Enterprise knowledge graph optimization method and system, electronic device and storage medium

Abstract

The application provides an enterprise knowledge graph optimization method and system, electronic equipment and a storage medium, and belongs to the technical field of knowledge graphs. The method comprises the following steps: obtaining target enterprise information comprising unstructured data and preprocessed structured data; extracting initial enterprise entities and relationship information from the target enterprise information, and constructing an initial enterprise knowledge graph based on the initial enterprise entities and relationship information; screening target structured data meeting growth requirements from the structured data based on a preset growth rating model; and replacing corresponding structured data in the initial entities and relationship information with the target structured data, so as to optimize the initial enterprise knowledge graph to obtain a target enterprise knowledge graph. Through the application, the redundancy of the basic data in the initial enterprise knowledge graph can be screened to optimize the initial knowledge graph, so that the selected indicators from the knowledge graph meet the customer's expectations, thereby improving the data quality and data utilization rate.

Description

Technical Field

[0001] This invention belongs to the technical field of knowledge graphs, specifically relating to an enterprise knowledge graph optimization method, system, electronic device, and storage medium. Background Technology

[0002] The rise of the big data era and the shift in the research focus of artificial intelligence from perceptual intelligence to cognitive intelligence have fueled the popularity of knowledge graphs. Knowledge graphs, also known as scientific knowledge graphs, are a series of various graphs that display the development process and structural relationships of knowledge. They use visualization technology to describe knowledge resources and their carriers, mining, analyzing, constructing, drawing, and displaying knowledge and their interrelationships. As one of the application areas of artificial intelligence technology, knowledge graphs, with their powerful semantic processing and data structuring capabilities, provide a foundation for intelligent information applications. Knowledge graphs integrate, cross-reference, analyze, and compare large-scale data / knowledge by constructing semantic networks of entities and relationships, deeply mining the data to support intelligent understanding, representation, reasoning, retrieval, and services of knowledge, providing users with self-service iterative analysis capabilities.

[0003] Currently, the common approach to building an enterprise knowledge graph is to acquire dimensional information data from enterprises in the domain, then use methods such as natural language processing or deep learning to extract entities and relationships from the data, and finally construct the enterprise knowledge graph based on the extracted enterprise entity and relationship information. However, current enterprise knowledge graph construction focuses more on the integration and correlation of various structured or unstructured basic data, without addressing the effective filtering of redundancy in the basic data. This results in the selected indicators in the constructed knowledge graph not meeting customer expectations, and requiring a large storage space, meaning the data quality of the constructed knowledge graph is poor and fails to meet customer requirements. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method, system, electronic device, and storage medium for optimizing enterprise knowledge graphs. This method can filter out redundancy in the basic data of an initial enterprise knowledge graph to optimize the initial knowledge graph, ensuring that the indicators selected from the knowledge graph meet customer expectations, thereby improving data quality and data utilization.

[0005] In a first aspect, embodiments of this application provide a method for optimizing enterprise knowledge graphs, including:

[0006] Acquire target enterprise information, including unstructured data and pre-processed structured data;

[0007] Initial enterprise entity and relationship information is extracted from the target enterprise information, and an initial enterprise knowledge graph is constructed based on the initial enterprise entity and relationship information;

[0008] Based on a preset growth rating model, target structured data that meets growth requirements is selected from the structured data.

[0009] The target structured data replaces the corresponding structured data in the initial entity and relationship information, thereby optimizing the initial enterprise knowledge graph to obtain the target enterprise knowledge graph.

[0010] Preferably, the step of extracting initial enterprise entity and relationship information from the target enterprise information and constructing an initial enterprise knowledge graph based on the initial enterprise entity and relationship information specifically includes:

[0011] The structured data is extracted according to preset conditions to obtain entity and relationship information that meets preset standards;

[0012] The unstructured enterprise data is then used to identify enterprise entities and extract relationships using natural language processing or deep learning techniques.

[0013] Based on the processing results of the structured data and the unstructured data, initial enterprise entity and relationship information is obtained;

[0014] An initial enterprise knowledge graph is constructed based on the initial enterprise entity and relationship information.

[0015] Preferably, the specific steps for selecting target structured data that meets growth requirements from the structured data based on a preset growth rating model include:

[0016] The structured data is normalized to obtain indicator data, and an enterprise indicator system is constructed based on the indicator data;

[0017] The objective weighting method is used to calculate the weights for the aforementioned indicator data;

[0018] Analyze the weighting results of the indicator data, and filter out the target indicator data from the indicator data;

[0019] The target indicator data is re-weighted, and the weighted sum of the values ​​of each indicator in the target indicator data and their corresponding weights is calculated.

[0020] The result obtained by the weighted sum and renormalization process is the growth score of the target company.

[0021] Output the target structured data corresponding to the target indicator data that meets the growth requirements.

[0022] Preferably, the specific steps of normalizing the structured data to obtain indicator data and constructing an enterprise indicator system based on the indicator data include:

[0023] The structured data is integrated and extracted, outliers are filtered out and removed, and missing values ​​in the structured data are supplemented based on the average of the target company over the past three years.

[0024] The integrated and extracted structured data is normalized to obtain indicator data, which includes financial indicators and non-financial indicators.

[0025] Construct an enterprise indicator system based on the aforementioned indicator data.

[0026] Preferably, the step of outputting the target structured data corresponding to the target indicator data that meets the growth requirements specifically includes:

[0027] Based on the growth score, the target companies are classified into growth stage, maturity stage, or decline stage.

[0028] The classification effect of the target enterprises is verified by the classification algorithm, wherein the verification refers to the verification of the precision, recall, F1 score and AUC score results;

[0029] Output the target structured data corresponding to the target indicator data that meets the growth requirements.

[0030] Preferably, the step of replacing the corresponding structured data in the initial entity and relationship information with the target structured data to optimize the initial enterprise knowledge graph and obtain the target enterprise knowledge graph specifically includes:

[0031] The target structured data is sorted by importance, and important structured data with a preset importance level are selected, wherein the importance is assessed according to the index score.

[0032] Based on the initial enterprise knowledge graph, the important structured data is used to replace the corresponding structured data in the initial enterprise entity and relationship information to optimize the initial enterprise knowledge graph, thereby obtaining the target enterprise knowledge graph.

[0033] Preferably, the preprocessing of structured data includes one or more of the following: missing value handling, outlier handling, or transform-merge value handling.

[0034] Secondly, embodiments of this application provide an enterprise knowledge graph optimization system, including:

[0035] The acquisition module is used to acquire target enterprise information, including unstructured data and pre-processed structured data.

[0036] The construction module is used to extract initial enterprise entity and relationship information from the target enterprise information, and construct an initial enterprise knowledge graph based on the initial enterprise entity and relationship information;

[0037] The filtering module is used to filter out target structured data that meets the growth requirements from the structured data based on a preset growth rating model.

[0038] An optimization module is used to replace the corresponding structured data in the initial entity and relationship information with the target structured data, so as to optimize the initial enterprise knowledge graph to obtain the target enterprise knowledge graph.

[0039] Preferably, the building module includes:

[0040] An extraction unit is used to extract entity and relationship information that conforms to preset standards from the structured data according to preset conditions.

[0041] The extraction unit is used to perform enterprise entity identification and relationship extraction on the unstructured enterprise data using natural language processing or deep learning techniques;

[0042] The acquisition unit is used to acquire initial enterprise entity and relationship information based on the processing results of the structured data and the unstructured data;

[0043] The construction unit is used to construct an initial enterprise knowledge graph based on the initial enterprise entities and relationship information.

[0044] Preferably, the filtering module includes:

[0045] The first processing unit is used to normalize the structured data to obtain indicator data, and to construct an enterprise indicator system based on the indicator data.

[0046] The first calculation unit is used to perform weight calculation on the indicator data using an objective weighting method;

[0047] A filtering unit is used to analyze the weighting results of the indicator data and filter out target indicator data from the indicator data;

[0048] The second calculation unit is used to recalculate the weights of the target indicator data and calculate the weighted sum of the values ​​of each indicator in the target indicator data and their corresponding weights.

[0049] The second processing unit is used to process the weighted sum and the result of the renormalization process, which is the growth score of the target enterprise.

[0050] The output unit is used to output the target structured data corresponding to the target indicator data that meets the growth requirements.

[0051] Preferably, the optimization module includes:

[0052] The sorting unit is used to sort the target structured data according to importance and select important structured data with a preset importance level, wherein the importance is assessed according to the index score.

[0053] The replacement unit is used to replace the structured data corresponding to the initial enterprise entity and relationship information with the important structured data based on the initial enterprise knowledge graph to optimize the initial enterprise knowledge graph and obtain the target enterprise knowledge graph.

[0054] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the enterprise knowledge graph optimization method as described in the first aspect above.

[0055] Fourthly, embodiments of this application provide a storage medium storing a computer program that, when executed by a processor, implements the enterprise knowledge graph optimization method as described in the first aspect above.

[0056] Compared to existing technologies, this application provides a method, system, electronic device, and storage medium for optimizing enterprise knowledge graphs. First, it collects structured data of the target enterprise's basic information and operational and financial information, processing invalid data, and collects unstructured textual data such as relevant policies and public opinion regarding the target enterprise to obtain target enterprise information. Second, it extracts entity and relationship information from the processed structured data according to preset conditions to obtain standardized enterprise entity and relationship information. It also uses natural language processing technology to perform entity recognition and relationship extraction on the unstructured data and identifies the nature of enterprise news and public opinion based on deep learning algorithms to obtain initial enterprise entity and relationship information, and constructs an initial enterprise knowledge graph based on this. Third, it constructs a growth rating model based on machine learning algorithms and selects target structured data that meets growth requirements based on the growth rating model. Finally, it sorts the target structured data by importance, selects important structured data of preset importance levels, and replaces the corresponding structured data in the initial enterprise entity and relationship information with the important structured data to obtain the target enterprise knowledge graph. By following the steps above, redundancy in the basic data of the initial enterprise knowledge graph can be filtered out to optimize the initial knowledge graph, ensuring that the indicators selected from the knowledge graph meet customer expectations and improve data quality and data utilization. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 A flowchart of the enterprise knowledge graph optimization method provided in Embodiment 1 of the present invention;

[0059] Figure 2 This is a detailed flowchart of step S103 provided in Embodiment 1 of the present invention;

[0060] Figure 3 This is a structural block diagram of an enterprise knowledge graph optimization system corresponding to the method in Embodiment 1, provided in Embodiment 2 of the present invention;

[0061] Figure 4 This is a schematic diagram of the hardware structure of the electronic device provided in Embodiment 3 of the present invention.

[0062] Explanation of reference numerals in the attached figures:

[0063] 10-Acquisition Module;

[0064] 20 - Building module, 21 - Extraction unit, 22 - Extraction unit, 23 - Acquisition unit, 24 - Building unit;

[0065] 30-Filtering module, 31-First processing unit, 32-First calculation unit, 33-Filtering unit, 34-Second calculation unit, 35-Second processing unit, 36-Output unit;

[0066] 40 - Optimization module, 41 - Sorting unit, 42 - Replacement unit;

[0067] 50 - Bus, 51 - Processor, 52 - Memory, 53 - Communication interface. Detailed Implementation

[0068] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain embodiments of the present invention, and should not be construed as limiting the present invention.

[0069] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0070] Currently, the construction of enterprise knowledge graphs focuses more on the integration and correlation of various structured or unstructured basic data, without addressing the redundancy in the basic data. This results in the selected indicators in the constructed knowledge graph not meeting customer expectations, as well as requiring a large storage space. In other words, the constructed knowledge graph has poor data quality and low data utilization.

[0071] To address this, the present invention proposes an enterprise knowledge graph optimization method, system, electronic device, and storage medium. By filtering out redundancy in the basic data of the initial enterprise knowledge graph, the initial knowledge graph is optimized, ensuring that the indicators selected from the knowledge graph meet customer expectations, thereby improving data quality and data utilization.

[0072] Example 1

[0073] This embodiment provides a method for optimizing enterprise knowledge graphs. Figure 1 This is a flowchart of the enterprise knowledge graph optimization method in this embodiment, such as... Figure 1 As shown, the process includes the following steps:

[0074] S101, Obtain target enterprise information including unstructured data and pre-processed structured data;

[0075] Structured data includes character-based data such as basic enterprise information and numerical data such as enterprise operating and financial information. Basic enterprise information refers to basic enterprise information such as company name, address, and registration date, while enterprise operating and financial information refers to financial information such as profit data, asset and liability data, and cash flow data. Unstructured data refers to text data such as enterprise-related policies and public opinion data, which serve as training data for downstream tasks and source data for enterprise public opinion data displayed on the front end.

[0076] Specifically, different methods are used for invalid data (document character type) in basic enterprise information: 1. Missing value handling: If the collected data has serious missing values, delete the data and fill in the missing values ​​using web crawlers. 2. Outlier handling: If the collected data has minor missing values, replace them with empirical numbers. 3. Transformed and merged value handling: For example, convert date data with formatting issues. Different methods are used for invalid data (numerical type) in enterprise operating and financial information: 1. Missing value handling: If the collected data has missing values, delete the data if the missing values ​​are particularly serious; if the missing values ​​are minor, fill in the missing values ​​using methods such as median, average, mode, or algorithmic fill. 2. Outlier handling: If the collected data has outliers, the handling methods include deletion, estimation using the average or median, and logarithmic transformation of the variables.

[0077] S102, extract initial enterprise entity and relationship information from the target enterprise information, and construct an initial enterprise knowledge graph based on the initial enterprise entity and relationship information;

[0078] Knowledge graphs essentially store objective experience within a vast network. Nodes represent entities or concepts, and edges represent semantic relationships between entities / concepts. Mature graph databases such as Neo4j, Dgraph, and JanusGraph can be used to store knowledge graphs. Companies like Google and Baidu use top-down algorithms to extract knowledge graphs and then categorize them into existing schemas; if a schema cannot be found, they need to generate a new schema for pattern matching. Besides the schema, the process of building a knowledge graph involves filling it with content—this process is called knowledge extraction.

[0079] Furthermore, the specific steps of step S102 include:

[0080] S1021, Extract entity and relationship information that conforms to preset standards from the structured data according to preset conditions;

[0081] The preset conditions refer to the entities and relationships related to the target company that are extracted. Entities include: the company's Chinese name, company ID, names of senior executives, and names of competitors; relationships include: the company's external investment relationships, supplier relationships, customer relationships, and shareholder holding relationships. The preset standards refer to the standards that meet the Neo4j graph data storage format.

[0082] Specifically, the collected character data such as basic enterprise information and numerical data such as enterprise operation and financial information of the target enterprises are processed into a data storage format that can meet the requirements of graph data storage. The purpose is to adjust the format of the target enterprises' structured data so that it can be used in the data format required for knowledge graph construction.

[0083] S1022, The unstructured enterprise data is used to perform enterprise entity identification and relationship extraction using natural language processing or deep learning technology;

[0084] Natural Language Processing (NLP) is a collective term for all technologies related to computer processing of natural language. Its purpose is to enable computers to understand and accept instructions input by humans in natural language, and to perform translation functions from one language to another. NLP technology can specifically employ LSTM-CRF, where LSTM stands for Long Shot-Term Memory Neural Network and CRF stands for Conditional Random Field. The nature of corporate news sentiment refers to whether the public sentiment surrounding the news item tends to be positive, neutral, or negative, and is used as a feature for downstream tasks.

[0085] S1023, Based on the processing results of the structured data and the unstructured data, obtain the initial enterprise entity and relationship information.

[0086] S1024, Construct an initial enterprise knowledge graph based on the initial enterprise entity and relationship information.

[0087] S103, Based on the preset growth rating model, select target structured data that meets the growth requirements from the structured data;

[0088] Enterprise growth assessment, in particular, is the judgment and evaluation of an enterprise's future growth potential. It involves identifying the enterprise's future development trends and potential, including measuring the direction, speed, capabilities, and results of its development. The core of enterprise growth assessment is the evaluation of an enterprise's growth space and potential.

[0089] Specifically, the pre-defined growth rating model reflects a company's comprehensive growth capability from different perspectives by setting indicators, including expansion capability, profitability, operational capability, cash strength, and technological innovation capability. All these indicators are positive, and there is a linear relationship between each growth evaluation indicator and the company's growth potential. Therefore, the established growth evaluation model is a multivariate linear evaluation model.

[0090] Furthermore, such as Figure 2 As shown, the specific steps of step S103 include:

[0091] S1031, Normalize the structured data to obtain indicator data, and construct an enterprise indicator system based on the indicator data;

[0092] Step S1031 is completed through the following processes:

[0093] S10311, Integrate and extract the structured data, filter and remove outliers in the structured data, and supplement the null values ​​in the structured data based on the average of the target company over the past three years;

[0094] Specifically, if there are outliers in the extracted structured data, it indicates that the data is unreliable. In this case, the outliers are directly removed to avoid the indicators corresponding to the outliers affecting the evaluation results. For the null values, the average value of the indicator for the target company over the past three years is used to fill the null values ​​so as to minimize the impact of the null values.

[0095] S10312, Normalize the integrated and extracted structured data to obtain indicator data, wherein the indicator data includes financial indicators and non-financial indicators;

[0096] Specifically, because the differences between different structured data are very large, such as growth rates, which are generally between 0 and 1, and absolute values, some indicators will have large values ​​and great influence if the data is not normalized, which is unfair to the evaluation of other structured data. By normalizing, the impact of various structured data on the final evaluation result is effectively balanced, making the final calculation of the enterprise growth score more reasonable.

[0097] S10313, Construct an enterprise indicator system based on the aforementioned indicator data.

[0098] S1032, The objective weighting method is used to calculate the weights for the indicator data;

[0099] Specifically, the objective weighting method determines weights based on the relationships between preprocessed indicator data using certain mathematical methods. Its judgment does not rely on subjective human judgment and has a strong mathematical theoretical basis. As an example, and not a limitation, the use of the combined objective weighting method for calculating preprocessed indicator data in this embodiment is merely a preferred calculation method; this application does not specifically limit the method for calculating the weights of preprocessed indicator data.

[0100] S1033, Analyze the weighting results of the indicator data, and filter out the target indicator data from the indicator data;

[0101] Specifically, by analyzing the weighting results of the indicator data, indicator data with low weight values ​​and high similarity are filtered out to make the filtered indicator data more representative.

[0102] S1034, recalculate the weights of the target indicator data, and calculate the weighted sum of the values ​​of each indicator in the target indicator data and their corresponding weights.

[0103] S1035, The result obtained by the weighted sum and renormalization process is the growth score of the target enterprise;

[0104] Specifically, by calculating the weighted sum of the values ​​of the selected indicators and their corresponding weights, and by normalizing the scoring method, the final enterprise growth score has a high accuracy rate, and the calculation method has good anti-overfitting and anti-noise characteristics. This makes the evaluation results of enterprise growth more reliable.

[0105] S1036, Output the target structured data corresponding to the target indicator data that meets the growth requirements;

[0106] Specifically, step S1036 is completed through the following processes:

[0107] S10361, Based on the growth score, the target company is divided into a growth stage, a mature stage, or a decline stage;

[0108] Specifically, by dividing target companies into growth, maturity, and decline stages, and training and evaluating the growth scores of target companies, the growth stage of the target companies is defined, thereby achieving the goal of rating the target companies.

[0109] S10362, The classification effect of the target enterprise is verified by the classification algorithm, wherein the verification refers to the verification of the precision, recall, F1 score and AUC score results;

[0110] The classification algorithms include the Random Forest algorithm and the XGBoost algorithm. The F1 score is a special case of the F-Measure; it is the harmonic mean of precision and recall, comprehensively reflecting the accuracy of the classification results. The closer the result is to 1, the higher the accuracy. The AUC (Area Under the Curve) score is a standard used to measure the quality of a classification model; the AUC score is a comprehensive indicator for evaluating the accuracy of a classification model.

[0111] Specifically, training and evaluation results for various partitioning combinations are obtained through random forest and Xgboost algorithms. Growth scores are divided into corresponding enterprise lifecycles based on locally optimal partitioning methods, and the final rating effect is verified through multi-classification algorithms.

[0112] S10363, Output the target structured data corresponding to the target indicator data that meets the growth requirements.

[0113] S104, replace the corresponding structured data in the initial entity and relationship information with the target structured data, so that the initial enterprise knowledge graph can be optimized to obtain the target enterprise knowledge graph;

[0114] Furthermore, the specific steps of step S104 include:

[0115] S1041, Sort the target structured data according to importance and select important structured data with a preset importance level, wherein the importance is evaluated according to the index score.

[0116] S1042, Based on the initial enterprise knowledge graph, replace the structured data corresponding to the initial enterprise entity and relationship information with the important structured data to optimize the initial enterprise knowledge graph and obtain the target enterprise knowledge graph.

[0117] By following the steps above, redundancy in the basic data of the initial enterprise knowledge graph can be filtered out to optimize the initial knowledge graph, ensuring that the indicators selected from the knowledge graph meet customer expectations and improve data quality and data utilization.

[0118] Example 2

[0119] This embodiment provides a structural block diagram of a system corresponding to the method described in Embodiment 1. Figure 3 This is a structural block diagram of an enterprise knowledge graph optimization system according to an embodiment of this application, such as... Figure 3 As shown, the system includes:

[0120] The acquisition module 10 is used to acquire target enterprise information, including unstructured data and preprocessed structured data;

[0121] The construction module 20 is used to extract initial enterprise entity and relationship information from the target enterprise information, and construct an initial enterprise knowledge graph based on the initial enterprise entity and relationship information;

[0122] The filtering module 30 is used to filter out target structured data that meets the growth requirements from the structured data based on a preset growth rating model.

[0123] The optimization module 40 is used to replace the corresponding structured data in the initial entity and relationship information with the target structured data, so as to optimize the initial enterprise knowledge graph to obtain the target enterprise knowledge graph.

[0124] Preferably, the building module 20 includes:

[0125] Extraction unit 21 is used to extract entity and relationship information that conforms to preset standards from the structured data according to preset conditions;

[0126] Extraction unit 22 is used to perform enterprise entity recognition and relationship extraction on the unstructured enterprise data using natural language processing or deep learning technology;

[0127] The acquisition unit 23 is used to acquire initial enterprise entity and relationship information based on the processing results of the structured data and the unstructured data;

[0128] Construction unit 24 is used to construct an initial enterprise knowledge graph based on the initial enterprise entity and relationship information.

[0129] Preferably, the 30 filtering modules include:

[0130] The first processing unit 31 is used to normalize the structured data to obtain indicator data, and to construct an enterprise indicator system based on the indicator data.

[0131] The first calculation unit 32 is used to perform weight calculation on the indicator data using an objective weighting method;

[0132] The filtering unit 33 is used to analyze the weight results of the indicator data and filter out the target indicator data from the indicator data;

[0133] The second calculation unit 34 is used to recalculate the weights of the target indicator data and calculate the weighted sum of the values ​​of each indicator in the target indicator data and their corresponding weights.

[0134] The second processing unit 35 is used to process the weighted sum and the result of the normalization process as the growth score of the target enterprise.

[0135] Output unit 36 ​​is used to output the target structured data corresponding to the target indicator data that meets the growth requirements.

[0136] Preferably, the optimization module 40 includes:

[0137] The sorting unit 41 is used to sort the target structured data according to importance and select important structured data with a preset importance level, wherein the importance is evaluated according to the index score.

[0138] Replacement unit 42 is used to replace the structured data corresponding to the initial enterprise entity and relationship information with the important structured data based on the initial enterprise knowledge graph to optimize the initial enterprise knowledge graph and obtain the target enterprise knowledge graph.

[0139] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can reside in the same processor; or the above modules can be located in different processors in any combination.

[0140] Example 3

[0141] Combination Figure 1 The enterprise knowledge graph optimization method of the present invention described herein can be implemented by electronic devices. Figure 4 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of this application.

[0142] The electronic device may include a processor 51 and a memory 52 storing computer program instructions.

[0143] Specifically, the processor 51 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0144] The memory 52 may include a large-capacity storage device for data or instructions. For example, and not limitingly, the memory 52 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 52 may include removable or non-removable (or fixed) media. Where appropriate, the memory 52 may be internal or external to a data processing device. In a particular embodiment, the memory 52 is non-volatile memory. In a particular embodiment, the memory 52 includes read-only memory (ROM) and random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), an electrically alterable read-only memory (EAROM), or flash memory, or a combination of two or more of these. Where appropriate, the RAM can be Static Random-Access Memory (SRAM) or Dynamic Random-Access Memory (DRAM). DRAM can be Fast Page Mode Dynamic Random-Access Memory (FPMDRAM), Extended Data Out Dynamic Random-Access Memory (EDODRAM), Synchronous Dynamic Random-Access Memory (SDRAM), etc.

[0145] The memory 52 can be used to store or cache various data files that need to be processed and / or used for communication, as well as possible computer program instructions executed by the processor 51.

[0146] The processor 51 implements the enterprise knowledge graph optimization method of Embodiment 1 by reading and executing computer program instructions stored in the memory 52.

[0147] In some embodiments, the electronic device may further include a communication interface 53 and a bus 50. For example, Figure 4 As shown, the processor 51, memory 52, and communication interface 53 are connected through bus 50 and complete communication with each other.

[0148] The communication interface 53 is used to enable communication between the various modules, devices, units, and / or equipment in the embodiments of this application. The communication interface 53 can also enable data communication with other components such as external devices, image / data acquisition devices, databases, external storage, and image / data processing workstations.

[0149] Bus 50 includes hardware, software, or both, that couples components of an electronic device together. Bus 50 includes, but is not limited to, at least one of the following: data bus, address bus, control bus, expansion bus, and local bus. For example, and not as a limitation, bus 50 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 50 may include one or more buses. Although specific buses are described and illustrated in the embodiments of this application, this application considers any suitable bus or interconnection.

[0150] The electronic device can execute the enterprise knowledge graph optimization method of Embodiment 1 of this application based on the enterprise knowledge graph optimization system.

[0151] Furthermore, in conjunction with the enterprise knowledge graph optimization method in Embodiment 1 above, this application embodiment can provide a storage medium for implementation. This storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement the enterprise knowledge graph optimization method of Embodiment 1 above.

[0152] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0153] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An enterprise knowledge graph optimization method, characterized in that, The method comprises the following steps: obtaining target enterprise information including unstructured data and preprocessed structured data; extracting initial enterprise entity and relationship information from the target enterprise information, and constructing an initial enterprise knowledge graph based on the initial enterprise entity and relationship information; screening target structured data meeting the growth requirement from the structured data based on a preset growth rating model; replacing the corresponding structured data in the initial enterprise entity and relationship information with the target structured data to optimize the initial enterprise knowledge graph and obtain a target enterprise knowledge graph; the step of extracting initial enterprise entity and relationship information from the target enterprise information and constructing an initial enterprise knowledge graph based on the initial enterprise entity and relationship information comprises the following steps: extracting entity and relationship information meeting the preset standard from the structured data according to the preset condition; performing enterprise entity recognition and relationship extraction on the unstructured data by using natural language processing or deep learning technology; obtaining initial enterprise entity and relationship information based on the processing results of the structured data and the unstructured data; constructing an initial enterprise knowledge graph based on the initial enterprise entity and relationship information; the specific steps of screening target structured data meeting the growth requirement from the structured data based on a preset growth rating model comprise the following steps: normalizing the structured data to obtain index data, and constructing an enterprise index system according to the index data; calculating the weight of the index data by using an objective weighting method; analyzing the weight result of the index data, and screening target index data from the index data; recalculating the weight of the target index data, and calculating the weighted sum of the numerical value of each index in the target index data and the corresponding weight; the result obtained by re-normalizing the weighted sum is the growth score of the target enterprise; outputting the target structured data corresponding to the target index data meeting the growth requirement. 2.The enterprise knowledge graph optimization method of claim 1, wherein, The specific steps of normalizing the structured data to obtain index data, and constructing an enterprise index system according to the index data comprise the following steps: integrating and extracting the structured data, screening and removing outliers in the structured data, and supplementing the null values in the structured data based on the average value of the target enterprise in the past three years; normalizing the integrated and extracted structured data to obtain index data, wherein the index data includes financial indicators and non-financial indicators; constructing an enterprise index system according to the index data. 3.The enterprise knowledge graph optimization method of claim 1, wherein, The step of outputting the target structured data corresponding to the target index data meeting the growth requirement comprises the following steps: dividing the target enterprise into a growth period, a mature period or a recession period according to the growth score; verifying the classification effect of the target enterprise by using a classification algorithm, wherein the verification refers to the verification of the precision, recall, F1 value and AUC value results; outputting the target structured data corresponding to the target index data meeting the growth requirement. 4.The enterprise knowledge graph optimization method of claim 1, wherein, The step of replacing the target structured data in the initial enterprise entity and relationship information to optimize the initial enterprise knowledge graph to obtain a target enterprise knowledge graph specifically comprises: The target structured data is sorted based on importance, and important structured data of a preset importance level is selected, wherein the importance is evaluated according to the high-low score of an index; The important structured data is replaced in the corresponding structured data in the initial enterprise entity and relationship information based on the initial enterprise knowledge graph to optimize the initial enterprise knowledge graph, and a target enterprise knowledge graph is obtained.

5. The enterprise knowledge graph optimization method of any one of claims 1-4, wherein, The preprocessing of the structured data includes one or a combination of more than two of missing value processing, abnormal value processing or transformed and merged value processing.

6. An enterprise knowledge graph optimization system, comprising: Comprises: An acquisition module configured to acquire target enterprise information including unstructured data and preprocessed structured data; A construction module configured to extract initial enterprise entity and relationship information from the target enterprise information, and construct an initial enterprise knowledge graph based on the initial enterprise entity and relationship information; The construction module is specifically configured to extract entity and relationship information meeting a preset standard from the structured data according to a preset condition; The unstructured data is subjected to enterprise entity recognition and relationship extraction by using natural language processing or deep learning technology; Initial enterprise entity and relationship information is acquired based on the processing results of the structured data and the unstructured data; An initial enterprise knowledge graph is constructed based on the initial enterprise entity and relationship information; A screening module configured to screen target structured data meeting growth requirements from the structured data based on a preset growth rating model; The screening module is specifically configured to normalize the structured data to obtain index data, and construct an enterprise index system according to the index data; An objective weighting method is used to calculate the weight of the index data; The weight result of the index data is analyzed, and target index data is screened from the index data; The target index data is reweighted, and the weighted sum of the numerical value of each index in the target index data and the corresponding weight is calculated; The result obtained by re-normalizing the weighted sum is the growth score of the target enterprise; Target structured data corresponding to the target index data meeting the growth requirements is outputted; An optimization module configured to replace the target structured data in the initial enterprise entity and relationship information to optimize the initial enterprise knowledge graph to obtain a target enterprise knowledge graph.

7. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to realize the enterprise knowledge graph optimization method of any one of claims 1 to 5.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to realize the enterprise knowledge graph optimization method of any one of claims 1 to 5.

Images