A digital economic data classification management method

Abstract

The present invention provides a digital economic data classification management method, which relates to the field of information management technology. The method includes: obtaining the data source name of the data to be managed; determining the source classification condition coefficient of the data to be managed; obtaining the relevant information of the data to be managed; training the data value prediction model to obtain the trained data value prediction model; obtaining the data basic value score and data circulation rate score of the data to be managed; determining the value classification condition coefficient; performing the first classification management on the data to be managed according to the source classification condition coefficient; and performing the second classification management on the data to be managed according to the value classification condition coefficient. According to the present invention, the value of the data can be determined based on the accurate analysis of the basic value and circulation rate of the data, and based on the basic value and circulation rate of the data. Further, the data can be classified according to the source status and value status of the data, thereby improving the accuracy and comprehensiveness of the classification management of digital economic data.

Description

Technical Field

[0001] The present invention relates to the technical field of information management, and in particular to a digital economic data classification management method. Background Art

[0002] In the related technology, CN117633897A discloses a digital economic information management system based on big data, which belongs to the field of economic information management technology, including an economic information acquisition module, a data mining and analysis module and an information management strategy determination module; in this scheme, an information management strategy determination module is provided in the system, which can clarify the focus of information management and ensure the management effect. At the same time, the data in the system can be quickly traced. At the same time, a specific data analysis module is also provided in the system, which can analyze the data in different time periods, so as to quickly understand the economic situation in different time periods, which is conducive to the subsequent targeted economic information management. A system security protection module is also provided in the system, which can effectively protect the economic information and effectively avoid information leakage. At the same time, it can also generate economic adjustment strategies, which are conducive to adjusting the economic situation, effectively solving the problems arising from economic information, and improving the actual application effect of the system.

[0003] CN116956353A discloses a multi-channel data collection method and device based on digital economy. Public information on the Internet is collected and acquired through a web crawler, and the collected digital economic data information is screened, and then the classified digital economic data information is privacy processed. In the privacy processing process, the digital economic data information needs to be desensitized first, and the desensitized information is legalized, and then the legalized data information is used as the digital economic data information to be extracted, so that the acquired data can avoid circumventing user privacy, and thus can avoid the problem of infringement of acquired information.

[0004] Based on the above-mentioned related technologies, economic information can be effectively protected. However, the related technologies cannot classify and manage digital economic data, that is, they cannot classify and manage the data to be managed according to the data source and data value of the data to be managed.

[0005] The information disclosed in the background technology section of this application is only intended to deepen the understanding of the general background technology of this application, and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to those skilled in the art. Summary of the invention

[0006] The present invention provides a digital economic data classification management method, which can solve the technical problem that related technologies are unable to classify and manage the data to be managed according to the data source and data value of the data to be managed.

[0007] According to a first aspect of the present invention, a digital economic data classification management method is provided, comprising:

[0008] Get the data source name of the data to be managed;

[0009] According to the data source name, determine the source classification condition coefficient of the data to be managed;

[0010] Obtain relevant information of the data to be managed, wherein the relevant information includes: data volume, number of quantity links, data update frequency, data accuracy and data usage scope;

[0011] Training the data value prediction model to obtain a trained data value prediction model;

[0012] The relevant information is processed through the trained data value prediction model to obtain the data basic value score and data circulation rate score of the data to be managed;

[0013] Determining a value classification condition coefficient according to the data basic value score and the data circulation rate score;

[0014] According to the source classification condition coefficient, the data to be managed is classified and managed for the first time;

[0015] After the first classification management is completed, the data to be managed is further classified for the second time according to the value classification condition coefficient.

[0016] According to a second aspect of the present invention, there is provided a digital economic data classification management system, comprising:

[0017] Data source name module, obtains the data source name of the data to be managed;

[0018] A source classification condition coefficient module determines the source classification condition coefficient of the data to be managed according to the data source name;

[0019] The data information module obtains relevant information of the data to be managed, wherein the relevant information includes: data volume, number of quantity links, data update frequency, data accuracy and data usage scope;

[0020] The prediction model module trains the data value prediction model to obtain the trained data value prediction model;

[0021] The data scoring module processes the relevant information through the trained data value prediction model to obtain the data basic value score and data circulation rate score of the data to be managed;

[0022] A value classification condition coefficient module determines a value classification condition coefficient according to the data basic value score and the data circulation rate score;

[0023] A first classification management module performs first classification management on the data to be managed according to the source classification condition coefficient;

[0024] The second classification management module, after completing the first classification management, continues to perform a second classification management on the management data according to the value classification condition coefficient.

[0025] Technical effect: According to the present invention, the data source of the data to be managed can be judged according to the data source name of the data to be managed, and the data basic value score and data circulation rate score of the data to be managed can be obtained through the trained data value prediction model. Further, the data value of the data to be managed can be evaluated according to the data basic value score and the data circulation rate score. Further, the data to be managed can be classified and managed according to the two aspects of data source and data value of the data to be managed, thereby improving the accuracy and comprehensiveness of the classification management of digital economic data. The source classification condition coefficient of the data to be managed can be determined according to the data source name vector, the known enterprise internal department name vector, the known external partner name vector, the known public data source name vector and the sorting sequence number. During the calculation process, the data source name vector can be respectively subjected to cosine similarity operations with the known enterprise internal department name vector, the known external partner name vector and the known public data source name vector to determine the specific source of the data to be managed, thereby improving the accuracy and comprehensiveness of the source classification condition coefficient. The training loss function of the data value prediction model can be determined based on the training data related information, the expert training data basic value score, the expert training data circulation rate score, the sample training data circulation rate score and the sample training data basic value score. In the calculation process, the influence of the three aspects of training data volume, training data link number and training data accuracy on the error of data basic value, as well as the influence of training data update frequency and training data usage scope on the error of training data circulation rate are fully considered to set the training loss function, so that the training loss function of the load forecasting model can be reduced during the training process, and the accuracy of the data value prediction model can be improved in a more targeted manner.

[0026] It should be understood that the above general description and the following detailed description are exemplary and explanatory only and do not limit the present invention. Other features and aspects of the present invention will become more apparent from the following detailed description of exemplary embodiments with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other embodiments can be obtained based on these drawings without creative work.

[0028] Figure 1 A flowchart of a method for classifying and managing digital economic data according to an embodiment of the present invention is exemplarily shown;

[0029] Figure 2 A block diagram of a digital economic data classification management system according to an embodiment of the present invention is exemplarily shown. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0032] Figure 1 A flowchart of a method for classifying and managing digital economic data according to an embodiment of the present invention is exemplarily shown, and the method includes:

[0033] Step S101, obtaining the data source name of the data to be managed;

[0034] Step S102, determining the source classification condition coefficient of the data to be managed according to the data source name;

[0035] Step S103, obtaining relevant information of the data to be managed, wherein the relevant information includes: data volume, number of quantity links, data update frequency, data accuracy and data usage scope;

[0036] Step S104, training the data value prediction model to obtain a trained data value prediction model;

[0037] Step S105, processing the relevant information through the trained data value prediction model to obtain the data basic value score and data circulation rate score of the data to be managed;

[0038] Step S106, determining a value classification condition coefficient according to the data basic value score and the data circulation rate score;

[0039] Step S107, performing a first classification management on the data to be managed according to the source classification condition coefficient;

[0040] Step S108, after completing the first classification management, continue to perform a second classification management on the data to be managed according to the value classification condition coefficient.

[0041] According to the digital economy data classification management method of an embodiment of the present invention, the data source of the data to be managed can be judged according to the data source name of the data to be managed, and the data basic value score and data circulation rate score of the data to be managed can be obtained through the trained data value prediction model. Furthermore, the data value of the data to be managed can be evaluated according to the data basic value score and the data circulation rate score. Furthermore, the data to be managed can be classified and managed according to the two aspects of the data source and data value of the data to be managed, thereby improving the accuracy and comprehensiveness of the digital economy data classification management.

[0042] According to an embodiment of the present invention, in step S101, the data source name of the data to be managed is obtained.

[0043] For example, through a professional data collection platform, obtain the data source name of the data to be managed.

[0044] According to an embodiment of the present invention, in step S102, the source classification condition coefficient of the data to be managed is determined according to the data source name.

[0045] According to one embodiment of the present invention, step S102 includes:

[0046] Obtain multiple known enterprise internal department names, known external partner names, and known public data source names;

[0047] Sort multiple known enterprise internal department names, known external partner names, and known public data source names respectively, and determine the sorting sequence numbers of the multiple known enterprise internal department names, known external partner names, and known public data source names;

[0048] Determine a data source name vector according to the data source name;

[0049] Determine a known enterprise internal department name vector according to the known enterprise internal department name;

[0050] Determine a known external partner name vector according to the known external partner name;

[0051] Determine a known public data source name vector according to the known public data source name;

[0052] The source classification condition coefficient of the data to be managed is determined based on the data source name vector, the known enterprise internal department name vector, the known external partner name vector, the known public data source name vector and the sorting sequence number.

[0053] For example, there are usually multiple departments within an enterprise that are involved in the generation of data. Through the enterprise's internal system, obtain the names of multiple known internal departments of the enterprise. Through the enterprise's cooperation agreements, business transaction records, project documents and other channels, obtain the names of known external partners. Obtain the names of known public data sources from public databases; sort the names of multiple known internal departments according to the importance of the department in the enterprise and the frequency of data generation. The more important the internal department of the enterprise, the smaller the sorting number. Sort the names of multiple known external partners according to the closeness of cooperation and the stability of cooperation. The more important the partner of the enterprise, the smaller the sorting number. Sort the names of multiple known public data sources according to the authority and accuracy of public data. The names are sorted. The higher the authority and accuracy of the public data, the smaller the sorting number. According to the data source name, the data source name is vectorized to determine the data source name vector. Multiple known internal department names of the enterprise are vectorized in a vectorized manner to determine the known internal department name vector. Multiple known external partner names are vectorized to determine the known external partner name vector. Multiple known public data source names are vectorized to determine the known public data source name vector. According to the data source name vector, the known internal department name vector, the known external partner name vector, the known public data source name vector and the sorting number, the data source of the data to be managed is evaluated to determine the source classification condition coefficient of the data to be managed.

[0054] According to an embodiment of the present invention, determining the source classification condition coefficient of the data to be managed according to the data source name vector, the known enterprise internal department name vector, the known external partner name vector, the known public data source name vector and the sorting sequence number includes: determining the source classification condition coefficient SC of the data to be managed according to formula (1),

[0055]

[0056] Among them, if is the conditional function, max is the maximum value function, CS T is the preset cosine similarity threshold, (DS) is the data source name vector, (DS) T is the transposed vector of (DS), (WD i) is the name vector of the i-th known internal department of the enterprise, (EP j ) is the jth known external partner name vector, (PS k ) is the kth known public data source name vector, i s To satisfy The sorting order of the known internal department names of the enterprise, j s To satisfy The sorted order of the known external partner name vector, k s To satisfy The sorting order of the known public data source name, n is the number of known internal department name vectors of the enterprise, i≤n, 1≤i s ≤n, m is the number of known external partner name vectors, j≤m, 1≤j s ≤m, K is the number of known public data source name vectors, k≤K, 1≤k s ≤K, i, i s , n, j, j s , m, k, k s and K are both positive integers.

[0057] According to one embodiment of the present invention, is the cosine similarity between the data source name vector and the i-th known internal department name vector. The greater the similarity, the greater the possibility that the data comes from the i-th known internal department of the enterprise. To obtain the maximum value of the cosine similarity between the data source name vector and any known enterprise internal department name vector, the above maximum value processing can be used to determine whether the data source name is the same as the known enterprise internal department name. , it means that the data comes from a known internal department of the enterprise. In formula (1), the conditional function The value of includes the following two cases, when satisfying When the condition is , it means that the data comes from a known internal department of the enterprise, and the value of the conditional function is (1,i s ),i s The sorting sequence number of the known internal departments of the enterprise where the data is sourced. When the condition is met, it means that the data does not come from any known internal department of the enterprise, and the value of the conditional function is (0,0).

[0058] According to one embodiment of the present invention, is the cosine similarity between the data source name vector and the j-th known external partner name vector. The greater the similarity, the greater the possibility that the data comes from the j-th known external partner. To obtain the maximum value of the cosine similarity between the data source name vector and any known external partner name vector, the above maximum value processing can be used to determine whether the data source name is the same as the known external partner name. , it means that the data comes from a known internal department of the enterprise. In formula (1), the conditional function The value of includes the following two cases, when satisfying When the condition is , it means that the data comes from one of the known external partners, and the value of the condition function is (2,j s ), j s The sorting sequence number of the known external partners of the data source. When the condition is met, it means that the data does not come from any known external partner, and the value of the condition function is (0,0).

[0059] According to one embodiment of the present invention, is the cosine similarity between the data source name vector and the k-th known public data source name vector. The greater the similarity, the greater the possibility that the data source name is the same as the k-th known public data source name. To obtain the maximum value of the cosine similarity between the data source name vector and any known public data source name vector, the above maximum value processing can be used to determine whether the data source name is the same as the known public data source name. , it means that the data source is the same as one of the known public data sources. In formula (1), the conditional function The value of includes the following two cases, when satisfying When the condition is , it means that the data comes from one of the known public data, and the value of the condition function is (3, k s ), k s The sorting sequence number of the known public data source of the data source, if it does not meet When the condition is met, it means that the data does not come from any known public data, and the value of the conditional function is (0,0).

[0060] According to one embodiment of the present invention,

[0061]

[0062] It means that the data source of the data to be managed is evaluated based on whether the data source name is the same as the known internal department name of the enterprise, the known external partner name and the known public data source name, and the source classification condition coefficient of the data to be managed is determined.

[0063] In this way, the source classification condition coefficient of the data to be managed can be determined based on the data source name vector, the known internal enterprise department name vector, the known external partner name vector, the known public data source name vector and the sorting sequence number. During the calculation process, the cosine similarity operation can be performed on the data source name vector with the known internal enterprise department name vector, the known external partner name vector and the known public data source name vector respectively to determine the specific source of the data to be managed, thereby improving the accuracy and comprehensiveness of the source classification condition coefficient.

[0064] According to an embodiment of the present invention, in step S103, relevant information of the data to be managed is obtained, wherein the relevant information includes: data volume, number of quantity links, data update frequency, data accuracy and data usage scope.

[0065] For example, through a professional data collection platform, the data volume, number of links, data update frequency, data accuracy and data usage scope of the data to be managed can be obtained.

[0066] According to one embodiment of the present invention, in step S104, the data value prediction model is trained to obtain a trained data value prediction model.

[0067] According to one embodiment of the present invention, step S104 includes:

[0068] Acquire a training data set for model training, wherein the training data set includes a plurality of training data;

[0069] Determine training data related information of the plurality of training data, wherein the training data related information includes: training data volume, number of training data links, training data update frequency, training data accuracy, and training data usage scope;

[0070] Obtain expert training data basic value scores and expert training data circulation rate scores of multiple training data;

[0071] Processing the training data related information through a data value prediction model to obtain a sample training data basic value score and a sample training data circulation rate score;

[0072] Determine a training loss function of a data value prediction model according to the training data related information, the expert training data basic value score, the expert training data circulation rate score, the sample training data circulation rate score and the sample training data basic value score;

[0073] The data value prediction model is trained according to the training loss function of the data value prediction model to obtain a trained data value prediction model.

[0074] For example, obtaining a training data set in a machine learning database to provide a basis for model learning and optimization, wherein the training data set has been preprocessed and labeled; determining the amount of training data for multiple training data (the amount of data can reflect the richness of the information contained), the number of training data links (the more links there are, the greater the value of the data is usually), the frequency of training data updates, the accuracy of training data, and the scope of use of training data; experts in related fields use a combination of qualitative (using the subjective judgment of experts to evaluate the basic value and circulation rate of training data) and quantitative (such as bibliometrics and data envelopment analysis, which are evaluated through quantitative analysis of training data) methods to evaluate the basic value and circulation rate of intelligence information, and obtain expert training data basic value scores and expert training data circulation rates. The circulation rate score is obtained by processing the training data volume, the number of training data links, the training data update frequency, the training data accuracy and the usage scope of the training data through the data value prediction model to obtain the sample training data basic value score and the sample training data circulation rate score; according to the training data related information, the expert training data basic value score, the expert training data circulation rate score, the sample training data circulation rate score and the sample training data basic value score, the error generated in the process of predicting the data value is evaluated to determine the training loss function of the data value prediction model; the data value prediction model is trained according to the training loss function of the data value prediction model to improve the accuracy of the data value prediction model in predicting the data value, and the trained data value prediction model is obtained.

[0075] According to one embodiment of the present invention, the training loss function of the data value prediction model is determined based on the training data related information, the expert training data basic value score, the expert training data circulation rate score, the sample training data circulation rate score and the sample training data basic value score, including: determining the training loss function LOSS of the data value prediction model according to formula (2): T ,

[0076]

[0077] Among them, EFV r Score the expert training data basic value of the rth training data, SFV r Score the basic value of the sample training data for the rth training data, ECR r SCR is the expert training data circulation rate score for the rth training data. r Score the sample training data circulation rate of the rth training data, TDV r is the amount of training data for the rth training data, TDL r is the number of training data links for the rth training data, TDA ris the training data accuracy of the rth training data, TDF r is the training data update frequency of the rth training data, TDR r is the training data usage range of the rth training data, R is the number of training data, r≤R, r and R are both positive integers.

[0078] According to one embodiment of the present invention, It is the ratio of the amount of training data of the rth training data to the average amount of training data. The larger the ratio is, the larger the amount of training data of the rth training data is, the greater the impact on the training process is, and the greater the impact of its change on the error of the basic value of the predicted data is. It is the ratio of the number of training data links of the r-th training data to the average number of training data links. The larger the ratio is, the more training data links the r-th training data has, the greater the impact on the training process, and the greater the impact of its change on the error of the basic value of the predicted data. It is the ratio of the training data accuracy of the rth training data to the average training data accuracy. The larger the ratio is, the higher the training data accuracy of the rth training data is, the greater the impact on the training process is, and the greater the impact of its change on the error of the basic value of the predicted data is. It means that the amount of training data, the number of training data links and the accuracy of training data are positively correlated with the basic value of training data. For example, when the amount of data is larger, the information contained is richer and the basic value of data is higher. When the number of data links is more, the relevance and integrity of the data are stronger and the basic value of data is higher. When the data accuracy is higher, the reliability of the data is higher and the basic value of data is higher. Therefore, the larger the amount of training data, the number of training data links and the accuracy of training data are relative to the average value, that is, and The larger the value of , the greater the impact on the basic value of the training data, and the greater the impact on the error of the predicted data value. It is the relative difference between the basic value score of the expert training data of the r-th training data and the basic value score of the sample training data of the r-th training data. The larger the ratio is, the greater the difference between the basic value score of the expert training data of the r-th training data and the basic value score of the sample training data of the r-th training data, and the greater the relative error of the predicted data value.

[0079] According to one embodiment of the present invention, It is the ratio of the training data update frequency of the rth training data to the average training data update frequency. The larger the ratio is, the higher the training data update frequency of the rth training data is, the greater the impact on the training process is, and the greater the impact of its change on the error of the predicted data circulation rate is. It is the ratio of the training data usage range of the rth training data to the average training data usage range. The larger the ratio is, the larger the training data usage range of the rth training data is, the greater the impact on the training process is, and the greater the impact of its change on the error of the predicted data circulation rate is. It indicates that the training data update frequency and the training data usage scope are positively correlated with the training data circulation rate. For example, when the data update frequency is faster, the data can be quickly disseminated and the data circulation rate can be improved. When the data usage scope is larger, the possibility of data circulation can be increased. Therefore, the larger the training data update frequency and the training data usage scope are relative to the average value, that is, and The larger the value of , the greater the impact on the circulation rate of training data, and the greater the impact on the error of the circulation rate of predicted data. It is the relative difference between the circulation rate score of the expert training data of the r-th training data and the circulation rate score of the sample training data of the r-th training data. The larger the ratio is, the greater the difference between the circulation rate score of the expert training data of the r-th training data and the circulation rate score of the sample training data of the r-th training data is, and the greater the relative error of the predicted data value is.

[0080] According to one embodiment of the present invention, using The relative errors of the basic value scores of the sample training data of each training data are averaged, and the The relative errors of the circulation rate scores of the sample training data of each training data are averaged and summed to obtain the training loss function. During the training process, the above training loss function is reduced, so that the error between the basic value score of the sample training data and the basic value score of the expert training data is reduced, and the error between the circulation rate score of the sample training data and the circulation rate score of the expert training data is reduced, so that the accuracy of the data value prediction model for the prediction of data circulation rate and data basic value is improved, thereby improving the accuracy of the data value prediction model.

[0081] In this way, the training loss function of the data value prediction model can be determined based on the training data related information, the expert training data basic value score, the expert training data circulation rate score, the sample training data circulation rate score and the sample training data basic value score. In the calculation process, the influence of the three aspects of training data volume, training data link number and training data accuracy on the error of data basic value, as well as the influence of training data update frequency and training data usage scope on the error of training data circulation rate are fully considered to set the training loss function, so that the load forecasting model can reduce the training loss function during the training process, and improve the accuracy of the data value prediction model in a more targeted manner.

[0082] According to an embodiment of the present invention, in step S105, the relevant information is processed by a trained data value prediction model to obtain a data basic value score and a data circulation rate score for the data to be managed.

[0083] For example, the data volume, number of data links, data update frequency, data accuracy and data usage scope of the data to be managed are input into the trained data value prediction model to determine the data basic value score and data circulation rate score of the data to be managed.

[0084] According to an embodiment of the present invention, in step S106, the value classification condition coefficient is determined based on the data basic value score and the data circulation rate score.

[0085] According to one embodiment of the present invention, step S106 includes:

[0086] Determine the data value score by taking a weighted sum based on the data basic value score and the data circulation rate score;

[0087] If the data value score is greater than or equal to the first set data value score threshold, the value classification condition coefficient is 1;

[0088] If the data value score is less than the first set data value score threshold, and the data value score is greater than the second set data value score threshold, the value classification condition coefficient is 2;

[0089] If the data value score is less than or equal to the second set data value score threshold, the value classification condition coefficient is 3.

[0090] For example, based on the different impacts of data basic value and data circulation rate on data value assessment, the data basic value score and the data circulation rate score are weighted and summed to determine the data value score; if the data value score is greater than or equal to the first set data value score threshold, it means that the data basic value and data circulation rate of the data are relatively high, and the value classification condition coefficient of the data is determined to be 1, and the data is high-value data; if the data value score is less than the first set data value score threshold, and the data value score is greater than the second set data value score threshold, it means that the data basic value of the data is not obvious and the data circulation rate is not high, and the value classification condition coefficient of the data is determined to be 2, and the data is medium-value data; if the data value score is less than or equal to the second set data value score threshold, it means that the data basic value and data circulation rate of the data are relatively low, and the value classification condition coefficient of the data is determined to be 3, and the data is low-value data.

[0091] According to an embodiment of the present invention, in step S107, the data to be managed is classified and managed for the first time according to the source classification condition coefficient.

[0092] According to one embodiment of the present invention, step S107 includes:

[0093] If the source classification condition coefficient of the data to be managed is (1,i s ), the data to be managed is divided into i in the internal data category s The corresponding known categories of internal departments of the enterprise;

[0094] If the source classification condition coefficient of the data to be managed is (2,j s ), the data to be managed is divided into j in the external data category s The corresponding categories of known external partners;

[0095] If the source classification condition coefficient of the data to be managed is (3, k s ), the data to be managed is divided into k categories of public data s The category of the corresponding public data source.

[0096] For example, when the data to be managed is classified for the first time, the data to be managed is classified according to the source classification condition coefficient. If the source classification condition coefficient of the data to be managed is (1, i s ), indicating that the data source name is the same as one of the known internal departments of the enterprise, and the data to be managed is classified into i in the internal data category. s The corresponding known internal department category of the enterprise; if the source classification condition coefficient of the data to be managed is (2,j s ), indicating that the data source name is the same as one of the known external partners, and the data to be managed is classified into j in the external data category. s The corresponding known external partner category; if the source classification condition coefficient of the data to be managed is (3,k s ), indicating that the data source name is the same as one of the known public data source names, and the data to be managed is divided into k public data categories. s The category of the corresponding public data source.

[0097] According to an embodiment of the present invention, in step S108, after the first classification management is completed, the data to be managed is continuously subjected to a second classification management according to the value classification condition coefficient.

[0098] According to one embodiment of the present invention, step S108 includes:

[0099] If the value classification condition coefficient of the data to be managed is 1, the data to be managed is classified into the high-value data category;

[0100] If the value classification condition coefficient of the data to be managed is 2, the data to be managed is classified into the medium value data category;

[0101] If the value classification condition coefficient of the data to be managed is 3, the data to be managed is classified into the low-value data category.

[0102] For example, when the data to be managed is classified for the second time, it is classified according to the value classification condition coefficient. If the value classification condition coefficient of the data to be managed is 1, it means that the data can directly bring significant economic benefits in economic terms and has high business value, and the data to be managed is classified into the high-value data category; if the value classification condition coefficient of the data to be managed is 2, it means that the data can bring certain economic benefits and have a certain supporting role in the enterprise business, and the data to be managed is classified into the medium-value data category; if the value classification condition coefficient of the data to be managed is 3, it means that the economic benefits brought by the data are very limited and have almost no business value, and the data to be managed is classified into the low-value data category.

[0103] According to the digital economic data classification management method of the embodiment of the present invention, the data source of the data to be managed can be judged according to the data source name of the data to be managed, and the data basic value score and data circulation rate score of the data to be managed can be obtained through the trained data value prediction model. Further, the data value of the data to be managed can be evaluated according to the data basic value score and the data circulation rate score. Further, the data to be managed can be classified and managed according to the two aspects of the data source and data value of the data to be managed, thereby improving the accuracy and comprehensiveness of the digital economic data classification management. The source classification condition coefficient of the data to be managed can be determined according to the data source name vector, the known enterprise internal department name vector, the known external partner name vector, the known public data source name vector and the sorting sequence number. In the calculation process, the data source name vector can be respectively subjected to cosine similarity operations with the known enterprise internal department name vector, the known external partner name vector and the known public data source name vector to determine the specific source of the data to be managed, thereby improving the accuracy and comprehensiveness of the source classification condition coefficient. The training loss function of the data value prediction model can be determined based on the training data related information, the expert training data basic value score, the expert training data circulation rate score, the sample training data circulation rate score and the sample training data basic value score. In the calculation process, the influence of the three aspects of training data volume, training data link number and training data accuracy on the error of data basic value, as well as the influence of training data update frequency and training data usage scope on the error of training data circulation rate are fully considered to set the training loss function, so that the training loss function of the load forecasting model can be reduced during the training process, and the accuracy of the data value prediction model can be improved in a more targeted manner.

[0104] Figure 2 A block diagram of a digital economic data classification management system according to an embodiment of the present invention is exemplarily shown, and the system includes:

[0105] Data source name module, obtains the data source name of the data to be managed;

[0106] A source classification condition coefficient module determines the source classification condition coefficient of the data to be managed according to the data source name;

[0107] The data information module obtains relevant information of the data to be managed, wherein the relevant information includes: data volume, number of quantity links, data update frequency, data accuracy and data usage scope;

[0108] The prediction model module trains the data value prediction model to obtain the trained data value prediction model;

[0109] The data scoring module processes the relevant information through the trained data value prediction model to obtain the data basic value score and data circulation rate score of the data to be managed;

[0110] A value classification condition coefficient module determines a value classification condition coefficient according to the data basic value score and the data circulation rate score;

[0111] A first classification management module performs first classification management on the data to be managed according to the source classification condition coefficient;

[0112] The second classification management module, after completing the first classification management, continues to perform a second classification management on the management data according to the value classification condition coefficient.

[0113] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0114] It should be understood by those skilled in the art that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments, and the embodiments of the present invention may be deformed or modified in any way without departing from the principles.

Claims

1. A digital economic data classification management method, characterized in that: include: Get the data source name of the data to be managed; According to the data source name, determine the source classification condition coefficient of the data to be managed; Obtain relevant information of the data to be managed, wherein the relevant information includes: data volume, number of quantity links, data update frequency, data accuracy and data usage scope; Training the data value prediction model to obtain a trained data value prediction model; The relevant information is processed through the trained data value prediction model to obtain the data basic value score and data circulation rate score of the data to be managed; Determining a value classification condition coefficient according to the data basic value score and the data circulation rate score; According to the source classification condition coefficient, the data to be managed is classified and managed for the first time; After the first classification management is completed, the data to be managed is continued to be classified for the second time according to the value classification condition coefficient; According to the data source name, the source classification condition coefficient of the data to be managed is determined, including: Obtain multiple known enterprise internal department names, known external partner names, and known public data source names; Sort multiple known enterprise internal department names, known external partner names, and known public data source names respectively, and determine the sorting sequence numbers of the multiple known enterprise internal department names, known external partner names, and known public data source names; Determine a data source name vector according to the data source name; Determine a known enterprise internal department name vector according to the known enterprise internal department name; Determine a known external partner name vector according to the known external partner name; Determine a known public data source name vector according to the known public data source name; Determine the source classification condition coefficient of the data to be managed according to the data source name vector, the known enterprise internal department name vector, the known external partner name vector, the known public data source name vector and the sorting sequence number; Determining the source classification condition coefficient of the data to be managed according to the data source name vector, the known enterprise internal department name vector, the known external partner name vector, the known public data source name vector and the sorting sequence number, including: According to the formula , Determine the source classification condition coefficient of the data to be managed , where if is the conditional function and max is the maximum value function. is the preset cosine similarity threshold, is the data source name vector, for The transposed vector of is the name vector of the i-th known internal department of the enterprise, is the jth known external partner name vector, is the kth known public data source name vector, To satisfy The sorting order of the known internal department names of the enterprise, To satisfy The sorted order of the vector of known external partner names, To satisfy The sorting sequence number of the known public data source name, n is the number of known enterprise internal department name vectors, i≤n, 1≤ ≤n, is the number of known external partner name vectors, j≤m, 1≤ ≤m, K is the number of known public data source name vectors, k≤K, 1≤ ≤K,i, , n, j, , m, k, and K are both positive integers; According to the data basic value score and the data circulation rate score, the value classification condition coefficient is determined, including: Determine the data value score by taking a weighted sum based on the data basic value score and the data circulation rate score; If the data value score is greater than or equal to the first set data value score threshold, the value classification condition coefficient is 1; If the data value score is less than the first set data value score threshold, and the data value score is greater than the second set data value score threshold, the value classification condition coefficient is 2; If the data value score is less than or equal to the second set data value score threshold, the value classification condition coefficient is 3.

2. The digital economic data classification management method according to claim 1 is characterized in that: The data value prediction model is trained to obtain a trained data value prediction model, including: Acquire a training data set for model training, wherein the training data set includes a plurality of training data; Determine training data related information of the plurality of training data, wherein the training data related information includes: training data volume, number of training data links, training data update frequency, training data accuracy, and training data usage scope; Obtain expert training data basic value scores and expert training data circulation rate scores of multiple training data; Processing the training data related information through a data value prediction model to obtain a sample training data basic value score and a sample training data circulation rate score; Determine a training loss function of a data value prediction model according to the training data related information, the expert training data basic value score, the expert training data circulation rate score, the sample training data circulation rate score and the sample training data basic value score; The data value prediction model is trained according to the training loss function of the data value prediction model to obtain a trained data value prediction model.

3. The digital economic data classification management method according to claim 2 is characterized in that: Determining a training loss function of a data value prediction model according to the training data related information, the expert training data basic value score, the expert training data circulation rate score, the sample training data circulation rate score and the sample training data basic value score includes: According to the formula , Determine the training loss function of the data value prediction model ,in, Score the basic value of the expert training data for the rth training data, Score the basic value of the sample training data for the rth training data, Score the expert training data circulation rate of the r-th training data, Score the sample training data circulation rate of the rth training data, is the amount of training data for the rth training data, is the number of training data links for the rth training data, is the training data accuracy of the rth training data, is the training data update frequency of the rth training data, is the training data usage range of the rth training data, R is the number of training data, r≤R, r and R are both positive integers.

4. The digital economic data classification management method according to claim 1 is characterized in that: According to the source classification condition coefficient, the data to be managed is classified and managed for the first time, including: If the source classification condition coefficient of the data to be managed is , the data to be managed is divided into the internal data categories The corresponding known categories of internal departments of the enterprise; If the source classification condition coefficient of the data to be managed is , the data to be managed is divided into the external data category The corresponding categories of known external partners; If the source classification condition coefficient of the data to be managed is , the data to be managed is divided into the public data category The category of the corresponding public data source.

5. The digital economic data classification management method according to claim 1 is characterized in that: After the first classification management is completed, the data to be managed is further classified and managed for the second time according to the value classification condition coefficient, including: If the value classification condition coefficient of the data to be managed is 1, the data to be managed is classified into the high-value data category; If the value classification condition coefficient of the data to be managed is 2, the data to be managed is classified into the medium value data category; If the value classification condition coefficient of the data to be managed is 3, the data to be managed is classified into the low-value data category.

6. A digital economic data classification management system for executing the method according to any one of claims 1 to 5, characterized in that: include: Data source name module, obtains the data source name of the data to be managed; A source classification condition coefficient module determines the source classification condition coefficient of the data to be managed according to the data source name; The data information module obtains relevant information of the data to be managed, wherein the relevant information includes: data volume, number of quantity links, data update frequency, data accuracy and data usage scope; The prediction model module trains the data value prediction model to obtain the trained data value prediction model; The data scoring module processes the relevant information through the trained data value prediction model to obtain the data basic value score and data circulation rate score of the data to be managed; A value classification condition coefficient module determines a value classification condition coefficient according to the data basic value score and the data circulation rate score; A first classification management module performs first classification management on the data to be managed according to the source classification condition coefficient; The second classification management module, after completing the first classification management, continues to perform a second classification management on the management data according to the value classification condition coefficient.

Images