigm graph. If the computation query system finds that the first prior entity is located in a computation scheduling knowledge unit with the same unit label in each first-paradigm graph, it means that these graphs are highly consistent in describing this entity.

[0133] Finally, if the above conditions are met, the computation query system will output a judgment result indicating that the knowledge association parameters of the first prior entity between at least two first-paradigm computation scheduling knowledge graphs meet the first preset requirements. This judgment result is based on the analysis of the entity's consistency across different graphs, proving the accuracy and reliability of these graphs in describing the first prior entity. If the conditions are not met, the computation query system may output a judgment result that does not meet the requirements, or take other measures, such as re-analyzing the graphs or updating the knowledge association parameters.

[0134] In one possible implementation, the target computation scheduling knowledge graph includes multiple uninterrupted target computation scheduling knowledge graphs, the target entity delimitation domain includes a target entity delimitation subdomain of each of the target computation scheduling knowledge graphs, and step S110 may include:

[0135] Step S111: For the target operation scheduling knowledge graph traversed in this round, obtain the forward-generated target operation scheduling knowledge graph and the target entity delimitation subdomain of the forward-generated target operation scheduling knowledge graph.

[0136] Step S112: Load the target operation scheduling knowledge graph traversed in this round, the target operation scheduling knowledge graph generated in the forward direction, and the target entity delimitation subdomain of the target operation scheduling knowledge graph generated in the forward direction into the entity recognition network to generate the target entity delimitation subdomain associated with each target node of the target operation scheduling knowledge graph traversed in this round.

[0137] In this embodiment, the computation query system begins processing the target computation scheduling knowledge graph for this round of traversal. Before processing, it first obtains the target computation scheduling knowledge graph generated in the previous round of traversal, as well as the associated target entity delimitation subdomains. These forward-generated knowledge graphs and entity delimitation subdomains are the results obtained by the computation query system in the previous round of processing, and they will serve as the input data for this round of processing.

[0138] The computation query system will then load the target computation scheduling knowledge graph traversed in this round, the target computation scheduling knowledge graph generated in the previous round, and the target entity delimitation subdomain generated in the previous round into the entity recognition network.

[0139] After loading the relevant data into the entity recognition network, the computational query system runs the network's recognition algorithm to generate target entity delimitation subdomains associated with each target node in the target computational scheduling knowledge graph for this round of traversal. These target entity delimitation subdomains are automatically generated by the entity recognition network based on the structural and relational information in the graph, and they accurately describe the role and meaning of each target node in the graph.

[0140] Through the above steps, the computational query system can utilize the entity recognition network to perform entity recognition and domain generation on the target computational scheduling knowledge graph, providing accurate foundational data for subsequent knowledge graph analysis and applications. Simultaneously, by continuously iterating and processing the knowledge graph, the computational query system can gradually optimize and improve the performance of the entity recognition network, enhancing the accuracy and efficiency of entity recognition.

[0141] Figure 2 shows a computational query system 100 provided in this embodiment of the application, including a processor 1001, a memory 1003, and program code stored in the memory 1003. The processor 1001 executes the program code to implement the steps of the knowledge graph-based information processing method.

[0142] The computational query system 100 shown in Figure 2 includes a processor 1001 and a memory 1003. The processor 1001 and the memory 1003 are connected, for example, via a bus 1002. Optionally, the computational query system 100 may further include a transceiver 1004, which can be used for data interaction between the computational query system and other computational query systems, such as sending and / or receiving data. It should be noted that in actual scheduling, the transceiver 1004 is not limited to one, and the structure of this computational query system 100 does not constitute a limitation on the embodiments of this application.

[0143] Processor 1001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 1001 may also be a combination that implements computing functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0144] Bus 1002 may include a pathway for transmitting information between the aforementioned components. Bus 1002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 1002 may be divided into address bus, data bus, control bus, etc.

[0145] The memory 1003 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media, other magnetic storage devices, or any other medium capable of having or storing program code and capable of being read by a computer, without limitation herein.

[0146] The memory 1003 is used to store program code for executing the embodiments of this application, and its execution is controlled by the processor 1001. The processor 1001 is used to execute the program code stored in the memory 1003 to implement the steps shown in the foregoing method embodiments.

[0147] This application provides a computer-readable storage medium storing program code, which, when executed by a processor, can implement the steps and corresponding content of the aforementioned method embodiments.

[0148] It should be understood that although arrows indicate various operation steps in the flowcharts of this application's embodiments, the order in which these steps are implemented is not limited to the order indicated by the arrows. Unless explicitly stated herein, in some implementation scenarios of this application's embodiments, the implementation steps in each flowchart can be executed in other orders based on requirements. Furthermore, some or all steps in each flowchart may include multiple sub-steps or multiple stages depending on the actual implementation scenario. Some or all of these sub-steps or stages can be executed simultaneously, and each sub-step or stage can also be executed at different times. In scenarios where execution times differ, the execution order of these sub-steps or stages can be flexibly configured based on requirements, and this application's embodiments do not limit this.

[0149] The above description is only an optional implementation method for some implementation scenarios of this application. It should be noted that for those skilled in the art, other similar implementation methods based on the technical concept of this application, without departing from the technical concept of this application, also fall within the protection scope of the embodiments of this application.

Claims

1. A knowledge graph-based information processing method, characterized in that, The method includes: Obtain the target computation scheduling knowledge graph and load the target computation scheduling knowledge graph into the entity recognition network to generate the target entity delimitation domain associated with each target node of the target computation scheduling knowledge graph; Based on the target entity definition domain, the target entity is identified from the target computation scheduling knowledge graph. The entity recognition network learns network parameters based on a first knowledge inference error. The first knowledge inference error is calculated based on the first estimated definition domain of each first node in at least two first paradigm computation scheduling knowledge graphs in the first paradigm sequence, which is identified by the entity recognition network, and the first labeled entity definition domain of each first node. The at least two first paradigm computation scheduling knowledge graphs include a prior computation scheduling knowledge graph and at least two reconstructed computation scheduling knowledge graphs generated by applying at least two differentiated knowledge reconstruction strategies to the first prior entity in the prior computation scheduling knowledge graph. The knowledge association parameters of the first prior entity between at least two first paradigm computation scheduling knowledge graphs meet a first preset requirement. The knowledge association parameters are parameters that describe the association or similarity between entities in the first paradigm computation scheduling knowledge graphs. The determination step of whether the knowledge association parameters of the first prior entity meet the first preset requirements among at least two first paradigm operation scheduling knowledge graphs includes: For each prior node of the first prior entity in the prior operation scheduling knowledge graph, obtain the link where the prior node is located in at least two first example operation scheduling knowledge graphs; if for each prior node, the deviation between the links where the prior node is located in any two first example operation scheduling knowledge graphs is less than a first set deviation, output that the knowledge association parameters of the first prior entity between at least two first example operation scheduling knowledge graphs meet the first preset requirements, where the first prior entity refers to an entity that is specially focused on or used as the basis for knowledge reconstruction in the prior operation scheduling knowledge graph, and the link is a different stage, task or subtask in the operation scheduling process; Alternatively, each of the first paradigm operation scheduling knowledge graphs can be decomposed into operation scheduling knowledge units, and unit labels can be assigned to the operation scheduling knowledge units according to the participating links of the operation scheduling knowledge units; if the first prior entity is in the operation scheduling knowledge unit associated with the same unit label in each of the first paradigm operation scheduling knowledge graphs, the knowledge association parameters of the first prior entity between at least two first paradigm operation scheduling knowledge graphs can be output to meet the first preset requirements.

2. The information processing method based on knowledge graphs according to claim 1, characterized in that, Based on the first knowledge inference error, network parameter learning is performed on the entity recognition network, including: Obtain a second paradigm operation scheduling knowledge graph sequence, the second paradigm operation scheduling knowledge graph sequence includes at least two second paradigm operation scheduling knowledge graphs parsed from the same operation scheduling knowledge graph, the second paradigm operation scheduling knowledge graph includes a second prior entity, the second paradigm operation scheduling knowledge graph has a second labeled entity delimitation domain associated with each second node in the second paradigm operation scheduling knowledge graph, and the second prior entity occupies a key position in the second paradigm operation scheduling knowledge graph; Each second paradigm operation scheduling knowledge graph is loaded into the entity recognition network to generate a second estimated bounding domain associated with each second node of the second paradigm operation scheduling knowledge graph; Based on the second example, the second estimated bounding domain and the second labeled entity bounding domain associated with each second node of the knowledge graph are operated and scheduled, and the second knowledge reasoning error is output. Based on the first knowledge inference error and the second knowledge inference error, network parameters are learned for the entity recognition network.

3. The information processing method based on knowledge graphs according to claim 2, characterized in that, The steps for generating the second labeled entity delimitation domain include: Each second paradigm operation scheduling knowledge graph is loaded into the prior recognition network of the entity recognition network to generate a third estimated bounding domain associated with each second node of the second paradigm operation scheduling knowledge graph, and the third estimated bounding domain is used as the second labeled entity bounding domain.

4. The information processing method based on knowledge graphs according to claim 3, characterized in that, The generation steps of the prior recognition network include: Obtain a third-paradigm operation scheduling knowledge graph sequence, the third-paradigm operation scheduling knowledge graph sequence including at least two third-paradigm operation scheduling knowledge graphs, the third-paradigm operation scheduling knowledge graphs including third prior entities, the third-paradigm operation scheduling knowledge graphs having third labeled entity delimitations associated with each third node in the third-paradigm operation scheduling knowledge graphs, and the count of the third-paradigm operation scheduling knowledge graphs being greater than the count of the first-paradigm operation scheduling knowledge graphs; Each of the third paradigm operation scheduling knowledge graphs is loaded into the prior recognition network to generate a fourth estimated bounding domain associated with each of the third nodes of the third paradigm operation scheduling knowledge graph. Based on the fourth estimated bounding domain and the third labeled entity bounding domain associated with each of the third nodes of the third paradigm operation scheduling knowledge graph, the third knowledge reasoning error is output. Based on the third knowledge inference error, network parameters are learned for the prior recognition network.

5. The information processing method based on knowledge graphs according to claim 1, characterized in that, Obtain the first-paradigm operation scheduling knowledge graph sequence, including: Obtain the prior computation scheduling knowledge graph, which includes the first prior entity and has the first annotation entity delimitation domain associated with each first node; The knowledge reconstruction strategy under the same stage is applied to the first prior entity and the first labeled entity bounding domain in the prior operation scheduling knowledge graph to generate the reconstructed operation scheduling knowledge graph and the associated first labeled entity bounding domain. The knowledge association parameters of the first prior entity between the prior operation scheduling knowledge graph and the reconstructed operation scheduling knowledge graph meet the first preset requirements. The knowledge reconstruction strategy under the same stage is designed for the first prior entity and the first labeled entity bounding domain, and aims to reorganize and express the knowledge in the prior operation scheduling knowledge graph without changing the essential meaning of the entity. The knowledge reconstruction strategy under the same stage is applied to the first prior entity and the first labeled entity in the reconstructed operation scheduling knowledge graph to generate an iterative reconstructed operation scheduling knowledge graph. The knowledge association parameters of the first prior entity in the associated reconstructed operation scheduling knowledge graphs meet the first preset requirements. The steps of applying the same knowledge reconstruction strategy to the first prior entity and the first labeled entity in the reconstructed operation scheduling knowledge graph and generating iterative reconstructed operation scheduling knowledge graphs are executed repeatedly until the count of the reconstructed operation scheduling knowledge graph meets the second preset requirement. Based on the prior operation scheduling knowledge graph and at least two reconstructed operation scheduling knowledge graphs, a first paradigm operation scheduling knowledge graph sequence is determined.

6. The information processing method based on knowledge graphs according to claim 1, characterized in that, The target computation scheduling knowledge graph includes multiple uninterrupted target computation scheduling knowledge graphs, and the target entity delimitation domain includes a target entity delimitation subdomain for each target computation scheduling knowledge graph. The step of loading the target computation scheduling knowledge graph into the entity recognition network to generate the target entity delimitation domain associated with each target node of the target computation scheduling knowledge graph includes: For the target operation scheduling knowledge graph traversed in this round, obtain the target operation scheduling knowledge graph generated in the forward direction and the target entity delimitation subdomain of the target operation scheduling knowledge graph generated in the forward direction; The target computation scheduling knowledge graph traversed in this round, the target computation scheduling knowledge graph generated in the forward pass, and the target entity delimitation subdomain of the target computation scheduling knowledge graph generated in the forward pass are loaded into the entity recognition network to generate the target entity delimitation subdomain associated with each target node of the target computation scheduling knowledge graph traversed in this round.

7. A computational query system, characterized in that, The method includes a processor and a computer-readable storage medium storing machine-executable instructions that, when executed by the processor, implement the knowledge graph-based information processing method according to any one of claims 1-6.

Images