Government affair index management method

Abstract

The invention discloses a government affair index management method, and the method comprises the steps: obtaining an original government affair index library, carrying out the preprocessing of the original government affair index library, obtaining the index information, enabling the index information to comprise an entity, a relation, and an entity attribute value, enabling the entity to be an index in the original government affair index library, and enabling the relation to be an index in the original government affair index library; the relation refers to the relation between the entities; constructing an index knowledge graph based on the entities, the relationships and the entity attribute values; and fusing the index knowledge graph into the original government affair index library to obtain a fused government affair index library which comprises index names, index definitions, relationships among indexes and index attributes.

Description

Technical Field

[0001] The present application relates to the field of big data, and in particular to a government affairs indicator management method. Background Art

[0002] Government affairs is a result of data analysis and is a standard used to measure government affairs. Indicators can be divided into atomic indicators, derived indicators and composite indicators according to their granularity. Derived indicators are composed of atomic indicators combined with analysis dimensions, and composite indicators are obtained through calculations of other indicators.

[0003] In the process of government affairs management in related technologies, there is no repeated retrieval mechanism for saving indicators. New indicators calculated will be re-entered into the database, but new indicators generated by multiple calculations will be repeated or invalid indicators; there is no indicator business dependency mechanism, and the generation of new indicators only depends on dimension division. The generated new indicators lack business integration, which affects the availability of generated indicators; the query efficiency is low. Most current indicator management systems use information such as indicator names and dimensions to query and obtain relevant results of indicators. However, due to the small amount of information such as indicator names and dimensions, it is impossible to accurately reflect the calculation logic of indicators and the calculation relationship between indicators and other indicators (such as derived indicators, combined indicators, etc.). It is necessary to query this information again, layer by layer, and after multiple queries, the complete calculation logic of the indicator can be obtained. This management method is relatively inefficient and cannot quickly and accurately reflect the complete information of indicators. Multiple queries also consume a lot of computing resources. Summary of the invention

[0004] The embodiment of the present application provides a method for managing government indicators to solve the problems of low management efficiency, repeated query and waste of resources in related technologies.

[0005] In a first aspect, an embodiment of the present application provides a method for managing government affairs indicators, including: Acquire an original government indicator database, and preprocess the original government indicator database to obtain indicator information, wherein the indicator information includes entities, relationships, and entity attribute values, wherein the entities are indicators in the original government indicator database, and the relationships are relationships between entities; Based on the entities, relationships, and entity attribute values, construct an indicator knowledge graph; The indicator knowledge graph is integrated into the original government indicator database to obtain a fused government indicator database, which includes indicator names, indicator definitions, relationships between indicators, and indicator attributes.

[0006] In a second aspect, an embodiment of the present application provides a government affairs indicator management device, including: An acquisition module is used to acquire an original government indicator database and preprocess the original government indicator database to obtain indicator information, wherein the indicator information includes entities, relationships, and entity attribute values. The entities are indicators in the original government indicator database, and the relationships are relationships between entities. A construction module, used to construct an indicator knowledge graph based on the entities, relationships, and entity attribute values; A fusion module is used to fuse the indicator knowledge graph into the original government indicator library to obtain a fused government indicator library, which includes indicator names, indicator definitions, relationships between indicators, and indicator attributes.

[0007] In a third aspect, an embodiment of the present application provides a terminal device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method described in the first aspect.

[0008] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0009] In a fifth aspect, an embodiment of the present application provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions, which, when executed by a computer, implement the steps of the method described in the first aspect.

[0010] In an embodiment of the present application, the original government indicator library is first obtained, and the original government indicator library is preprocessed to obtain indicator information, which includes entities, relationships, and entity attributes. Entities are indicators in the original government indicator library, and relationships refer to the associations between entities. Then, based on entities, relationships, and entity attribute values, an indicator knowledge graph is constructed, and finally the indicator knowledge graph is integrated into the original government indicator library to obtain a fused government indicator library. The fused government indicator library includes indicator names, indicator descriptions, relationships between indicators, and indicator attributes. The embodiment of the present application can extract valuable information from a complex government indicator library and present it in an intuitive and easy-to-understand form in the form of a graph, thereby improving the accuracy of indicator queries, improving the efficiency of data indicator management, and avoiding repeated queries that waste resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1It is a flow chart of the government affairs indicator management method provided in the embodiment of the present application; Figure 2 A schematic diagram of knowledge graph embedding provided in an embodiment of the present application; Figure 3 is a schematic diagram of a government affairs indicator management device provided in an embodiment of the present application; Figure 4 It is a schematic diagram of the structure of the terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0012] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0013] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here. In addition, the "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally represents that the objects associated with each other are in an "or" relationship.

[0014] Government affairs is a result of data analysis and a standard for measuring government affairs business. Indicators can be divided into atomic indicators, derived indicators and composite indicators according to their granularity. Derived indicators are composed of atomic indicators combined with analysis dimensions, and composite indicators are obtained by calculation of other indicators. For example, annual income is an atomic indicator, average monthly income is a derived indicator, and annual profit is a composite indicator, which is the result of calculating the annual income (atomic indicator) minus the cost of the previous year (atomic indicator). Usually, a data indicator includes the indicator name, indicator definition, analysis dimension of the indicator and calculation logic of the indicator. The calculation logic description includes specific data fields (atomic indicators / derived indicators) and the calculation relationship between data fields.

[0015] As data is increasingly used in enterprises, data indicators are often needed to guide decision-making for a specific business. Data indicators are statistics and summary results for certain business data. As business complexity increases, the query and calculation tasks of business indicators will gradually increase. How to accurately locate data indicators and understand the true meaning and calculation methods of indicators becomes a challenge.

[0016] The following is combined with Figures 1 to 4, a government affairs indicator management method provided by an embodiment of the present application is described in detail through specific embodiments and application scenarios.

[0017] like Figure 1 As shown, it is a flow chart of a government affairs indicator management method provided by an embodiment of the present application. Figure 1 As shown, the government affairs indicator management method may include: contents shown from S101 to S103.

[0018] In S101, the original government indicator database is obtained and preprocessed to obtain indicator information, which includes entities, relationships, and entity attribute values. The entity is an indicator in the original government indicator database, and the relationship refers to the relationship between entities.

[0019] Among them, the triple of entity, relationship, and entity attribute value is an indicator, which is obtained by extracting indicator system information based on the existing government indicator database.

[0020] In S102, an indicator knowledge graph is constructed based on entities, relationships, and entity attribute values.

[0021] In this embodiment, by constructing an indicator knowledge graph, the data in the government indicator library can be presented intuitively.

[0022] In S103, the indicator knowledge graph is integrated into the original government indicator database to obtain a fused government indicator database, which includes indicator names, indicator definitions, relationships between indicators, and indicator attributes.

[0023] In this embodiment, by integrating the constructed indicator knowledge graph with the original government indicator database, the indicators in the original government indicator database are easier to manage and convenient for subsequent queries.

[0024] In an embodiment of the present application, the original government indicator library is first obtained, and the original government indicator library is preprocessed to obtain indicator information, which includes entities, relationships, and entity attributes. Entities are indicators in the original government indicator library, and relationships refer to the associations between entities. Then, based on entities, relationships, and entity attribute values, an indicator knowledge graph is constructed, and finally the indicator knowledge graph is integrated into the original government indicator library to obtain a fused government indicator library. The fused government indicator library includes indicator names, indicator descriptions, relationships between indicators, and indicator attributes. The embodiment of the present application can extract valuable information from a complex government indicator library and present it in an intuitive and easy-to-understand form in the form of a graph, thereby improving the accuracy of indicator queries, improving the efficiency of data indicator management, and avoiding repeated queries that waste resources.

[0025] In a possible implementation of the present application, preprocessing the original government indicator library to obtain indicator information may include: preprocessing the original government indicator library, extracting indicator names from the original government indicator library, identifying entities, and the entities include entity definitions and application rules; determining the relationship between entities based on the dependency or association relationship between entities; extracting entity attribute values ​​of the entities based on the entity definitions and application rules, and the entity attribute values ​​include at least one of the following: calculation logic, unit, spatial dimension, time dimension, and update frequency.

[0026] In this embodiment, triples for constructing the indicator knowledge graph are obtained through entity extraction, relationship extraction, and attribute extraction. Through the above process, valuable information can be effectively extracted from the complex government indicator library and presented in an intuitive and easy-to-understand form, thereby supporting more in-depth data analysis and decision making.

[0027] The process of entity extraction is as follows: At this stage, the original indicator system needs to be preprocessed. This usually includes cleaning and formatting the data so that meaningful information can be extracted from it. Then, by analyzing the indicator name column, specific entities (i.e., the indicator itself) can be identified. Each entity represents a unique indicator, which will become the basic unit for the subsequent construction of the knowledge graph.

[0028] The process of relationship extraction is as follows: Relationship refers to the relationship between entities, which is one of the key steps in building a knowledge graph. In this embodiment, rules can be set to determine which entities are dependent or associated. For example, some indicators may be calculated based on other indicators, or they may have a common trend of change in a specific time period. By applying these rules, the relationships between entities can be identified and recorded.

[0029] The process of attribute extraction is as follows: This step involves defining and applying rules to extract specific attribute values ​​for each entity. These attributes may include, but are not limited to: the calculation logic of the indicator, units, spatial dimensions (such as regions), time dimensions (such as years), update frequency, and departments responsible for management or release. Understanding these details helps to more fully describe each entity and provide support for building complex relationships.

[0030] After determining the entity, relationship, and entity attribute value, you can start building a triple, which consists of a head entity (h), a relationship (r), and a tail entity (t). In a possible implementation of the present application, based on the entity, relationship, and entity attribute value, an indicator knowledge graph is constructed, including: Construct the indicator knowledge graph according to the construction rules shown in the following formula: CKG={(h,r,t)∣h,t∈E, r∈R} Among them, h is the head entity, r is the tail entity; E is the set of all entities; r is the relationship; R is the set of all relationships.

[0031] In order to highlight important indicators that are frequently cited or used, a knowledge graph of popular indicators can be constructed. First, determine which indicators are "popular", usually by sorting them according to the number of citations and selecting the top preset percentage, such as the top 30% of indicators. Then, establish new entity-relationship-attribute triples for these popular indicators and their associated projects. In a possible implementation of the present application, the government indicator management method also includes: Construct a knowledge graph of popular indicators as shown below: KGitem={(h1,r1,t1)∣h1∈Ehot, t1∈Et, r1∈Rt} Among them, h1 is the entity corresponding to the hot indicator, and the hot indicator refers to the indicator that ranks in the top preset percentage among all the referenced indicators in the original government indicator library; t1 is the entity corresponding to the project that references the hot indicator; r1 is the relationship between h1 and t1; Ehot is the set of hot indicators; Et is the set of all projects, and Rt is the set of relationships between Ehot and Et.

[0032] In this embodiment, two knowledge graphs of different dimensions are obtained, among which the indicator knowledge graph integrates the indicators and their attributes into one knowledge graph, which is more conducive to mining the implicit information between the indicators; the popular indicator knowledge graph contains indicators with more interactions, which can enhance the project characteristics when making recommendations.

[0033] In order to use the neural network model to mine features in the knowledge graph, such as the characteristic information of users and projects, it is necessary to map entities and relationships into a low-dimensional vector space, namely, knowledge graph embedding. The knowledge graph embedding algorithm is based on multi-scale dilated convolution and attention mechanism, as shown below: In a possible implementation of the present application, the government indicator management method also includes: randomly initializing the entities and relationships in the indicator knowledge graph or the popular indicator knowledge graph into a multidimensional vector; reshaping the multidimensional vector to obtain a two-dimensional vector; setting two different void rates, splitting the two-dimensional vectors of the entities and relationships into convolution kernels of different scales, and performing void rate convolution respectively to obtain the feature matrix of the entity and the feature matrix of the relationship; adding the original information of the entity and the relationship to the feature matrix of the entity and the feature matrix of the relationship; using the attention mechanism to aggregate the feature information of the feature matrices of entities and the feature matrices of relationships of different scales to obtain a weighted matrix; stretching the weighted matrix and mapping it to the dimension of the tail entity embedding after full connection to obtain the tail entity matrix; performing an inner product operation on the feature matrix and the tail entity matrix to calculate the similarity, obtaining the scores of the entities, relationships, and entity attribute values, and obtaining the embedded representation of the entities and relationships of the indicator knowledge graph and the popular indicator knowledge graph through back propagation.

[0034] In this embodiment, the entities and relationships in the triples are randomly initialized to a d-dimensional vector, and then the entity and relationship vectors are reshaped into a two-dimensional vector D; two different void rates are set to split the embedding of the entity and relationship into convolution kernels of different scales. With D as input, dilated convolutions are performed respectively to obtain feature matrices after two layers of six dilated convolutions; the residual network is used to add the original information of the entities and relationships; the attention mechanism SKNet is used to aggregate feature information of different scales. This process includes two parts: feature fusion and feature selection. In the feature fusion part, the two groups of six feature matrices obtained by two layers of multi-scale dilated convolutions are pooled and fully connected to obtain the feature vector v; the feature selection part calculates the weight of the feature information at each scale, and finally performs weighted aggregation; the weighted feature matrix is ​​stretched, and after passing through the fully connected layer, the result is mapped to the dimension of the tail entity embedding, and finally the feature matrix and the tail entity matrix are inner-producted to calculate the similarity, and the score of the given triple is obtained, and the embedding representation of the entity and relationship is updated by back-propagation. As Figure 2 shown.

[0035] After obtaining the embedded representations of the two knowledge graphs, in order to solve the problems of easy loss of original information, failure to distinguish the importance of neighborhood weights and neighborhood information at different distances in the feature information propagation part of the current knowledge graph-based recommendation algorithm, the graph attention network technology is used to propagate entity features, as shown below: In a possible implementation of the present application, the government indicator management method also includes: using a multi-head graph attention network to aggregate the neighborhood information of the entity and assigning different weights to different neighborhoods; using an attenuation mechanism to aggregate the feature information of different neighborhoods of the entity and obtain the weight of the feature information.

[0036] That is to say, in the process of knowledge graph embedding, the embedded representations of entities and relationships of the government indicators knowledge graph and the popular indicators knowledge graph are obtained. First, the entity features of the two knowledge graphs are fused, and after the feature fusion, the embedded representations of users and projects are obtained respectively; then, the neighborhood information of the entity is aggregated using a multi-headed graph attention network, and different weights are assigned to different neighborhoods; the attenuation mechanism is used to aggregate the feature information of different neighborhoods of the entity, and for the neighborhood with a distance of l, its weight is aggregation method addition, splicing or enhanced aggregation.

[0037] By assigning different weights to different neighborhoods and aggregating feature information of different neighborhoods of an entity through this embodiment, it is possible to solve the problems existing in the feature information propagation part of the current knowledge graph-based recommendation algorithm, such as easy loss of original information, failure to distinguish neighborhood weights and importance of neighborhood information at different distances.

[0038] In a possible implementation of the present application, the indicator knowledge graph is integrated into the original government indicator library to obtain a fused government indicator library, including: updating the indicator knowledge graph based on actual data, and linking the updated indicator knowledge graph entities to the original government indicator library; using a correlation algorithm to determine the strength of association between different entities, and performing knowledge fusion on different entities.

[0039] In this embodiment, entity ambiguity is eliminated by entity linking entities, relationships between entities, and entity attribute values.

[0040] In one implementation, the indicator knowledge graph is updated based on actual data, including: extracting indicator attributes of each indicator in the actual data from the nodes of the indicator knowledge graph; obtaining other indicators associated with the current indicator based on the relationship of the indicator knowledge graph; and updating the indicator knowledge graph based on the indicator attributes of each indicator and other indicators associated with the current indicator.

[0041] The process of knowledge generation and fusion aims to enhance the accuracy and practicality of the knowledge graph and improve its quality by eliminating entity ambiguity and integrating related knowledge. Specifically, it includes the contents shown in S1 to S3 below.

[0042] S1: Indicator calculation At this stage, the focus is on using the constructed knowledge graph to perform actual data processing and analysis. Specifically: • Get basic properties: Extract the basic properties of each indicator (such as unit, time dimension, etc.) from the graph node to ensure that the correct parameters are used during calculation.

[0043] • Read calculation dependencies: Use the edges in the graph (i.e., the relationships between entities) to identify which other indicators the current indicator calculation depends on, and establish the calculation process accordingly.

[0044] • Perform calculations: Based on the above information, the indicators are calculated according to the predetermined logic or formula to obtain the latest results.

[0045] •Update graph nodes: Save the new calculated results back to the corresponding graph nodes to keep the knowledge graph up to date.

[0046] S2: Entity Linking The key to this step is to reduce or eliminate the phenomenon of homonymous entities or heteronymous entities in data from different sources, that is, entity ambiguity. The method is as follows: • Link to existing libraries: For each newly identified entity, try to match it with an entry in an existing government indicator library. This may involve techniques such as text similarity calculation and semantic analysis.

[0047] • Disambiguation: When there are multiple possible matches, certain rules or algorithms (e.g., context analysis, domain expertise, etc.) are used to determine the most appropriate match, thereby ensuring that each entity has a unique and accurate representation in the knowledge graph.

[0048] S3: Knowledge Fusion The final step is to synthesize the processed entities and their related information to achieve deeper knowledge discovery. The following measures can be taken: • Apply correlation algorithms: Use machine learning or other statistical methods to evaluate the strength of associations between different entities and identify underlying patterns or trends.

[0049] •Merge redundant information: If it is found that some entities actually refer to the same concept or thing, you should consider merging them into a unified entity while retaining their unique attributes or characteristics.

[0050] • Expand the knowledge graph: Based on newly discovered relevance, add new connections to the knowledge graph or adjust the existing structure so that it can better reflect the complex relationships in the real world.

[0051] Through the above steps, not only can the quality and reliability of the knowledge graph be improved, but also cross-domain and cross-departmental collaboration and communication can be promoted, providing strong support for government decision-making. In addition, the continuous process of knowledge generation and integration helps the knowledge graph to evolve continuously and adapt to changing needs and development.

[0052] In a possible implementation of the present application, the government indicator management method also includes: when update information is received, determining the confidence of the update information, the update information being at least one of an entity, a relationship, and an entity attribute value; when the confidence of the update information is greater than or equal to a confidence threshold, inputting the update information into the indicator knowledge graph, and using the update information to update the indicator knowledge graph; when the confidence of the update information is less than the confidence threshold, deleting the update information.

[0053] In other words, the Chauvel criterion can be used to verify and update the predictive knowledge to verify the accuracy of the triples, making the information in the knowledge graph more accurate.

[0054] In an example, update information is first input, and the update information includes one of entities, entity relationships or entity attributes; then the confidence judgment is made on the update information, and all indicators are used as samples to construct a data set using the Chauvelet criterion, and a probability band centered on the mean of the normal distribution is determined. Any sample data value that is not within the probability band will be judged as an outlier and removed from the data set. The calculation formula of the Chauvelet criterion is: Dmax≥|x-μ| / δ, where Dmax is the set maximum deviation value, x is the suspected outlier, μ is the sample mean, and δ is the sample standard deviation. All triples are input into the Trans-E training model, and the noise-aware knowledge graph model is trained using random negative sampling. A triplet scoring formula is preset in the Trans-E model. According to the scores of each triplet after training, all triples are input into the triplet classification model for training. After the training, the confidence of each triplet is refreshed, and different triples have different confidences. A preset confidence threshold, that is, when the confidence of a triplet is greater than a preset threshold, the triplet is judged to be correct and retained; when the confidence of a triplet is less than a preset threshold, the triplet is judged to be wrong and removed. The Sigmoid function is used in the triple classification model to constrain the output of the classifier to 0-1. The Trans-E model and the triple classification model are combined for iterative training until the knowledge graph training model and the triple classification model are fully converged. Finally, if the confidence is within the set threshold range, the update information is input into the knowledge graph for incremental update, otherwise, the update information is filtered out.

[0055] In a possible implementation of the present application, knowledge fusion includes at least one of the following: name fusion, connotation fusion and dimension fusion.

[0056] Among them, name fusion is to conduct consistency analysis between the indicator name of the newly added indicator and the existing indicators; connotation fusion is to conduct consistency analysis between the indicator definition of the newly added single indicator and the existing indicators; dimension fusion is to conduct consistency analysis between the indicator time dimension, space dimension, update dimension of the newly added single indicator and the existing indicators.

[0057] Specifically, when the indicator names of the two are the same but the indicator descriptions are different, the indicator name of the single indicator to be added will be reminded and modified, and the single indicator will be added to the government indicator database; when the indicator names and indicator connotations of the two are different, the single indicator to be added will be added to the integrated indicator database, and the basic information of the new single indicator will be improved; when the indicator names of the two are different but the indicator connotations are the same, an alias will be added to the single indicator to be added, and the next step of integrated analysis will be carried out; when the indicator names and indicator definitions of the two are the same, the next step of updating dimension analysis will be carried out; Update dimension fusion is to conduct consistency analysis between the update dimension of the newly added single indicator and the existing indicator; When the update dimensions of the two indicators are different, the update dimensions of the existing indicators are adjusted; when the update dimensions of the two indicators are the same, the next step is time dimension analysis. Time dimension fusion is to conduct consistency analysis between the time dimension of the newly added single indicator and the existing indicators; When the time dimensions of the two are different, the update dimension of the existing indicator is adjusted; when the time dimensions of the two are the same, the next dimension analysis is carried out; The applicable spatial dimension fusion is to conduct consistency analysis between the applicable area of ​​the newly added single indicator and the existing indicators; When the applicable spatial dimensions of the two are the same, it is determined that the single indicator to be added is already in the integrated government business data indicator library, and the addition is completed.

[0058] In this embodiment, a fusion analysis is performed on the new indicators calculated based on the graph, and multiple scenarios are integrated through name comparison, connotation comparison, and dimension comparison, so as to effectively improve the utilization rate of the new indicators.

[0059] like Figure 3 As shown, it is a schematic diagram of a government affairs indicator management device provided in an embodiment of the present application. Figure 3 As shown, the government affairs indicator management device may include: an acquisition module 301, a construction module 302, and a fusion module 303.

[0060] Among them, the acquisition module 301 is used to obtain the original government indicator library and pre-process the original government indicator library to obtain indicator information, which includes entities, relationships, and entity attribute values. The entity is the indicator in the original government indicator library, and the relationship refers to the relationship between entities; the construction module 302 is used to construct an indicator knowledge graph based on entities, relationships, and entity attribute values; the fusion module 303 is used to fuse the indicator knowledge graph into the original government indicator library to obtain a fused government indicator library, which includes indicator names, indicator definitions, relationships between indicators, and indicator attributes.

[0061] In the embodiment of the present application, first, the acquisition module 301 acquires the original government indicator library, and pre-processes the original government indicator library to obtain indicator information, which includes entities, relationships, and entity attributes. Entities are indicators in the original government indicator library, and relationships refer to the associations between entities. Then, the construction module 302 constructs an indicator knowledge graph based on entities, relationships, and entity attribute values. Finally, the fusion module 303 fuses the indicator knowledge graph into the original government indicator library to obtain a fused government indicator library. The fused government indicator library includes indicator names, indicator descriptions, relationships between indicators, and indicator attributes. The embodiment of the present application can extract valuable information from a complex government indicator library and present it in an intuitive and easy-to-understand form in the form of a graph, thereby improving the accuracy of indicator queries, improving the efficiency of data indicator management, and avoiding repeated queries that waste resources.

[0062] In a possible implementation of the present application, the acquisition module 301 is used to: preprocess the original government indicator library, extract the indicator name from the original government indicator library, identify the entity, and the entity includes the definition and application rules of the entity; determine the relationship between the entities based on the dependency or association relationship between the entities; based on the definition and application rules of the entity, extract the entity attribute value of the entity, and the entity attribute value includes at least one of the following: calculation logic, unit, spatial dimension, time dimension, and update frequency.

[0063] In a possible implementation of the present application, the construction module 302 is used to: construct an indicator knowledge graph according to the construction rule shown in the following formula: CKG={(h,r,t)∣h,t∈E, r∈R} Among them, h is the head entity, r is the tail entity; E is the set of all entities; r is the relationship; R is the set of all relationships.

[0064] In a possible implementation of the present application, the construction module 302 is used to: construct a hot index knowledge graph according to the following formula: KGitem={(h1,r1,t1)∣h1∈Ehot, t1∈Et, r1∈Rt} Among them, h1 is the entity corresponding to the hot indicator, and the hot indicator refers to the indicator that ranks in the top preset percentage among all the referenced indicators in the original government indicator library; t1 is the entity corresponding to the project that references the hot indicator; r1 is the relationship between h1 and t1; Ehot is the set of hot indicators; Et is the set of all projects, and Rt is the set of relationships between Ehot and Et.

[0065] In a possible implementation of the present application, the government affairs indicator management device may also include: a knowledge prediction module.

[0066] Among them, the knowledge prediction module is used to: randomly initialize the entities and relationships in the indicator knowledge graph or the popular indicator knowledge graph into a multidimensional vector; reshape the multidimensional vector to obtain a two-dimensional vector; set two different void rates, split the two-dimensional vectors of entities and relationships into convolution kernels of different scales, and perform void rate convolution respectively to obtain the feature matrix of the entity and the feature matrix of the relationship; add the original information of the entity and the relationship to the feature matrix of the entity and the feature matrix of the relationship; use the attention mechanism to aggregate the feature information of the feature matrices of entities and the feature matrices of relationships of different scales to obtain a weighted matrix; stretch the weighted matrix and map it to the dimension of the tail entity embedding after full connection to obtain the tail entity matrix; perform an inner product operation on the feature matrix and the tail entity matrix to calculate the similarity, obtain the scores of the entity, relationship, and entity attribute value, and obtain the embedded representation of the entities and relationships of the indicator knowledge graph and the popular indicator knowledge graph through back propagation.

[0067] In a possible implementation of the present application, the knowledge prediction module is used to: utilize a multi-head graph attention network to aggregate the neighborhood information of an entity and assign different weights to different neighborhoods; utilize an attenuation mechanism to aggregate feature information of different neighborhoods of an entity and obtain the weight of the feature information.

[0068] In a possible implementation of the present application, the fusion module 303 is used to: update the indicator knowledge graph based on actual data, and link the updated indicator knowledge graph entities to the original government indicator library; use the correlation algorithm to determine the association strength between different entities, and perform knowledge fusion on different entities.

[0069] In a possible implementation of the present application, the fusion module 303 is used to: extract the indicator attributes of each indicator in the actual data in the node of the indicator knowledge graph; obtain other indicators associated with the current indicator based on the relationship of the indicator knowledge graph; update the indicator knowledge graph based on the indicator attributes of each indicator and other indicators associated with the current indicator.

[0070] In a possible implementation manner of the present application, the fusion module 303 is used for: knowledge fusion includes at least one of the following: name fusion, connotation fusion and dimension fusion.

[0071] In a possible implementation of the present application, the government affairs indicator management device may also include: an update module.

[0072] Among them, the update module is used to: determine the confidence of the update information when the update information is received, and the update information is at least one of the entity, relationship, and entity attribute value; when the confidence of the update information is greater than or equal to the confidence threshold, input the update information into the indicator knowledge graph, and use the update information to update the indicator knowledge graph; when the confidence of the update information is less than the confidence threshold, delete the update information.

[0073] The function of the government affairs index management device of this application has been Figure 1-2 The method embodiment shown is described in detail, so for any details not provided in the description of this embodiment, please refer to the relevant descriptions in the aforementioned embodiments, which will not be repeated here.

[0074] like Figure 4 As shown, an embodiment of the present application also provides a terminal device 400, including a processor 401, a memory 402, and a program or instruction stored in the memory 402 and executable on the processor 401. When the program or instruction is executed by the processor 401, each process of the above-mentioned government affairs indicator management processing method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0075] Optionally, the embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, each process of the above-mentioned government affairs index management method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0076] Optionally, an embodiment of the present application also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the various processes of the above-mentioned government affairs indicator management method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, they will not be repeated here.

[0077] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0078] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0079] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

Claims

1. A government affairs indicator management method, characterized in that: include: Acquire an original government indicator database, and preprocess the original government indicator database to obtain indicator information, wherein the indicator information includes entities, relationships, and entity attribute values, wherein the entities are indicators in the original government indicator database, and the relationships are relationships between entities; Based on the entities, relationships, and entity attribute values, construct an indicator knowledge graph; The indicator knowledge graph is integrated into the original government indicator database to obtain a fused government indicator database, which includes indicator names, indicator definitions, relationships between indicators, and indicator attributes.

2. The method according to claim 1, characterized in that The preprocessing of the original government affairs indicator database to obtain indicator information includes: Preprocessing the original government affairs indicator database, extracting indicator names from the original government affairs indicator database, and identifying entities, wherein the entities include entity definitions and application rules; Determine the relationship between entities based on the dependencies or associations between them; Based on the definition and application rules of the entity, entity attribute values ​​of the entity are extracted, and the entity attribute values ​​include at least one of the following: calculation logic, unit, spatial dimension, time dimension, and update frequency.

3. The method according to claim 1, characterized in that The constructing of an indicator knowledge graph based on the entities, relationships, and entity attribute values ​​includes: Construct the indicator knowledge graph according to the construction rules shown in the following formula: CKG={(h,r,t)∣h,t∈E, r∈R} Among them, h is the head entity, r is the tail entity; E is the set of all entities; r is the relationship; R is the set of all relationships.

4. The method according to claim 3, characterized in that The method further comprises: Construct a knowledge graph of popular indicators as shown below: KGitem={(h1,r1,t1)∣h1∈Ehot, t1∈Et, r1∈Rt} Among them, h1 is the entity corresponding to the hot indicator, and the hot indicator refers to the indicator ranked in the top preset percentage among all the referenced indicators in the original government indicator library; t1 is the entity corresponding to the project that references the hot indicator; r1 is the relationship between h1 and t1; Ehot is the set of hot indicators; Et is the set of all projects, and Rt is the set of relationships between Ehot and Et.

5. The method according to claim 4, characterized in that The method further comprises: Randomly initialize entities and relationships in the indicator knowledge graph or the popular indicator knowledge graph into a multidimensional vector; Reshape the multidimensional vector to obtain a two-dimensional vector; Setting two different dilation rates, splitting the two-dimensional vectors of the entity and the relationship into convolution kernels of different scales, and performing dilation rate convolution respectively to obtain the feature matrix of the entity and the feature matrix of the relationship; Adding original information of the entity and the relationship to the feature matrix of the entity and the feature matrix of the relationship; An attention mechanism is used to aggregate feature information of the feature matrix of the entity and the feature matrix of the relationship at different scales to obtain a weighted matrix; The weighted matrix is ​​stretched and mapped to the dimension of the tail entity embedding after full connection to obtain a tail entity matrix; The similarity is calculated by performing an inner product operation on the feature matrix and the tail entity matrix to obtain the scores of the entities, relationships, and entity attribute values, and the embedded representations of the entities and relationships of the indicator knowledge graph and the popular indicator knowledge graph are obtained through back propagation.

6. The method according to claim 5, characterized in that The method comprises: Use a multi-head graph attention network to aggregate the neighborhood information of entities and assign different weights to different neighborhoods; The feature information of different neighborhoods of an entity is aggregated using the attenuation mechanism to obtain the weight of the feature information.

7. The method according to claim 1, characterized in that The step of fusing the indicator knowledge graph into the original government indicator database to obtain a fused government indicator database includes: Update the indicator knowledge graph based on actual data, and link the updated indicator knowledge graph entity to the original government indicator library; The correlation algorithm is used to determine the association strength between different entities and to fuse the knowledge of different entities.

8. The method according to claim 7, characterized in that The updating process of the indicator knowledge graph based on actual data includes: Extracting the indicator attributes of each indicator in the actual data from the nodes of the indicator knowledge graph; Based on the relationship of the indicator knowledge graph, other indicators associated with the current indicator are obtained; Based on the indicator attributes of each indicator and other indicators associated with the current indicator, the indicator knowledge graph is updated.

9. The method according to claim 7, characterized in that: The knowledge fusion includes at least one of the following: name fusion, connotation fusion and dimension fusion.

10. The method according to claim 1, characterized in that The method further comprises: Upon receiving update information, determining a confidence level of the update information, wherein the update information is at least one of an entity, a relationship, and an entity attribute value; When the confidence of the updated information is greater than or equal to the confidence threshold, inputting the updated information into the indicator knowledge graph, and updating the indicator knowledge graph using the updated information; If the confidence of the update information is less than the confidence threshold, the update information is deleted.

11. A computer program product, characterized in that The computer program product comprises a computer program stored on a non-transitory computer-readable storage medium, wherein the computer program comprises program instructions, and when the program instructions are executed by a computer, the steps of the method according to any one of claims 1 to 10 are implemented.

Images