Information management method and device based on smart business

Abstract

The invention relates to the technical field of data management, and discloses an information management method and device based on smart business. By means of the method, multi-source heterogeneous data compatibility can be effectively achieved, original data are obtained through a preset standardization protocol, a standardized data set is formed through a format conversion algorithm, the problems that data sources are diversified and formats are different are solved, and a foundation is laid for follow-up processing. And the clustering algorithm performs grouping processing on the characteristic attribute data, so that the heterogeneous threshold requirement is met, the data is further structured, and the data has higher availability.

Description

Technical Field

[0001] The present invention relates to the field of data management technology, and in particular to an information management method and device based on smart business. Background Art

[0002] As the core pillar of smart business development, information management is directly related to the decision-making efficiency and competitive advantage of enterprises in a complex market environment. Its importance is self-evident. With the increasing diversification of business scenarios, the integration and intelligent processing of multi-source data has become the key to promoting business innovation.

[0003] In the existing technology, information or data streams are usually processed by processing single data separately. However, since the data collection layer cannot unify the standards, compatibility issues between heterogeneous data frequently occur; the data processing layer lacks the ability to model dynamic relationships, making it difficult to capture deep information relevance; and the data analysis layer and the display layer are not well connected, making it difficult to quickly transform analysis conclusions into actionable business insights.

[0004] In summary, the existing technology is separated in the data processing and analysis links, resulting in low efficiency of information flow and failure to meet the requirements of real-time and accuracy. Summary of the invention

[0005] The present invention provides an information management method and device based on smart business to achieve compatibility, data modeling and dynamic presentation of multi-source heterogeneous data.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides an information management method based on smart business, comprising: Obtain multi-source raw data through a preset standardized protocol, map it into a unified structure through a format conversion algorithm, and obtain a standardized data set; Extracting characteristic attribute data according to the normalized data set, grouping the characteristic attribute data using a clustering algorithm, and taking the grouping results that meet the heterogeneity threshold as the classified data set; Acquire time series features according to the classified data set, and perform dynamic change sampling through a sliding window technology according to the difficulty of collecting the time series features to obtain dynamic data samples; Construct a relationship network based on the dynamic data samples and model and mine implicit associations, optimize the information flow path and transmission based on the results, and obtain target transmission data; Extract key indicators according to the target transmission data, obtain the predicted value of short-term resource demand and the real-time load of the current system resource usage status, and fuse the predicted value and the real-time load through a weighted average algorithm to obtain an adjusted resource demand assessment value; Determine the weight allocation ratio according to the adjusted resource requirement evaluation value, and perform error prediction on the weight allocation ratio through a regression algorithm to obtain analysis result data; Generate a multi-dimensional feature vector according to the analysis result data, perform dimensionality reduction processing on the data analysis connection requirements through a principal component analysis algorithm, and optimize feature extraction to adjust the display parameter range to obtain associated data; Construct a visualization model based on the dimensionality reduction analysis data, convert the multi-dimensional feature vector into graphic elements through data mapping, and adjust the display parameters of the graphic elements according to the intuitive presentation ability requirements to obtain visualization presentation parameters; Obtain user feedback according to the visualization presentation parameters, optimize the system-level collaboration requirements through an iterative update algorithm to obtain feedback data, and compare the feedback data with preset improvement conditions, and adjust the system level to obtain optimized system data.

[0007] In an alternative embodiment, extract feature attribute data according to the normalized data set, group the feature attribute data through a clustering algorithm, and use the grouping result that meets the heterogeneity threshold as a classification data set, including: Obtain structural features according to the classification data set, and perform category distribution processing through statistical tools to obtain a distribution data set; Extract keyword fields according to the distribution data set. If the keyword fields meet the consistency check, determine the consistency data set; if not, adjust the field content through a field mapping tool to obtain an adjusted data set; Adjust according to the category distribution of the data set, and identify outliers through a clustering algorithm to obtain an outlier marked data set; Correct outliers through a mean filling tool according to the marking information of the outlier marked data set to obtain a corrected data set; Extract associated fields according to the corrected data set, and construct a query structure through an indexing tool to obtain a query data set; According to the complete records of the query data set, adjust the order according to the category distribution through a sorting tool to obtain a sorted data set; Perform integrity verification on the sorted data set through a consistency check tool according to the structural features of the sorted data set to obtain a verified data set.

[0008] In an alternative embodiment, construct a relationship network based on the dynamic data sample and model to mine implicit associations, and optimize the information flow path and transmission according to the results to obtain target transmission data, including: Construct a relationship network based on the dynamic data samples, perform multi-layer data relationship modeling through the graph neural network algorithm, and mine the implicit associations between nodes with deep data processing requirements to obtain relationship-enhanced data; Obtain the information flow path according to the relationship-enhanced data, optimize the data transmission order through the shortest path algorithm, and adjust the path weights according to the information flow efficiency requirements to obtain the target transmission data; In an alternative embodiment, extract key indicators from the target transmission data, obtain the predicted value of the short-term resource requirements and the real-time load of the current system resource usage status, and fuse the predicted value and the real-time load through the weighted average algorithm to obtain the adjusted resource requirement evaluation value, including: Obtain key indicators from the target transmission data, combine the predicted value of the short-term resource requirements, and perform real-time load fusion through the weighted average method to obtain the preliminary resource requirement data; Judge whether there is abnormal fluctuation according to the sudden increase trend of the preliminary resource requirement data to obtain the fluctuation detection data; Adjust the resource allocation order through the sorting tool according to the downward trend in the fluctuation detection data to obtain the rearranged resource data; Obtain the balanced load data by obtaining the load balancing distribution according to the system state of the rearranged resource data; Obtain the data fusion characteristics according to the balanced load data. If the sudden increase trend exceeds the preset threshold, adjust the resource requirements to obtain the corrected requirement data; Judge whether the resource requirements meet the system state according to the estimated value of the corrected requirement data to obtain the verified requirement data; Adjust the resource allocation through the mean filling tool according to the key indicators of the verified requirement data to obtain the resource requirement evaluation value.

[0009] In an alternative embodiment, determine the weight allocation ratio according to the adjusted resource requirement evaluation value, and perform error prediction on the weight allocation ratio through the regression algorithm to obtain the analysis result data, including: determine the weight allocation ratio according to the adjusted resource requirement evaluation value, and perform prediction on the real-time requirement and the accuracy requirement through the regression analysis algorithm to obtain the preliminary prediction data; If the prediction error in the preliminary prediction data is lower than the preset threshold, determine the analysis result data. If it is higher than the preset threshold, adjust the model parameters for re-prediction to obtain the corrected prediction data; Obtain the trend analysis data by analyzing the change trend of the allocation ratio according to the real-time requirement of the corrected prediction data; According to the accuracy requirements of the trend analysis data, the weight distribution is adjusted through a mean filling tool to obtain balanced distribution data; According to the analysis results of the balanced distribution data, by detecting the fluctuation of resource requirements, fluctuation detection data is obtained; According to the change trend of the fluctuation detection data, by adjusting the estimated value of resource requirements, the final demand data is obtained; Key indicators are extracted according to the final demand data, and by judging whether the real-time requirements meet the system state, verification result data is obtained as the analysis result data.

[0010] In an alternative embodiment, a multi-dimensional feature vector is generated according to the analysis result data, the dimensionality reduction processing is performed on the data analysis connection requirements through a principal component analysis algorithm, and the display parameter range is adjusted by optimizing feature extraction to obtain associated data, including: A multi-dimensional feature vector is generated according to the analysis result data, and the dimensionality reduction processing is performed on the data analysis connection requirements through a principal component analysis algorithm to obtain dimensionality reduction analysis data; According to the optimized feature extraction process of the dimensionality reduction analysis data, the display parameter range is adjusted through the data change trend determination threshold to obtain associated data of resource fluctuation and visualization structure, including: According to the dimensionality reduction analysis data, the feature extraction process is adjusted, and by judging the sudden increase trend of a preset threshold, trend determination data is obtained; According to the downward trend of the trend determination data, by adjusting the display parameter range, parameter optimization data is obtained; According to the resource fluctuation of the parameter optimization data, by detecting the distribution of the visualization structure, distribution detection data is obtained; According to the associated data of the distribution detection data, by extracting the core dimension of data optimization, dimension extraction data is obtained; If the dimension extraction data meets the determination threshold, then the trend determination is adjusted through feature extraction to obtain adjusted analysis data; According to the visualization structure of the adjusted analysis data, by optimizing the allocation ratio of resource fluctuation, ratio optimization data is obtained; According to the parameter range of the ratio optimization data, by judging the change of the sudden increase trend, change detection data is obtained.

[0011] In an alternative embodiment, a visualization model is constructed according to the dimensionality reduction analysis data, the multi-dimensional feature vector is transformed into graphic elements through data mapping, and the display parameters of the graphic elements are adjusted according to the presentation ability requirements to obtain visualization presentation parameters, including: Obtain feature vectors by parsing according to the dimensionality reduction analysis data; Perform vector conversion on the basis of the vector features through data mapping technology to obtain graphic elements; Adjust the display parameters according to the presentation ability to determine the display form of the graphic elements; Construct a visualization model according to the mapping technology to obtain presentation data; If the presentation data meets the preset threshold, optimize the distribution of the graphic elements through parameter adjustment to obtain the adjusted presentation data; Judge the improvement of the presentation ability by detecting the change trend of the feature vectors according to the adjusted presentation data; Update the visualization model according to the change trend to obtain optimized presentation data.

[0012] In an alternative implementation, obtain user feedback according to the visualization presentation parameters, optimize the system-level collaboration requirements through an iterative update algorithm to obtain feedback data, and compare the feedback data with preset improvement conditions, and adjust the system level to obtain optimized system data, including: Obtain user feedback according to user interaction data, and obtain the distribution characteristics of the feedback content through a recording tool; Judge the feedback trigger situation according to the distribution characteristics. If the preset improvement condition is triggered, adjust the system-level parameters through an iterative algorithm to obtain the adjusted collaboration requirement data; Obtain the update direction of the processing parameters by parsing the process according to the adjusted collaboration requirement data; Adjust the processing parameters according to the update direction to obtain optimized system-level data; Judge whether the feedback trigger decreases by detecting the change trend of the interaction data according to the optimized system-level data; If the change trend shows a decrease in the trigger, update the presentation form through the parsing process to obtain optimized data; Detect the matching degree of the collaboration requirements through a verification tool according to the optimized data to determine the optimized system data output by the final system.

[0013] In a second aspect, the present invention further provides an information management device based on intelligent commerce, including: A data acquisition module, configured to acquire multi-source raw data through a preset standard protocol, map it into a unified structure through a format conversion algorithm to obtain a normalized data set; A data grouping module, configured to extract feature attribute data according to the normalized data set, perform grouping processing on the feature attribute data through a clustering algorithm, and use the grouping result that meets the heterogeneity threshold as a classification data set; A dynamic analysis module, configured to obtain time series features according to the classification data set, and perform dynamic change sampling through a sliding window technique according to the acquisition difficulty of the time series features to obtain dynamic data samples; A modeling and optimization module, configured to construct a relationship network and perform modeling to mine implicit associations according to the dynamic data samples, and optimize the information flow path and transmission according to the results to obtain target transmission data; A resource evaluation module, configured to extract key indicators according to the target transmission data, obtain a predicted value of short-term resource requirements and a real-time load of the current system resource usage status, and fuse the predicted value and the real-time load through a weighted average algorithm to obtain an adjusted resource requirement evaluation value; An evaluation and prediction module, configured to determine a weight allocation ratio according to the adjusted resource requirement evaluation value, and perform error prediction on the weight allocation ratio through a regression algorithm to obtain analysis result data; A vector association module, configured to generate a multi-dimensional feature vector according to the analysis result data, perform dimensionality reduction processing on the data analysis connection requirements through a principal component analysis algorithm, and optimize the feature extraction to adjust the display parameter range to obtain associated data; A parameter modeling module, configured to construct a visualization model according to the dimensionality reduction analysis data, convert the multi-dimensional feature vector into graphic elements through data mapping, and adjust the display parameters of the graphic elements according to the intuitive presentation ability requirements to obtain visualization presentation parameters; A feedback and optimization module, configured to obtain user feedback according to the visualization presentation parameters, optimize the system-level collaboration requirements through an iterative update algorithm to obtain feedback data, compare the feedback data with preset improvement conditions, and adjust the system level to obtain optimized system data.

[0014] In a third aspect, the present invention further provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the information management method based on intelligent commerce described in any one of the above is implemented.

[0015] In a fourth aspect, the present invention further provides a computer-readable storage medium, where the computer-readable storage medium includes a stored computer program. When the computer program runs, the device where the computer-readable storage medium is located is controlled to execute the information management method based on intelligent commerce described in any one of the above.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) The method of the present invention obtains multi-source raw data through a preset standardized protocol, maps it into a unified structure through a format conversion algorithm, and uses a clustering algorithm to group the feature attribute data. While improving the consistency and efficiency of data processing, it can screen out the grouping results that meet the heterogeneity threshold as the classification data set, which helps to effectively classify the data, facilitates subsequent analysis, and improves the pertinence of data processing.

[0017] (2) The method of the present invention performs dynamic data sampling through a sliding window technique, which can adapt to the dynamic variability of data. At the same time, by modeling the dynamic data, it realizes the mining of implicit associations, optimizes the information flow path and transmission, and obtains the target transmission data, which can improve the information flow efficiency, make the data transmission more accurate, and meet the accuracy requirements.

[0018] (3) The method of the present invention displays data according to the intuitive presentation requirements through a visualization model and parameter adjustment. Finally, it optimizes the system in combination with user feedback, and adjusts the system level through iterative updates to make the system more in line with actual business needs and improve the efficiency and quality of intelligent business operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a schematic flowchart of an information management method based on intelligent commerce provided by the first embodiment of the present invention; Figure 2 is a structural diagram of an information management device based on intelligent commerce provided by the first embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0021] Refer to Figure 1 , the first embodiment of the present invention provides an information management method based on intelligent commerce, including the following steps: S101, obtaining multi-source raw data through a preset standardized protocol, mapping it into a unified structure through a format conversion algorithm, and obtaining a normalized data set; S102, extracting feature attribute data according to the normalized data set, grouping the feature attribute data through a clustering algorithm, and using the grouping results that meet the heterogeneity threshold as the classification data set; S103. Obtain time series features according to the classified data set, and perform dynamic change sampling through the sliding window technique according to the acquisition difficulty of the time series features to obtain dynamic data samples; S104. Construct a relationship network based on the dynamic data samples and model to mine implicit associations, and optimize the information flow path and transmission according to the results to obtain target transmission data; S105. Extract key indicators according to the target transmission data, obtain the predicted value of short-term resource requirements and the real-time load of the current system resource usage status, and fuse the predicted value and the real-time load through the weighted average algorithm to obtain an adjusted resource requirement evaluation value; S106. Determine the weight allocation ratio according to the adjusted resource requirement evaluation value, and perform error prediction on the weight allocation ratio through the regression algorithm to obtain analysis result data; S107. Generate a multi-dimensional feature vector according to the analysis result data, perform dimensionality reduction processing on the data analysis connection requirements through the principal component analysis algorithm, and optimize the feature extraction to adjust the display parameter range to obtain associated data; S108. Construct a visualization model according to the dimensionality reduction analysis data, convert the multi-dimensional feature vector into graphic elements through data mapping, and adjust the display parameters of the graphic elements according to the intuitive presentation ability requirements to obtain visualization presentation parameters; S109. Obtain user feedback according to the visualization presentation parameters, optimize the system-level collaboration requirements through the iterative update algorithm to obtain feedback data, compare the feedback data with the preset improvement conditions, and adjust the system level to obtain optimized system data.

[0022] In step S101, obtain the original data of multi-source data through a preset standardization protocol, and map the original data from different sources to a unified structure through a format conversion algorithm to obtain a normalized data set; It should be noted that the original data set is extracted from multi-source data through a preset protocol, and the data content is obtained by using a standard interface to obtain the initial data set. For the heterogeneous data in the initial data set, a format conversion algorithm is applied for processing, mapping the data to a preset unified structure to obtain the first processed data set. If the first processed data set still contains unaligned fields, the data format is adjusted through a field matching tool to obtain the second processed data set. According to the structural characteristics of the second processed data set, a clustering algorithm is used to group the data to obtain a grouped data set. Through the category distribution in the grouped data set, it is judged whether there are abnormal data points. If there are abnormal data points, a mean filling method is used for correction to obtain an optimized data set. The key fields in the optimized data set are obtained, and consistency verification is performed on the key fields to obtain a normalized data set. The complete records in the normalized data set are used to construct a query structure through a data indexing tool to obtain the final data set.

[0023] Specifically, extracting the original data set from multi-source data through a preset protocol is actually pulling the data scattered in different systems together.

[0024] For example, assume that an enterprise's data comes from a sales system, an inventory system, and a customer management system. The sales records, inventory quantities, and customer information can be obtained through an API interface or a database query language according to the agreed field standards to form an initial data set. The initial data set may include sales data in JSON format, an inventory table in CSV format, and customer profiles in XML format. These data have different structures, and the field names and types are not unified. For the heterogeneous data in the initial data set, a format conversion algorithm is applied for processing, aiming to map different formats to a unified structure.

[0025] For example, a standard template can be designed, including fields such as "time", "product ID", "quantity", "source", etc. The "date" field in the sales data is converted to "time", and the "item_code" in the inventory data is converted to "product ID". Through a field mapping table and conversion rules, the first processed data set is generated. There may still be problems with the first processed data set. For example, the "phone" field in the customer data is missing in the sales data, and unaligned fields still exist. If the first processed data set contains unaligned fields, the format is adjusted through a field matching tool.

[0026] For example, fuzzy matching technology can be used to identify "phone" and "tel" as the same field, fill in the missing values or remove redundant items to form the second processed data set. The advantage of this method is to improve data consistency and lay a foundation for subsequent analysis.

[0027] Specifically, all records can be scanned through a matching tool to find cases where the field semantics are similar but the names are different, and after unified adjustment, redundancy can be reduced. According to the structural characteristics of the second processed data set, a clustering algorithm is used to group the data.

[0028] For example, based on the "quantity" and "time" fields, the K-means clustering is used to divide the sales records into two groups: high-frequency transactions and low-frequency transactions, obtaining a grouped data set. After grouping, the data distribution pattern can be clearly seen, which helps to discover potential patterns.

[0029] Preferably, if the data volume is large, dimensionality reduction can be first performed through principal component analysis and then clustering can be carried out to improve efficiency. Through the category distribution of the grouped data set, abnormal data points are judged.

[0030] For example, in the high-frequency transaction group, a record shows that the "quantity" is 10,000, while the average value of the same category is only 50, which can be regarded as abnormal. The mean filling method is adopted to correct it to the within-group mean value of 50, forming an optimized data set. This way can smooth out abnormal fluctuations and improve data reliability.

[0031] It can be understood that if there are too many abnormal points, they can also be eliminated rather than filled in combination with business rules. The key fields in the optimized data set, such as "product ID", "time", and "quantity", are obtained, and consistency verification is performed.

[0032] For example, check whether the "time" conforms to the date format and whether the "quantity" is a positive integer. The records that do not conform can be marked or corrected to obtain a normalized data set.

[0033] In one embodiment, the field format can be verified through regular expressions to ensure data integrity. The complete records in the normalized data set are used to construct a query structure.

[0034] For example, using the database indexing technology, a composite index is created according to the "product ID" and "time" to generate the final data set. The advantage of this structure is that the query efficiency is significantly improved. Especially in the big data scenario, it can quickly respond to business requirements.

[0035] In one possible implementation, a distributed index can also be introduced to further optimize the performance.

[0036] Exemplarily, the whole process from multi-source extraction to the final set can gradually solve the problems of heterogeneity, abnormality, and consistency. While the final data set has a unified format, it also has high query capabilities. The method of the present invention can significantly improve the decision support efficiency in enterprise data integration and reduce the data processing cost at the same time.

[0037] In step S102, characteristic attribute data is extracted according to the normalized data set, and the characteristic attribute data is grouped by a clustering algorithm, and the grouping result that meets the heterogeneity threshold is used as a classification data set, including: S1021, obtaining structural features according to the classification data set, and performing category distribution processing through a statistical tool to obtain a distribution data set; S1022, extracting key fields according to the distribution data set. If the key fields meet the consistency check, a consistency data set is determined; if not, the field content is adjusted through a field mapping tool to obtain an adjusted data set; S1023, adjusting according to the category distribution of the data set, and identifying outliers through a clustering algorithm to obtain an outlier marked data set; S1024, correcting outliers through a mean filling tool according to the marking information of the outlier marked data set to obtain a corrected data set; S1025, extracting associated fields according to the corrected data set, and constructing a query structure through an indexing tool to obtain a query data set; S1026, adjusting the order according to the category distribution through a sorting tool according to the complete records of the query data set to obtain a sorted data set; S1027, performing integrity verification through a consistency check tool according to the structural features of the sorted data set to obtain a verified data set.

[0038] Specifically, when obtaining structural features through the classification data set, the category distribution can be analyzed with the help of a statistical tool.

[0039] For example, in a sales data set of an enterprise, fields such as "product category", "sales region", and "transaction volume" are included. The statistical tool can calculate the proportion of the transaction volume of each product category to form a distribution data set.

[0040] Exemplarily, if the data set shows that the transaction volume of a certain product in a certain region accounts for 60%, its distribution characteristics can be initially judged. This method is convenient for quickly grasping the overall law of the data. When extracting key fields from the distribution data set, assume that "transaction volume" and "sales region" are selected as key fields.

[0041] In a possible implementation manner, it is necessary to check whether these fields are consistent, such as whether the "transaction volume" is all positive and there are no null values. If the consistency check is met, a consistency data set is directly formed; if it is found that the "transaction volume" of some records is negative, adjustment is required.

[0042] Specifically, a field mapping tool can be used to map negative value records to zero or infer reasonable values based on business rules, and finally an adjusted data set is obtained. For the category distribution of the adjusted data set, clustering algorithms can be used to identify outliers.

[0043] For example, based on the "transaction volume" and "region" fields, the DBSCAN algorithm is used to group the data. If the transaction volume of a record far exceeds the average value of the same region, it is marked as an outlier, forming an outlier-marked data set.

[0044] It should be noted that this kind of marking depends on the setting of clustering parameters, and adjusting the parameters can change the sensitivity of outlier identification. Based on the outlier-marked data set, the outlier points are corrected through a mean filling tool.

[0045] In one embodiment, if the "transaction volume" of an outlier point is 5000 and the average value of the same group is 200, it is corrected to 200 to obtain a corrected data set.

[0046] It can be understood that this correction method can smooth data fluctuations and facilitate the stability of subsequent analysis. Associated fields such as "product category" and "transaction volume" are extracted from the corrected data set, and an indexing tool is used to construct a query structure.

[0047] For example, an index is created according to the "product category" to generate a query data set. This structure can speed up the retrieval by category, especially when the data volume is large. By querying the complete records of the query data set, the sorting tool can adjust the order according to the category distribution.

[0048] Preferably, if the "transaction volume" is sorted from high to low, it can intuitively reflect which categories are outstanding, forming a sorted data set. This kind of adjustment helps to prioritize the processing of core data. According to the structural characteristics of the sorted data set, a consistency verification tool can verify the integrity.

[0049] For example, check whether there are still null values or abnormal formats in the "transaction volume" field. If problems are found, they are marked and corrected, and finally a verified data set is obtained.

[0050] In one embodiment, a visualization tool can be used to assist in verification to visually display data consistency. This method can effectively improve the credibility of the data and provide reliable support for business decisions.

[0051] Step S103, obtain time series features according to the classification data set, and perform dynamic change sampling through a sliding window technique according to the acquisition difficulty of the time series features to obtain dynamic data samples, including: S1031, obtain dynamic change data from the time series through a window technique to obtain dynamic data samples; S1032. Calculate the sequence distribution using statistical tools according to the change trend of the dynamic data sample to obtain a distribution data sample. S1033. Extract time features from the distribution data sample and identify dynamic change patterns through a clustering algorithm to obtain a pattern data sample. S1034. Construct a time series query structure using an indexing tool according to the sequence distribution in the pattern data sample to obtain a query data sample. S1035. If the change trend in the query data sample meets a preset threshold, determine a consistency data sample; if not, adjust the dynamic sample through a mean filling tool to obtain an adjusted data sample.

[0052] S1036. Adjust the time features in the adjusted data sample and use a sorting tool to adjust the order according to the sequence distribution to obtain a sorted data sample.

[0053] S1037. Verify the integrity through a consistency verification tool according to the dynamic change pattern in the sorted data sample to obtain a verified data sample.

[0054] Specifically, the sliding window technique is often used to extract dynamic change data from a time series.

[0055] For example, the sales data of an enterprise includes the daily transaction volume with a time span of 30 days. Through the sliding window technique, the window size can be set to 7 days and slide from the 1st day to the 24th day to gradually obtain the dynamic data sample of each week. This method can capture the short-term fluctuation characteristics of the transaction volume.

[0056] Exemplarily, if the transaction volume within a certain window gradually rises from 100 to 150, it can be initially judged that there is an upward trend. For the change trend in the dynamic data sample, statistical tools can calculate the sequence distribution.

[0057] Specifically, the average transaction volume and standard deviation within each window can be statistically calculated to obtain a distribution data sample.

[0058] In a possible implementation, the average transaction volume of a certain window is 120 and the standard deviation is 15, indicating that the data is relatively concentrated.

[0059] It should be noted that this distribution analysis helps to judge the stability of the data. When extracting time features from the distribution data sample, the start time of the window and the change rate can be selected as key points. Identify dynamic change patterns through a clustering algorithm such as K-Means.

[0060] For example, divide multiple windows into three categories: "stable", "rising", and "falling" to obtain a pattern data sample.

[0061] In one embodiment, a certain pattern shows that the trading volume of three consecutive windows rises from 100 to 130, which is classified as the "rising" pattern. This classification facilitates grasping the change rule. According to the sequence distribution in the pattern data sample, the indexing tool can construct a time series query structure.

[0062] Preferably, an index is created according to the timestamp to generate a query data sample.

[0063] For example, querying all "rising" pattern windows within a certain period of time can quickly locate relevant records. If the change trend of the query data sample meets a preset threshold, such as the rising rate being greater than 10%, it is determined as a consistent data sample. On the contrary, if the rising rate of a certain window is only 5%, it can be adjusted by the mean filling tool.

[0064] For example, filling the abnormal low value of 80 to the window mean of 100 to obtain an adjusted data sample. By adjusting the time characteristics in the data sample, the sorting tool can adjust the order according to the sequence distribution.

[0065] For example, sorting by the trading volume change rate from high to low to generate a sorted data sample.

[0066] In one embodiment, the change rate of a certain window is 20% and ranks first, which preferentially reflects the significant trend.

[0067] It can be understood that this sorting can highlight the key change points. According to the dynamic change pattern in the sorted data sample, the consistency verification tool verifies the integrity.

[0068] For example, checking whether all window data is complete and without missing values, and finally obtaining a verified data sample.

[0069] Exemplarily, if the trading volume of a certain window is empty, it can be marked and filled with the mean value of 120. This method ensures data reliability and provides support for subsequent analysis.

[0070] Step S104, constructing a relationship network based on the dynamic data sample and modeling to mine implicit associations, and optimizing the information flow path and transmission according to the results to obtain target transmission data, including: S1041, constructing a relationship network based on the dynamic data sample, performing multi-layer data relationship modeling through the graph neural network algorithm, and mining the implicit associations between nodes with deep data processing requirements to obtain relationship-enhanced data; S1042, obtaining the information flow path according to the relationship-enhanced data, optimizing the data transmission order through the shortest path algorithm, and adjusting the path weights according to the information flow efficiency requirements to obtain the target transmission data; According to step S1041, a relationship network is constructed through dynamic data, and a graph neural network algorithm is used to mine multi-layer implicit associations to obtain enhanced relationship data. The association strength between nodes is obtained from the enhanced relationship data, and statistical tools are used to calculate the distribution characteristics to obtain distribution characteristic data. According to the changes in the implicit associations in the distribution characteristic data, a clustering algorithm is used to identify node groups to obtain group division data. Through the relationship network characteristics in the group division data, the hierarchical structure between nodes is obtained to obtain hierarchical structure data. If the association strength in the hierarchical structure data meets the preset threshold, consistent data is determined; if not, the distribution characteristics are adjusted through a mean filling tool to obtain adjusted distribution data. According to the node group characteristics in the adjusted distribution data, a sorting tool is used to adjust the hierarchical order to obtain sorted hierarchical data. Through the multi-layer association characteristics in the sorted hierarchical data, an enhanced relationship network structure is obtained to obtain verified network data.

[0071] According to step S1042, the information flow path is obtained through relationship-enhanced data, and the shortest path algorithm is used to adjust the transmission order to obtain preliminary transmission data. The path weights are extracted from the preliminary transmission data, and a mean filling tool is used to adjust the weight distribution to obtain balanced weight data. According to the information flow characteristics in the balanced weight data, the path calculation result is calculated to obtain optimized path data. If the transfer efficiency in the optimized path data meets the preset threshold, consistent data is determined; if not, the transmission order is adjusted through a sorting tool to obtain rearranged transmission data. Through the path weight distribution in the rearranged transmission data, an optimized transmission structure is obtained to obtain verified transmission data. According to the data calculation characteristics in the verified transmission data, it is judged whether the efficiency requirement is met to obtain the final transmission data.

[0072] Specifically, when obtaining the information flow path through relationship-enhanced data, it can be regarded as the transaction transfer process in an enterprise sales network.

[0073] Exemplarily, assume a sales chain includes customer A, product X, and customer B. The information flow path is that A purchases X and then transfers it to B. The principle of the shortest path algorithm to adjust the transmission order lies in identifying the optimal connection between nodes.

[0074] For example, the direct transaction path weight from customer A to product X is 10, while the path weight through middleman C is 15. The shortest path algorithm will preferentially select the direct path with a weight of 10 to generate preliminary transmission data.

[0075] Specifically, the preliminary transmission data may show that the total path weight of A-X-B is 20. After extracting the path weights from the preliminary transmission data, the implementation method of the mean filling tool to adjust the weight distribution can be to balance the outliers.

[0076] In a possible implementation, if the weight of A-X is 10 and the weight of X-B is 30, the mean filling tool will increase the lower weight of 10 to be close to the mean of 20, obtaining balanced weight data. This adjustment can smooth the fluctuations in information flow.

[0077] For example, the balanced weight data may show that the weights of both A-X and X-B are 20, reflecting more stable transfer characteristics. When calculating the path result based on the information flow characteristics in the balanced weight data, it can be understood as analyzing the transmission efficiency.

[0078] Preferably, if the total weight of the path A-X-B is 40 and the preset efficiency threshold is 50, then optimization is required. In an embodiment where the sorting tool adjusts the transmission order, the priority of X-B can be increased to reduce the intermediate delay, generating rearranged transmission data.

[0079] For example, after rearrangement, the path becomes X-B-A, and the total weight drops to 35 but the efficiency increases. When obtaining the optimized transmission structure through the path weight distribution in the rearranged transmission data, it is verified that the generation of the transmission data depends on the weight balance.

[0080] Specifically, if the weight of X-B is 15 and that of B-A is 20, the optimized transmission structure may reveal that B is the core node.

[0081] In one embodiment, the verification of the calculation characteristics of the transmission data shows that the transfer time is reduced from 5 days to 3 days, and the efficiency requirement is met.

[0082] It should be noted that the judgment of the final transmission data is based on the efficiency threshold.

[0083] For example, if the threshold is 3 days, and the verification data meets the requirements, then the consistency data is confirmed. This method can help enterprises optimize the sales path.

[0084] For example, as the core node, customer B can be preferentially allocated resources to improve the overall transfer efficiency.

[0085] It can be understood that this kind of optimization can also reveal potential bottlenecks. For example, the high weight of the X-B path may indicate the need for logistics improvement, thereby supporting sales decisions.

[0086] According to step S105, extract key indicators from the target transmission data, and obtain the predicted value of short-term resource requirements and the real-time load of the current system resource usage status. Through the weighted average algorithm, fuse the predicted value and the real-time load to obtain the adjusted resource requirement evaluation value, including: Obtain key indicators according to the target transmission data, combine the predicted value of short-term resource requirements, and perform real-time load fusion through the weighted average method to obtain preliminary resource requirement data; Based on the sudden increase trend of the preliminary resource demand data, determine whether there is abnormal fluctuation to obtain fluctuation detection data; According to the downward trend in the fluctuation detection data, adjust the resource allocation order through a sorting tool to obtain rearranged resource data; According to the system state of the rearranged resource data, obtain balanced load data by acquiring the load balancing distribution; Obtain the data fusion characteristics according to the balanced load data. If the sudden increase trend exceeds the preset threshold, adjust the resource demand to obtain corrected demand data; According to the estimated value of the corrected demand data, determine the verified demand data by judging whether the resource demand meets the system state; According to the key indicators of the verified demand data, adjust the resource allocation through a mean filling tool to obtain the resource demand evaluation value.

[0087] Step S106: Determine the weight allocation ratio according to the adjusted resource demand evaluation value, and perform error prediction on the weight allocation ratio through a regression algorithm to obtain analysis result data, including: S1061: Determine the weight allocation ratio according to the adjusted resource demand evaluation value, and perform prediction on the real-time requirement and accuracy requirement through a regression analysis algorithm to obtain preliminary prediction data; S1062: If the prediction error in the preliminary prediction data is lower than the preset threshold, determine the analysis result data. If it is higher than the preset threshold, adjust the model parameters for re-prediction to obtain corrected prediction data; S1063: According to the real-time requirement of the corrected prediction data, obtain trend analysis data by analyzing the change trend of the allocation ratio; S1064: According to the accuracy requirement of the trend analysis data, adjust the weight allocation through a mean filling tool to obtain balanced allocation data; S1065: According to the analysis result of the balanced allocation data, detect the fluctuation of the resource demand to obtain fluctuation detection data; S1066: According to the change trend of the fluctuation detection data, adjust the estimated value of the resource demand to obtain the final demand data; S1067: Extract the key indicators according to the final demand data, and determine the verification result data as the analysis result data by judging whether the real-time requirement meets the system state.

[0088] Specifically, in resource demand prediction, the weight allocation ratio is determined based on the adjusted resource demand estimate. For example, the weight for real-time requirements is set to 6, and the weight for accuracy requirements is set to 4. The multiple linear regression analysis algorithm is used, with historical resource usage data as the training set to construct a regression model. Suppose the current system's disk I / O utilization rate is 45% and the network latency is 30 ms. The regression model predicts that the disk I / O utilization rate may rise to 60% and the network latency may increase to 40 ms within the next half hour. By calculating the error between the predicted value and the actual value, if the error is lower than the preset threshold of 5%, the predicted result is directly adopted; if the error is higher than the threshold, the regression model parameters are adjusted, such as increasing the time window of historical data or adjusting the feature weights, and prediction is carried out again.

[0089] For example, after adjustment, the predicted disk I / O utilization rate is 58% and the network latency is 38 ms, and the error drops to 4%, meeting the requirements. Finally, the analysis result data is combined with the system resource scheduling strategy to dynamically adjust resource allocation. For example, according to the prediction result, the disk I / O bandwidth is increased in advance or the network routing is optimized to ensure stable system performance. This prediction method based on regression analysis can effectively balance real-time and accuracy, providing a reliable basis for resource scheduling.

[0090] Step S107: Generate a multi-dimensional feature vector based on the analysis result data, perform dimensionality reduction processing on the data analysis connection requirements through the principal component analysis algorithm, and optimize the feature extraction to adjust the display parameter range to obtain associated data, including: S1071: Generate a multi-dimensional feature vector based on the analysis result data, perform dimensionality reduction processing on the data analysis connection requirements through the principal component analysis algorithm to obtain dimensionality reduction analysis data; Construct a multi-dimensional feature vector through the analysis result data, use the principal component analysis algorithm to perform dimensionality reduction processing on the connection requirements to obtain dimensionality reduction analysis data. For the core information in the dimensionality reduction analysis data, use the mean filling tool to adjust the information dimension to obtain balanced adjustment data. According to the transformation requirements in the balanced adjustment data, calculate the change trend of business insights to obtain trend analysis data. If the connection requirements in the trend analysis data meet the preset threshold, determine the allocation ratio to obtain ratio allocation data. Detect the fluctuation situation of the data construction through the retained dimensions in the ratio allocation data to obtain fluctuation detection data. According to the change trend in the fluctuation detection data, adjust the distribution structure of the multi-dimensional features to obtain optimized distribution data. For the information dimension in the optimized distribution data, extract key indicators to determine whether the business insights meet the system state to obtain verification result data.

[0091] Specifically, in the scenario of business data analysis, first extract the original features from the user behavior logs, including 20 initial dimensions such as the daily access frequency (e.g., user A averages 12 times), the page stay duration (e.g., the product page averages 95 seconds), the conversion rate (e.g., 2%), etc., to form the original feature matrix. Use the principal component analysis algorithm, set the variance contribution rate threshold at 85%, calculate the eigenvalues and eigenvectors through the covariance matrix, and find that the cumulative contribution rate of the first 5 principal components reaches 87%. Among them, the first principal component (PC1) has the highest loadings for the access frequency (62) and the stay duration (58), reflecting user activity; the second principal component (PC2) highlights the conversion rate (71) and the unit price per customer (65), representing the consumption quality. Project the 20-dimensional data into a 5-dimensional space. For example, the original feature vector of user B [15, 120, 5%...] is reduced to [34, -87, 0...] after linear transformation. For the dimensionality-reduced data, combine the K-means clustering algorithm (k = 4, Euclidean distance) to divide the user groups. Cluster 1 (high activity, low conversion) accounts for 32%, and cluster 2 (low activity, high conversion) accounts for 18%. Verify the clustering effectiveness through the silhouette coefficient of 68. Finally, output the dimensionality-reduced data and the clustering labels to the recommendation system. For example, give priority to pushing high-unit-price products to users in cluster 2 to achieve a balance between business requirements and computational efficiency.

[0092] S1072. According to the optimized feature extraction process of the dimensionality reduction analysis data, determine the display parameter range by the data change trend threshold to obtain the associated data of resource fluctuation and visualization structure, including: S10721. Adjust the feature extraction process according to the dimensionality reduction analysis data, and judge the sudden increase trend of the preset threshold to obtain the trend judgment data; S10722. According to the downward trend of the trend judgment data, adjust the display parameter range to obtain the parameter optimization data; S10723. According to the resource fluctuation of the parameter optimization data, detect the distribution of the visualization structure to obtain the distribution detection data; S10724. According to the associated data of the distribution detection data, extract the core dimension of the data optimization to obtain the dimension extraction data; S10725. If the dimension extraction data meets the judgment threshold, judge the feature extraction adjustment trend through the trend judgment to obtain the adjustment analysis data; S10726. According to the visualization structure of the adjustment analysis data, optimize the allocation ratio of the resource fluctuation to obtain the ratio optimization data; S10727. According to the parameter range of the ratio optimization data, judge the change of the sudden increase trend to obtain the change detection data.

[0093] Specifically, in the resource monitoring scenario, by analyzing historical resource usage data, 15 initial dimensions such as key features like CPU usage rate (e.g., server A averages 75%), memory occupancy (e.g., average 45GB), network bandwidth (e.g., average 120Mbps), etc. are extracted to construct an original feature matrix. The Isolation Forest algorithm is adopted, and the anomaly score threshold is set to 65. By calculating the feature values and anomaly scores, anomaly points where resource usage suddenly increases or decreases are identified. For example, the CPU usage rate of server B suddenly increases to 95% during a certain period, and the memory occupancy drops to 30GB. According to the distribution of anomaly points, the range of visualization parameters is adjusted. The display range of CPU usage rate is set to 0% to 100%, the display range of memory occupancy is set to 0GB to 64GB, and the display range of network bandwidth is set to 0Mbps to 200Mbps. By dynamically adjusting the display parameters and combining with time series analysis algorithms (such as ARIMA model), the future resource usage trend is predicted. For example, it is predicted that the CPU usage rate of server C will stabilize between 80% and 85% within the next 24 hours. Finally, the resource fluctuation data is associated with the visualization structure to generate a dynamic monitoring chart, which reflects the resource usage status in real time and provides data support for resource scheduling.

[0094] Step S108, construct a visualization model based on the dimensionality reduction analysis data, convert the multi-dimensional feature vectors into graphic elements through data mapping, and adjust the display parameters of the graphic elements according to the presentation ability requirements to obtain visualization presentation parameters, including: S1081, according to the dimensionality reduction analysis data, obtain feature vectors through parsing; S1082, according to the vector features, perform vector conversion through data mapping technology to obtain graphic elements; S1083, adjust the display parameters according to the presentation ability to determine the display form of the graphic elements; S1084, construct a visualization model according to the mapping technology to obtain presentation data; S1085, if the presentation data meets the preset threshold, optimize the distribution of the graphic elements through parameter adjustment to obtain the adjusted presentation data; S1086, according to the adjusted presentation data, detect the change trend of the feature vectors to judge the improvement of the presentation ability; S1087, update the visualization model according to the change trend to obtain optimized presentation data.

[0095] Specifically, in the scenario of financial transaction data analysis, the original data contains 20 feature dimensions, including trading volume (such as the average daily trading volume of stock X is 12 million shares), price volatility (such as the standard deviation is 15), bid-ask spread (such as the average is 0.3 yuan), etc. The dimensions are reduced to 5 through the principal component analysis (PCA) algorithm, retaining 95% of the variance contribution rate. The weight of the first principal component is 42, reflecting the dominance of price fluctuations. The t-SNE algorithm is used to map the feature vectors after dimensionality reduction to a two-dimensional space, setting the perplexity parameter to 30, the learning rate to 200, and iterating 1000 times to generate a scatter plot clustering distribution. For example, the center coordinates of the high-frequency trading cluster are (-2, 8), and the center coordinates of the low-frequency trading cluster are (5, -6). The color gradient of the heat map is dynamically adjusted according to the clustering results. The trading volume interval is divided into 0 - 5 million (blue), 5 - 15 million (green), and over 15 million (red). The price volatility interval is divided into 0 - 1 (light color), 1 - 3 (medium color), and over 3 (dark color). Combining with the DBSCAN clustering algorithm, setting the neighborhood radius to 5 and the minimum number of samples to 10, abnormal trading patterns are identified, such as an outlier with a trading volume suddenly increasing to 30 million and a volatility reaching 4 during a certain period. By dynamically adjusting the time granularity of the candlestick chart, the 1-minute line is switched to a 5-minute line to smooth the noise, and at the same time, the Bollinger Bands indicator (period 20, standard deviation multiple 2) is superimposed to display the price channel. When the price breaks through the upper track, a red warning mark is triggered. Finally, an interactive visualization panel is generated to synchronously display the results of dimensionality reduction clustering, heat map, and candlestick analysis, realizing the presentation of multi-dimensional trading feature associations.

[0096] Step S109: Obtain user feedback according to the visualization presentation parameters, optimize the system-level collaboration requirements through an iterative update algorithm to obtain feedback data, and compare the feedback data with preset improvement conditions. Adjust the system level to obtain optimized system data, including: S1091: Obtain user feedback according to user interaction data, and obtain the distribution characteristics of the feedback content through a recording tool; S1092: Judge the feedback trigger situation according to the distribution characteristics. If the preset improvement conditions are triggered, adjust the system-level parameters through an iterative algorithm to obtain the adjusted collaboration requirement data; S1093: According to the adjusted collaboration requirement data, obtain the update direction of the processing parameters through an analysis process; S1094: According to the update direction, adjust the processing parameters to obtain optimized system-level data; S1095: According to the optimized system-level data, judge whether the feedback trigger decreases by detecting the change trend of the interaction data; S1096: If the change trend shows that the trigger decreases, update the presentation form through an analysis process to obtain optimized data; S1097. Detect the matching degree of the collaboration requirements through a verification tool according to the optimized data, and determine the optimized system data output by the final system.

[0097] Specifically, during the interaction between the user and the visualization system, the system captures the user's click behavior on the hotspots of the clustering distribution in real time. For example, a certain click focuses on the data points within a radius of 5 near the coordinates (2, -8), triggering the recalculation of local data based on density. The Gaussian Mixture Model (GMM) is used to recluster the data in the clicked area, with the initial number of clusters set to 3, the covariance type to full covariance, the maximum number of iterations to 500, and the convergence threshold to 1e-4, obtaining new mean vectors (1, -7), (3, -9), and (0, -6), and covariance matrices [[2, 05], [05, 3]], [[3, 1], [1, 4]], and [[1, 02], [02, 2]] respectively. According to the new clustering results, the system automatically adjusts the weight distribution of the parallel coordinate axes, increasing the original weights of order thickness (original weight 15) and liquidity gap (original weight 08) in the original features to 22 and 12 respectively, while reducing the weight of the position change rate from 2 to 1. By analyzing the user's continuous operation sequence through the Hidden Markov Model (HMM), with the number of states set to 4, the number of observation symbols to 6, and the initial value of the transition probability matrix set to a uniform distribution, after iterative update through the forward-backward algorithm, the user preference pattern of "first view the volatility distribution and then check for trading volume anomalies" is identified, with a probability of 78%. Based on this pattern, the system dynamically loads the LSTM prediction module, with an input window length of 10, 64 hidden layer units, and outputs the predicted values of the liquidity risk indicators for the next 5 time steps (e.g., the predicted value at time t+1 is 45, and the confidence interval is [38, 52]), and superimposes the prediction results on the existing visualization view as a semi-transparent ribbon layer. When the user stays on a certain asset classification for more than 30 seconds, it triggers the feature importance analysis based on the Gradient Boosting Decision Tree (GBDT), with the tree depth set to 6, the learning rate to 1, and the number of iterations to 200. It is calculated that the buy-sell pressure ratio (importance score 35) and the order book slope (importance score 28) are the key influencing factors in the current view, and then the real-time change curves of these two indicators are prominently displayed in the sidebar of the interface, with the sampling frequency increased from the default 1 second / time to 2 seconds / time.

[0098] Through the method of the present invention, first, it realizes the compatibility of multi-source heterogeneous data. Through operations such as normalization processing and clustering grouping, data from different sources can be effectively integrated and utilized. Secondly, it uses a variety of algorithms for data modeling, such as constructing a relationship network to mine implicit associations and regression analysis to predict errors, improving the depth and accuracy of data processing. Furthermore, it can dynamically display data. Through the visualization model and parameter adjustment, data is presented according to the intuitive presentation requirements. Finally, it optimizes the system in combination with user feedback. By iteratively updating and adjusting the system hierarchy, the system better meets the actual business needs, improving the efficiency and quality of intelligent business operations.

[0099] An embodiment of the present invention also provides an information management device based on intelligent commerce, such as Figure 2 shown, including: A data acquisition module, configured to obtain multi-source raw data through a preset standard protocol, map it into a unified structure through a format conversion algorithm, and obtain a normalized data set; A data grouping module, configured to extract characteristic attribute data according to the normalized data set, perform grouping processing on the characteristic attribute data through a clustering algorithm, and use the grouping result that meets the heterogeneity threshold as a classification data set; A dynamic analysis module, configured to obtain time series characteristics according to the classification data set, and perform dynamic change sampling through a sliding window technique according to the acquisition difficulty of the time series characteristics to obtain a dynamic data sample; A modeling optimization module, configured to construct a relationship network and perform modeling to mine implicit associations according to the dynamic data sample, optimize the information flow path and transmission according to the result, and obtain target transmission data; A resource evaluation module, configured to extract key indicators according to the target transmission data, obtain a predicted value of short-term resource requirements and a real-time load of the current system resource usage status, and fuse the predicted value and the real-time load through a weighted average algorithm to obtain an adjusted resource requirement evaluation value; An evaluation prediction module, configured to determine a weight allocation ratio according to the adjusted resource requirement evaluation value, perform error prediction on the weight allocation ratio through a regression algorithm, and obtain analysis result data; A vector association module, configured to generate a multi-dimensional feature vector according to the analysis result data, perform dimensionality reduction processing on the data analysis connection requirements through a principal component analysis algorithm, and optimize the feature extraction to adjust the display parameter range to obtain associated data; A parameter modeling module, configured to construct a visualization model according to the dimensionality reduction analysis data, convert the multi-dimensional feature vector into graphic elements through data mapping, and adjust the display parameters of the graphic elements according to the intuitive presentation ability requirement to obtain visualization presentation parameters; A feedback optimization module, configured to obtain user feedback according to the visualization presentation parameters, optimize the system-level collaboration requirements through an iterative update algorithm to obtain feedback data, compare the feedback data with preset improvement conditions, and adjust the system level to obtain optimized system data.

[0100] It should be noted that an information management device based on intelligent commerce provided by an embodiment of the present invention is used to execute all process steps of an information management method based on intelligent commerce in the above embodiment, and the working principles and beneficial effects of the two correspond one by one, so they will not be elaborated here.

[0101] An embodiment of the present invention also provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a data acquisition program. When the processor executes the computer program, it implements the steps in each of the above-described embodiments of the information management method based on intelligent commerce, such as Figure 1 the step S101 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in each of the above-described device embodiments, such as a data acquisition module.

[0102] Exemplarily, the computer program may be divided into one or more modules / units, and the one or more modules / units are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device.

[0103] The electronic device may be a computing device such as a desktop computer, a notebook, a palm computer, and a smart tablet. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the above components are only examples of the electronic device and do not constitute a limitation on the electronic device. It may include more or fewer components than the above, or combine certain components, or different components. For example, the electronic device may further include input / output devices, network access devices, a bus, etc.

[0104] The so-called processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the electronic device and connects various parts of the entire electronic device through various interfaces and lines.

[0105] The memory can be used to store the computer program and / or modules. By running or executing the computer program and / or modules stored in the memory, and invoking the data stored in the memory, the processor can implement various functions of the electronic device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store the operating system, application programs required for at least one function (such as the sound playback function, the image playback function, etc.); the data storage area can store the data created according to the use of the mobile phone (such as audio data, phone book, etc.). In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage devices.

[0106] Among them, if the modules / units integrated in the electronic device are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above-mentioned embodiment methods of the present invention, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0107] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the attached drawings of the device embodiments provided by the present invention, the connection relationship between modules indicates that they have a communication connection, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0108] The specific embodiments described above further elaborate on the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. In particular, for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An information management method based on intelligent commerce, characterized in that, Including: Obtain multi-source raw data through a preset standard protocol, map it into a unified structure through a format conversion algorithm, and obtain a normalized data set; Extract feature attribute data according to the normalized data set, group the feature attribute data through a clustering algorithm, and use the grouping result that meets the heterogeneity threshold as a classification data set; Obtain time series features according to the classification data set, and perform dynamic change sampling through a sliding window technique according to the acquisition difficulty of the time series features to obtain dynamic data samples; Construct a relationship network based on the dynamic data samples and perform modeling to mine implicit associations, optimize the information flow path and transmission according to the results, and obtain target transmission data; Extract key indicators according to the target transmission data, obtain the predicted value of short-term resource requirements and the real-time load of the current system resource usage status, and fuse the predicted value and the real-time load through a weighted average algorithm to obtain an adjusted resource requirement evaluation value; Determine the weight allocation ratio according to the adjusted resource requirement evaluation value, and perform error prediction on the weight allocation ratio through a regression algorithm to obtain analysis result data; Generate a multi-dimensional feature vector according to the analysis result data, perform dimensionality reduction processing on the data analysis connection requirements through a principal component analysis algorithm, and optimize the feature extraction to adjust the display parameter range to obtain associated data; Construct a visualization model according to the dimensionality reduction analysis data, convert the multi-dimensional feature vector into graphic elements through data mapping, and adjust the display parameters of the graphic elements according to the intuitive presentation ability requirements to obtain visualization presentation parameters; Obtain user feedback according to the visualization presentation parameters, optimize the system-level collaboration requirements through an iterative update algorithm to obtain feedback data, compare the feedback data with preset improvement conditions, and adjust the system level to obtain optimized system data.

2. The information management method based on intelligent commerce according to claim 1, wherein Extract feature attribute data according to the normalized data set, group the feature attribute data through a clustering algorithm, and use the grouping result that meets the heterogeneity threshold as a classification data set, including: Obtain structural features according to the classification data set, perform category distribution processing through statistical tools, and obtain a distribution data set; Extract keyword fields according to the distribution data set. If the keyword fields meet the consistency check, determine a consistency data set; if not, adjust the field content through a field mapping tool to obtain an adjusted data set; Adjust according to the category distribution of the data set, identify outliers through a clustering algorithm, and obtain an outlier marked data set; Correct outliers through a mean filling tool according to the marking information of the outlier marked data set to obtain a corrected data set; Extract associated fields according to the corrected data set, and construct a query structure through an indexing tool to obtain a query data set; Adjust the order according to the category distribution through a sorting tool according to the complete records of the query data set to obtain a sorted data set; Perform integrity verification through a consistency check tool according to the structural features of the sorted data set to obtain a verification data set.

3. The information management method based on intelligent commerce according to claim 1, characterized in that Construct a relationship network based on the dynamic data samples, model and mine the implicit associations, and optimize the information flow path and transmission according to the results to obtain the target transmission data, including: Construct a relationship network based on the dynamic data samples, perform multi-layer data relationship modeling through the graph neural network algorithm, and mine the implicit associations between nodes with deep data processing requirements to obtain relationship-enhanced data; Obtain the information flow path according to the relationship-enhanced data, optimize the data transmission order through the shortest path algorithm, and adjust the path weights according to the information flow efficiency requirements to obtain the target transmission data.

4. The information management method based on intelligent commerce according to claim 1, wherein Extract key indicators from the target transmission data, obtain the predicted value of the short-term resource requirements and the real-time load of the current system resource usage status, and fuse the predicted value and the real-time load through the weighted average algorithm to obtain the adjusted resource requirement evaluation value, including: Obtain key indicators from the target transmission data, combine the predicted value of the short-term resource requirements, and perform real-time load fusion through the weighted average method to obtain the preliminary resource requirement data; Judge whether there is abnormal fluctuation according to the sudden increase trend of the preliminary resource requirement data to obtain the fluctuation detection data; Adjust the resource allocation order through the sorting tool according to the downward trend in the fluctuation detection data to obtain the rearranged resource data; Obtain the balanced load data by obtaining the load balancing distribution according to the system state of the rearranged resource data; Obtain the data fusion characteristics according to the balanced load data. If the sudden increase trend exceeds the preset threshold, adjust the resource requirements to obtain the corrected requirement data; Obtain the verified requirement data by judging whether the resource requirements meet the system state according to the estimated value of the corrected requirement data; Adjust the resource allocation through the mean filling tool according to the key indicators of the verified requirement data to obtain the resource requirement evaluation value.

5. The information management method based on intelligent commerce according to claim 4, wherein, Determine the weight allocation ratio according to the adjusted resource requirement evaluation value, and predict the error of the weight allocation ratio through the regression algorithm to obtain the analysis result data, including: Determine the weight allocation ratio according to the adjusted resource requirement evaluation value, and predict the real-time requirement and the accuracy requirement through the regression analysis algorithm to obtain the preliminary prediction data; If the prediction error in the preliminary prediction data is lower than the preset threshold, determine the analysis result data. If it is higher than the preset threshold, adjust the model parameters for re-prediction to obtain the corrected prediction data; Obtain the trend analysis data by analyzing the change trend of the allocation ratio according to the real-time requirement of the corrected prediction data; Adjust the weight allocation through the mean filling tool according to the accuracy requirement of the trend analysis data to obtain the balanced allocation data; Detect the fluctuation of the resource requirements according to the analysis result of the balanced allocation data to obtain the fluctuation detection data; Adjust the estimated value of the resource requirements according to the change trend of the fluctuation detection data to obtain the final requirement data; Extract key indicators from the final requirement data, and obtain the verification result data as the analysis result data by judging whether the real-time requirement meets the system state.

6. The information management method based on intelligent commerce according to claim 1, characterized in that, Generate a multi-dimensional feature vector based on the analysis result data, perform dimensionality reduction processing on the data analysis connection requirements through the principal component analysis algorithm, and optimize the feature extraction to adjust the display parameter range to obtain associated data, including: Generate a multi-dimensional feature vector based on the analysis result data, and perform dimensionality reduction processing on the data analysis connection requirements through the principal component analysis algorithm to obtain dimensionality reduction analysis data; Based on the optimized feature extraction process of the dimensionality reduction analysis data, adjust the display parameter range through the data change trend determination threshold to obtain associated data on resource fluctuations and visualization structures, including: Adjust the feature extraction process according to the dimensionality reduction analysis data, and judge the sudden increase trend of the preset threshold to obtain trend determination data; Based on the downward trend of the trend determination data, adjust the display parameter range to obtain parameter optimization data; Based on the resource fluctuations of the parameter optimization data, detect the distribution of the visualization structure to obtain distribution detection data; Based on the associated data of the distribution detection data, extract the core dimension of data optimization to obtain dimension extraction data; If the dimension extraction data meets the determination threshold, adjust the trend determination through feature extraction to obtain adjusted analysis data; Based on the visualization structure of the adjusted analysis data, optimize the allocation ratio of resource fluctuations to obtain ratio optimization data; Based on the parameter range of the ratio optimization data, judge the change of the sudden increase trend to obtain change detection data.

7. The information management method based on intelligent commerce according to claim 1, characterized in that Construct a visualization model based on the dimensionality reduction analysis data, convert the multi-dimensional feature vector into graphic elements through data mapping, and adjust the display parameters of the graphic elements according to the presentation ability requirements to obtain visualization presentation parameters, including: Obtain the feature vector by parsing according to the dimensionality reduction analysis data; Perform vector conversion through data mapping technology according to the vector feature to obtain graphic elements; Adjust the display parameters according to the presentation ability to determine the display form of the graphic elements; Construct a visualization model according to the mapping technology to obtain presentation data; If the presentation data meets the preset threshold, optimize the distribution of the graphic elements through parameter adjustment to obtain adjusted presentation data; Based on the adjusted presentation data, detect the change trend of the feature vector to judge the improvement of the presentation ability; Update the visualization model according to the change trend to obtain optimized presentation data.

8. The information management method based on intelligent commerce according to claim 1, wherein Obtain user feedback based on the visualization presentation parameters, optimize the system-level collaboration requirements through the iterative update algorithm to obtain feedback data, and compare the feedback data with the preset improvement conditions, and adjust the system level to obtain optimized system data, including: Obtain user feedback based on user interaction data, and obtain the distribution characteristics of the feedback content through a recording tool; Judge the feedback trigger situation according to the distribution characteristics. If the preset improvement condition is triggered, adjust the system-level parameters through the iterative algorithm to obtain adjusted collaborative requirement data; Based on the adjusted collaborative requirement data, obtain the update direction of the processing parameters by parsing the process; According to the update direction, obtain optimized system-level data by adjusting processing parameters; According to the optimized system-level data, judge whether the feedback trigger is reduced by detecting the change trend of interaction data; If the change trend shows a reduction in the trigger, update the presentation form through the parsing process to obtain optimized data; According to the optimized data, detect the matching degree of the collaboration requirements through a verification tool to determine the optimized system data output by the final system.

9. An information management device based on intelligent commerce, characterized in that, Including: A data acquisition module, configured to obtain multi-source raw data through a preset standard protocol, map it into a unified structure through a format conversion algorithm, and obtain a normalized data set; A data grouping module, configured to extract feature attribute data according to the normalized data set, group the feature attribute data through a clustering algorithm, and use the grouping result that meets the heterogeneity threshold as a classification data set; A dynamic analysis module, configured to obtain time series features according to the classification data set, and perform dynamic change sampling through a sliding window technique according to the acquisition difficulty of the time series features to obtain dynamic data samples; A modeling optimization module, configured to construct a relationship network and model to mine implicit associations according to the dynamic data samples, and optimize the information flow path and transmission according to the results to obtain target transmission data; A resource evaluation module, configured to extract key indicators according to the target transmission data, obtain the predicted value of short-term resource requirements and the real-time load of the current system resource usage status, and fuse the predicted value and the real-time load through a weighted average algorithm to obtain an adjusted resource requirement evaluation value; An evaluation prediction module, configured to determine the weight allocation ratio according to the adjusted resource requirement evaluation value, and perform error prediction on the weight allocation ratio through a regression algorithm to obtain analysis result data; A vector association module, configured to generate a multi-dimensional feature vector according to the analysis result data, perform dimensionality reduction processing on the data analysis connection requirements through a principal component analysis algorithm, and optimize the feature extraction to adjust the display parameter range to obtain associated data; A parameter modeling module, configured to construct a visualization model according to the dimensionality reduction analysis data, convert the multi-dimensional feature vector into graphic elements through data mapping, and adjust the display parameters of the graphic elements according to the intuitive presentation ability requirement to obtain visualization presentation parameters; A feedback optimization module, configured to obtain user feedback according to the visualization presentation parameters, optimize the system-level collaboration requirements through an iterative update algorithm to obtain feedback data, compare the feedback data with preset improvement conditions, and adjust the system level to obtain optimized system data.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein when the computer program runs, it controls the device where the computer-readable storage medium is located to execute the information management method based on intelligent commerce according to any one of claims 1 to 8.

Images