Data integration method and system for multi-classification sports events and medium

Abstract

The invention relates to the technical field of data classification storage, in particular to a data integration method and system for multi-classification sports events and a medium. The method comprises the following steps: firstly, carrying out clustering analysis on a time sequence acquisition frequency of each professional data, and identifying a multi-modal frequency characteristic of the time sequence acquisition frequency; secondly, dividing high-relevance data clustering clusters by calculating collaborative indexes among frequency clustering clusters of different professional data, reducing mixed storage of low-relevance data and reducing the null value rate of a relational database; furthermore, in combination with the frequency matching strength in the data clustering cluster and the storage occupation space, the write-in demand quantity of each cluster is dynamically calculated, so that the professional data is integrated and stored. In conclusion, clustering is carried out through the collaboration of collection frequency changes of different types of professional data, the null value problem in the storage process of the relational database is reduced, meanwhile, the storage occupied space of the professional data is analyzed, and the write-in pressure of the server is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of data classification and storage, and in particular to a data integration method, system and medium for multi-category sports events. Background Art

[0002] In the current field of sports data analysis, with the rapid development of sensor technology, IoT devices, and data acquisition systems, diverse sports (such as track and field, swimming, ball games, and cycling) have generated massive amounts of multi-dimensional time-series data. This multi-dimensional professional data (such as cadence, speed, and athlete heart rate in cycling events) is widely collected, stored, and applied during training and competitions, serving scenarios such as event management, optimizing the audience experience, making business decisions, and improving athlete performance.

[0003] In the existing technology, relational databases are usually used to store multi-category sports event data. However, since the frequency of collecting professional data at different stages during a sports event varies greatly, this leads to the database model design generating a large number of redundant field null values. Especially in the face of high-concurrency real-time data updates, relational databases may have bottlenecks in scalability and response speed, making it difficult to efficiently process massive and changing data types. Summary of the Invention

[0004] To address the technical issues of large differences in the frequency of professional data collection at different stages during sports events, which results in a large number of redundant fields with empty values in database schema design, and the potential bottlenecks in scalability and response speed of relational databases in the face of highly concurrent real-time data updates, making it difficult to efficiently process massive and diverse data types, the present invention aims to provide a data integration method, system, and medium for multi-category sports events. The technical solutions adopted are as follows:

[0005] A data integration method for multi-category sports events, comprising:

[0006] In each sports event, obtain a match data set for each athlete, wherein the match data set includes time series data of the acquisition frequency of various professional data and the storage space occupied by each professional data;

[0007] In each event data set, cluster analysis is performed on the frequency values at all moments in each collected frequency time series data to obtain frequency clusters. In the frequency clusters of different professional data, the similarities between frequency clusters and the quantitative characteristics of the moments in the frequency clusters are comprehensively considered to determine the synergy index between any two professional data in each event data set. This is used to perform cluster analysis on the professional data to obtain data clusters.

[0008] In the data clusters of each event data set, the write demand of each data cluster is determined based on the matching between the frequency clusters of professional data and the storage space occupied by professional data;

[0009] All event data sets of all types of sports events are integrated and stored based on the write demand of data clusters.

[0010] Furthermore, the method for obtaining the collaborative indicator includes:

[0011] In each event data set, for any two professional data, the frequency clusters in the two professional data are combined in pairs to obtain all non-repeated combination clusters;

[0012] In each combined cluster, the similarity of frequency distribution and time between frequency clusters is analyzed to determine the frequency overlap index between the two frequency clusters;

[0013] In each event data set, one professional data is selected as the target data. Under the target data and any other professional data, among all the combination clusters corresponding to each frequency cluster in the target data, the maximum value of the frequency overlap index is used as the frequency domain overlap degree of each frequency cluster in the target data;

[0014] Based on the number of moments in the frequency clusters in the target data, the frequency domain overlap corresponding to the frequency clusters is weightedly fused to obtain the similarity index between the target data and the professional data;

[0015] For any two professional data, the normalized value of the mean of the similarity index between the two professional data is used as the synergy index between the two professional data.

[0016] Furthermore, the method for obtaining the frequency overlap index includes:

[0017] In each combined cluster, the mean of all frequency values in each frequency cluster is used as the mean eigenvalue;

[0018] The absolute value of the difference between the mean eigenvalues of the two frequency clusters in each combination cluster is negatively correlated and normalized, and is used as the first similarity factor between the two frequency clusters in each combination cluster;

[0019] Compare the moments of the frequency values in the two frequency clusters in each combination cluster, and use the proportion of the same moments as the second similarity factor between the two frequency clusters in each combination cluster;

[0020] The value obtained by normalizing the product of the first similarity factor and the second similarity factor is used as the frequency overlap index between the two frequency clustering clusters in each combined cluster.

[0021] Furthermore, the method for obtaining the write demand includes:

[0022] In each data cluster, the product of the mean of all frequency values in each frequency cluster in each professional data and the storage space occupied by each professional data is used as the write factor of each frequency cluster in each professional data;

[0023] In each data cluster, select one professional data as the test data, and select one frequency cluster in the test data as the test cluster;

[0024] In each data cluster, between the data to be tested and any remaining professional data, the frequency cluster with the largest frequency overlap index with the cluster to be tested in the professional data is used as the matching cluster of the cluster to be tested; in the data to be tested and the professional data, the sum of the write factors of the cluster to be tested and the matching cluster is normalized by the product of the frequency overlap index between the cluster to be tested and the matching cluster, and the result is used as the write index between the cluster to be tested and the matching cluster;

[0025] Under the conditions of the data to be tested and any other professional data, the maximum value of the write index of all frequency clusters in the data to be tested is used as the simultaneous write demand factor between the data to be tested and the professional data;

[0026] In each data cluster, the mean of the simultaneous write demand factors between the data to be tested and all other professional data is used as the write demand indicator corresponding to the data to be tested;

[0027] In each data cluster, the maximum value of all write demand indicators of all professional data is taken as the write demand of each data cluster.

[0028] Furthermore, the method for obtaining the frequency clusters includes:

[0029] In the collection frequency time series data of each professional data in each event data set, the DBSCAN clustering algorithm is used to perform cluster analysis on the frequency values at all times to obtain frequency cluster clusters; among them, the neighborhood radius and the minimum number of points in the neighborhood are both preset values.

[0030] Furthermore, the method for obtaining the data clusters includes:

[0031] In each event data set, the value after negative correlation mapping of the synergy index between any two professional data is used as the difference index between the two professional data;

[0032] Based on the K-means clustering algorithm and the optimal K value, all professional data in each event data set are clustered to obtain data clusters; among them, the distance metric is the difference index between professional data.

[0033] Furthermore, among all event data sets of all kinds of sports events, all event data sets are integrated and stored based on the write demand of data clusters, including:

[0034] Sort the data clusters in all data sets of all sports events in ascending order according to the write demand to obtain a sorted sequence;

[0035] In the sorting sequence, first-order difference calculation is performed on the write demand of the data clusters to obtain a difference sequence, and the latter of the two data clusters corresponding to the maximum value in the difference sequence in the sorting sequence is used as a breakpoint;

[0036] In the sorting sequence, the data clusters at and after the breakpoint are used as distributed storage data, and the data clusters before the breakpoint are used as overall storage data;

[0037] Perform cluster analysis on the data clusters in the distributed storage data based on the K-means clustering algorithm and the preset K value to obtain distributed clusters, where the distance metric is the absolute value of the difference in write demand between data clusters, and the preset K value is the number of servers;

[0038] The sum of the write requirements of all data clusters in each distributed cluster is used as the storage requirement value of each distributed cluster. All distributed clusters are sorted in descending order according to the storage requirement value to obtain a storage requirement sequence. All servers are sorted in descending order according to storage capacity to obtain a storage capacity sequence. Each distributed cluster in the storage requirement sequence is stored in the server at the same position in the storage capacity sequence.

[0039] Update the storage capacity of each server and store the overall storage data in the server with the largest storage capacity.

[0040] Furthermore, the method for obtaining the optimal K value includes:

[0041] The optimal K value is obtained based on the elbow method.

[0042] A data integration system for multi-category sports events includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and when the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor, the steps of a data integration method for multi-category sports events are implemented.

[0043] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a data integration method for multi-category sports events.

[0044] The present invention has the following beneficial effects:

[0045] Given that the frequency of collecting professional data varies between different events, and even between different schedules for the same event, storing it in a relational database can lead to null value issues. This null value problem stems significantly from the mismatch in data collection frequency. Therefore, for each sporting event, we first obtain a data set for each athlete, including the time series data on the collection frequencies of various professional data and the storage space occupied by each type of professional data. Then, within each event data set, we perform a cluster analysis on the frequency values in the time series data for each type of professional data, generating frequency clusters that can identify the collection frequency characteristics of each type of professional data. Furthermore, within the frequency clusters of different professional data, we calculate synergy indicators based on the similarity and moment quantity characteristics of the frequency clusters, and perform cluster analysis on the professional data to generate data clusters. This accurately quantifies the correlation between professional data in each event data set, reduces the mixed storage of low-correlation data in subsequent processes, and optimizes the null value problem in relational database storage. Furthermore, in order to avoid the problem of excessive server pressure caused by the simultaneous writing of a large amount of data in a certain stage of a highly collaborative data set, the present invention combines the matching between the frequency clustering clusters of professional data with the storage space occupied by professional data in the data clustering cluster of the event data set, and determines the write demand of each data clustering cluster, which is used to avoid the problem of local storage overload in the subsequent storage process. Finally, the integration of all event data sets of all kinds of sports events is driven by the write demand of the data clustering cluster, and all event data sets are stored. In summary, the present invention clusters professional data by analyzing the synergy of the changes in the collection frequency of different types of professional data during the event, thereby reducing the null value problem in the storage process of the relational database, and analyzing the storage space occupied by professional data, effectively reducing the write pressure on the server. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0047] Figure 1A method flow chart of a data integration method for multi-category sports events provided by one embodiment of the present invention;

[0048] Figure 2 A flowchart of a method for obtaining a collaborative indicator provided by an embodiment of the present invention;

[0049] Figure 3 A system block diagram of a data integration system for multiple sports events provided by one embodiment of the present invention;

[0050] Figure 4 A schematic diagram of the system structure of a data integration system for multi-category sports events provided by one embodiment of the present invention;

[0051] Figure 5 A schematic diagram of a computer-readable storage medium provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0052] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail a data integration method, system, and medium for multi-category sports events, including its specific implementation, structure, features, and effectiveness. In the following description, references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0053] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0054] The following describes in detail a data integration method, system and medium for multi-category sports events provided by the present invention with reference to the accompanying drawings.

[0055] See also Figure 1 , which shows a method flow chart of a data integration method for multi-category sports events provided by an embodiment of the present invention, the method comprising the following steps:

[0056] Step S1: Under each sports event, obtain the event data set of each athlete, where the event data set includes the acquisition frequency time series data of various professional data and the storage space occupied by each professional data.

[0057] At present, sports events cover a variety of sports types, such as cycling, marathon, triathlon and biathlon, etc., which are characterized by significant differences in rules, data dimensions and technical requirements among different events. The data of multi-category sports events include data of multiple sports events. When integrating and storing the data of multiple sports events, it is necessary to collect data of each sports event. The relevant data of sports events include project type, event name, time, venue, group, player name, number, and professional data of players during the competition. Among them, professional data, taking cycling events as an example, include cadence, speed, power, etc. In this embodiment of the present invention, the objects of integrated storage are mainly for the professional data generated under each sports event.

[0058] When existing technologies use relational databases to store professional data of sports events such as cycling, marathon, triathlon and biathlon, there will be a null value problem. The null value problem is significantly caused by the mismatch of data collection frequency. For example, bicycles need to collect power, cadence, speed, etc. at a high frequency, and the collection frequency is multiple times per second, while marathons only need to periodically record pace or heart rate, for example, once per kilometer. If high-frequency fields are preset in the statistical table, such as "second-level power value", the corresponding fields of low-frequency items such as marathons will inevitably generate a large number of null values due to data gaps; conversely, if sparse fields are reserved for low-frequency items, such as "time per kilometer", the original data with high sampling rates such as cycling cannot be directly adapted and need to be split into multiple rows for storage or forced to reduce the frequency, further aggravating the data. Data redundancy or information loss; and currently in sports events, in order to reduce invalid monitoring data during the event and to ensure the data integrity of the main monitoring phase of the event, the sensor recording frequency of endurance events has implemented "multi-dimensional scenario-driven dynamic adjustment", that is, the sensor recording frequency automatically adjusts the recording frequency according to the stage and scenario of the event. For example, in the climbing section of a bicycle, the power meter sampling rate jumps from the conventional 1Hz to 10Hz to capture subtle fluctuations in pedaling force. In the sprint phase of a marathon, the plantar pressure sensor switches from 5Hz to 50Hz to monitor changes in ground contact time. This data collection method of changing the collection frequency in real time will cause the collection frequency of even the same professional data to change significantly at different stages of the race. Therefore, in an embodiment of the present invention, in order to reduce null values in a relational database, it is necessary to determine the synergy between each two professional data (cadence, power, speed, etc.) of the same event. Synergy mainly refers to the consistency of the collection frequency and the consistency of the frequency change during the event, so as to perform cluster analysis on the professional data and finally integrate and store them.

[0059] First, in each sports event, obtain the event data set of each athlete. The event data set includes the collection frequency time series data of various professional data and the storage space occupied by each professional data. For example, in a cycling competition, the types of professional data may include the athlete's heart rate, cadence, speed, power, etc. Then, relevant sensors can be deployed on the athletes and racing cars to collect the athlete's heart rate collection frequency time series data, cadence collection frequency time series data, power collection frequency time series data, speed collection frequency time series data, etc.

[0060] It should be noted that the collection and acquisition of data in the embodiments of the present invention have been authorized by relevant users, and the process does not violate relevant laws and regulations and does not violate public order and good morals.

[0061] Step S2: In each event data set, cluster analysis is performed on the frequency values at all moments in each collected frequency time series data to obtain frequency cluster clusters; in the frequency cluster clusters of different professional data, the similarities between frequency cluster clusters and the quantitative characteristics of the moments in the frequency cluster clusters are comprehensively considered to determine the synergy index between any two professional data in each event data set, which is used to perform cluster analysis on the professional data to obtain data cluster clusters.

[0062] In each event dataset, we can first cluster the time series frequency values of each professional data point, dividing them into different frequency patterns (such as high frequency, low frequency, or specific fluctuation phases), thereby revealing the state characteristics of each professional data point in different time periods. By comparing the frequency clusters of different professional data points, evaluating their similarities and quantitative characteristics of the moments, we can quantify the synergy between any two professional data points in the event dataset. For example, heart rate and speed may increase synchronously during the sprint phase, forming a synergistic cluster. Then, based on the synergy indicators, the professional data points are clustered into data clusters to reduce redundant information and provide structured multi-source data grouping for subsequent analysis.

[0063] First, in each event data set, the frequency values of the collected frequency time series data of each professional data are clustered to identify the frequency pattern of each professional data and obtain the frequency clustering clusters.

[0064] Preferably, in one embodiment of the present invention, the method for obtaining frequency clusters includes:

[0065] In the collection frequency time series data of each professional data in each event data set, the DBSCAN clustering algorithm is used to perform cluster analysis on the frequency values at all times to obtain frequency cluster clusters; among them, the neighborhood radius and the minimum number of points in the neighborhood are both preset values.

[0066] It should be noted that in this embodiment of the present invention, the neighborhood radius is set to 0.3, and the minimum number of points in the neighborhood is set to 5. The specific values can be adjusted according to the implementation scenario and are not limited here. The DBSCAN clustering algorithm is a well-known technology and the specific process will not be repeated here.

[0067] After obtaining the frequency clusters corresponding to each professional data, since the athlete's state will change dynamically as the competition progresses (such as warm-up, sprint, and recovery), different professional data may be difficult to align directly due to differences in collection frequency or timestamps; therefore, based on the frequency pattern, the covariation pattern between different professional data can be analyzed, and the implicit correlation between professional data can be discovered. The synergy index can be obtained to help determine which professional data has a more consistent frequency distribution pattern, enhance the robustness of state recognition, and achieve soft alignment at the pattern level to avoid the error of hard time matching.

[0068] Therefore, in the frequency clusters of different professional data, the similarities between the frequency clusters and the quantitative characteristics of the moments in the frequency clusters can be integrated to determine the synergy index between any two professional data in each event data.

[0069] Preferably, in one embodiment of the present invention, the method for obtaining the collaboration indicator includes:

[0070] See also Figure 2 , which shows a flow chart of a method for obtaining a collaborative indicator in one embodiment of the present invention, the method comprising the following steps:

[0071] Step S201: In each event data set, for any two professional data, the frequency clustering clusters in the two professional data are combined in pairs to obtain all non-repeated combined clusters.

[0072] For example, in a certain event set, professional data a has three frequency clusters 1, 2, and 3, and professional data b has two frequency clusters 4 and 5. The final combination clusters are 6, namely (1, 4), (1, 5), (2, 4), (2, 5), (3, 4), and (3, 5).

[0073] Step S202: In each combined cluster, the similarity of frequency distribution and time between frequency clusters is analyzed to determine the frequency overlap index between two frequency clusters.

[0074] The similarity of frequency patterns between professional data can be characterized by the similarity between frequency clustering clusters.

[0075] First, if the frequency distributions of two frequency clusters are closer, then it can be intuitively reflected that the frequency patterns between the two frequency clusters are more similar. Therefore, in each combined cluster, the mean of all frequency values in each frequency cluster is used as the mean eigenvalue to reflect the average level of the frequency distribution in each frequency cluster. The absolute value of the difference between the mean eigenvalues of the two frequency clusters in each combined cluster is calculated. The smaller the absolute value of the difference, the more similar the frequency distributions between the two frequency clusters. Therefore, the absolute value of the difference is negatively correlated and normalized to achieve logical relationship correction, thereby obtaining the first similarity factor between the two frequency clusters in each combined cluster. The larger the first similarity factor, the higher the similarity of the frequency distributions between the two frequency clusters. The negative correlation mapping and normalization here can be performed using the formula exp(-x), where exp() represents an exponential function with the natural constant e as the base, and x represents the independent variable.

[0076] Then, if the time distribution of the frequency values between the two frequency clustering clusters is more similar, it means that the two frequency clustering clusters are closer in time distribution, and the time overlap will be higher. Therefore, in each combined cluster, the proportion of the number of frequency values in the two frequency clustering clusters that are the same at the time is calculated as the second similarity factor between the two frequency clustering clusters in each combined cluster. Specifically, the time overlap of the frequency values in the two frequency clustering clusters is compared, and the number of overlapping (same) moments is used as the numerator, and the sum of the number of frequency values in the two frequency clustering clusters is used as the denominator. The obtained ratio is used as the second similarity factor. The larger the second similarity factor, the higher the time overlap between the two frequency clustering clusters in the combined cluster.

[0077] Finally, in each combined cluster, the product of the first and second similarity factors between the two frequency clusters is normalized to obtain the frequency overlap index between the two frequency clusters. A larger frequency overlap index indicates more similar frequency distributions between the two frequency clusters in the combined cluster. Normalization is a technique well known to those skilled in the art, and the normalization function can be linear normalization or standard normalization, among others. The specific normalization method is not limited here.

[0078] So far, in each event data set, for any two professional data, there is a frequency overlap index between any two frequency clusters under these two professional data.

[0079] Step S203: In each event data set, select any one type of professional data as the target data. In the target data and any remaining professional data, based on the frequency overlap index between the target data and the frequency clustering clusters in the data, and combined with the number of moments in the frequency clustering clusters in the target data, determine the similarity index between the target data and the professional data.

[0080] Based on the analysis of the previous steps, it can be seen that the larger the frequency overlap index between the two clusters, the closer the frequency changes are, and the more likely it is that the sensor acquisition requirements have changed due to changes in the scene during the event stage, which in turn has caused a coordinated change in the sensor acquisition frequency. Therefore, based on the frequency overlap index between the frequency clusters, the similarity index between the professional data can be further determined to reflect the degree of coordinated change.

[0081] For the convenience of explanation and illustration, in each event data set, any one professional data is selected as the target data. Under the target data and any remaining professional data, each frequency cluster in the target data will correspond to multiple combination clusters. Take the example in step S201 as an example: professional data a is the target data. In the target data a and professional data b, the combination clusters corresponding to the frequency cluster 1 in the target data a are (1, 4) and (1, 5).

[0082] Then, among all the combined clusters corresponding to each frequency cluster in the target data, the maximum value of the frequency overlap index is used as the frequency domain overlap of each frequency cluster in the target data, which is used to reflect the maximum degree of similar frequency changes. At this time, each frequency cluster in the target cluster corresponds to a frequency domain overlap.

[0083] Finally, based on the number of moments in the frequency clusters in the target data, the frequency domain overlap corresponding to the frequency clusters is weighted and fused to obtain the similarity index between the target data and the professional data. Specifically: the normalized value of the number of moments in each frequency cluster in the target cluster is used as the weight factor. The larger the weight factor, the more moments in the cluster. Then, the reference degree for analyzing and obtaining similarity indicators on this basis will be higher, and thus the accuracy will be improved. The product of the weight factor of each frequency cluster and the frequency domain overlap is used as the weighted overlap. The larger the weighted overlap, the higher the similarity between the target data and the professional data under the frequency cluster. Then, the normalized value of the mean of the weighted overlap of all frequency clusters in the target data is used as the similarity index between the target data and the professional data. It is easy to understand that the larger the similarity index, the greater the similarity between the frequency distribution, change and other characteristics between the target data and the professional data. Normalization is a technical means well known to those skilled in the art. The normalization function may be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0084] Step S204: For any two types of professional data, analyze the similarity index between the two types of professional data and determine the synergy index between the two types of professional data.

[0085] Based on the above steps, it can be seen that for any two professional data sets, there will be two similarity indicators between the two professional data sets. Therefore, the average value of the similarity indicators between the two professional data sets is normalized and used as the synergy indicator between the two professional data sets. Normalization is a technical means well known to those skilled in the art. The normalization function can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0086] At this point, in each event data set, the synergy index between professional data can be obtained based on the frequency change of professional data. Then, based on this index, the professional data can be clustered and analyzed to obtain data clusters, so that professional data with the same frequency change characteristics are grouped into a data cluster for easy subsequent integration and storage.

[0087] Preferably, in one embodiment of the present invention, the method for obtaining data clusters includes:

[0088] In each event data set, the value after negative correlation mapping of the synergy indicators between any two professional data is used as the difference indicator between the two professional data. The negative correlation mapping and normalization processing here can be performed using the formula exp(-x), where exp() represents an exponential function with the natural constant e as the base, and x represents the independent variable.

[0089] Then, all professional data in each event data set are clustered based on the K-means clustering algorithm and the optimal K value to obtain data clusters; among them, the distance metric is the difference index between professional data; the optimal K value can be obtained based on the elbow method.

[0090] It should be noted that the K-means clustering algorithm and the elbow method are both well-known technologies, and the specific processes are not described here in detail.

[0091] Step S3: In the data clusters of each event data set, based on the matching between the frequency clusters of the professional data and combined with the storage space occupied by the professional data, the write demand of each data cluster is determined.

[0092] Based on the above steps, highly collaborative professional data can be clustered together. However, when these highly collaborative professional data are stored together, due to changes in the event scene, these highly collaborative fields may experience a simultaneous surge in collection frequency, which can lead to excessive server bandwidth usage at certain times, or even an inability to meet normal write requirements. Therefore, when merging and storing professional data, the write demand for professional data in the data cluster during the event should be considered, and the professional data in the data cluster should be split, integrated, and stored based on the write demand.

[0093] Therefore, in the data clusters of each event data set, the writing demand of each data cluster is determined based on the matching between the frequency clusters of professional data and the storage space occupied by professional data.

[0094] Preferably, in one embodiment of the present invention, the method for obtaining the write demand includes:

[0095] In each data cluster, the product of the mean of all frequency values in each frequency cluster in each professional data and the storage occupied space of each professional data is used as the write factor of each frequency collection cluster in each professional data. The mean of the frequency value is used to reflect the data collection density, and the storage occupied space reflects the amount of data written in a single time. The write obtained by combining the two directly reflects the unit storage pressure of the frequency cluster.

[0096] In each data cluster, one kind of professional data is selected as the test data, and one frequency cluster is selected in the test data as the test cluster.

[0097] In each data cluster, for any frequency cluster under a certain professional data, the more likely the frequency cluster in other professional data that is most similar to this frequency cluster is to be in the same competition stage, the greater the probability that they will be written at the same time, and the writing pressure will also increase.

[0098] Therefore, in each data cluster, between the data to be tested and any remaining professional data, the frequency cluster with the largest frequency overlap index with the cluster to be tested in the professional data is used as the matching cluster of the cluster to be tested. In the data to be tested and this type of professional data, the sum of the write factors of the cluster to be tested and the matching cluster is used as the write sum. The larger the write sum, the greater the bandwidth when the cluster to be tested and the matching cluster are written simultaneously, and the greater the write pressure. When the frequency overlap index between the cluster to be tested and the matching cluster is larger, it means that the two clusters are more likely to be clusters of the same event stage, and the simultaneous writing is more likely to occur, and the write pressure will also increase. Therefore, the normalized value of the write sum and the product of the frequency overlap index between the cluster to be tested and the matching cluster is used as the write index between the cluster to be tested and the matching cluster. The larger the write index, the greater the write demand when the cluster to be tested and the matching cluster are written simultaneously. Normalization is a technical means well known to those skilled in the art. The normalization function may be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0099] Throughout the competition, the acquisition frequency is determined by the moment of maximum data write bandwidth. If the maximum bandwidth for writing a specific data type can be met, then normal write requirements can be met. Therefore, the simultaneous write requirements for two specific data types are measured based on the maximum write index of the frequency clusters between the two specific data types. Therefore, for the test data and any other specific data type, the maximum write index of all frequency clusters in the test data is used as the simultaneous write requirement factor between the test data and that specific data type.

[0100] At this time, in each data cluster, for the data to be tested, it has a simultaneous write demand factor with other professional data. Then the average of the simultaneous write demand factors between the data to be tested and all remaining professional data is used as the write demand index corresponding to the data to be tested. The write demand index reflects the collaborative write demand between the data to be tested and all professional data with high synergy. The larger the value, the greater the simultaneous write pressure.

[0101] At this point, in the data cluster, each type of professional data has a write demand indicator. Based on the aforementioned theory that if the maximum bandwidth for professional data writing can be met, normal write demand can be met, the maximum write demand indicator of all professional data in each data cluster is used as the write demand of each data cluster.

[0102] Step S4: among all the event data sets of all kinds of sports events, all the event data sets are integrated and stored based on the write demand of the data clusters.

[0103] Based on the above steps, the write demand of each data cluster can be obtained. In this step, all event data of all types of sports events can be integrated and stored based on the write demand of the data cluster.

[0104] Preferably, in one embodiment of the present invention, the process of integrating and storing all event data includes:

[0105] Given that the overall write demand for professional data in some data clusters is relatively small, normal write demand can be met without server distributed storage. Therefore, all data clusters need to be divided into clusters that require distributed storage and clusters that do not require distributed storage.

[0106] First, the data clusters in all data sets of all sports events are sorted in ascending order according to the write demand to obtain a sorted sequence. In the sorted sequence, the write demand of the data cluster is calculated by first-order difference to obtain a differential sequence. The value in the differential sequence represents the difference between the write demand of two adjacent data clusters; the maximum value in the differential sequence is selected. The maximum value represents the position where the write demand of the data cluster has the largest change. Therefore, the latter data cluster of the two data clusters corresponding to the maximum value in the sorted sequence is used as the breakpoint.

[0107] In the sorting sequence, the data clusters at and after the breakpoint have a large write demand, so the data is stored as a distribution; the data clusters before the breakpoint have a small write demand, so the data is stored as a whole.

[0108] Then, based on the K-means clustering algorithm and the preset K value, cluster analysis is performed on the data clusters in the distributed storage data to obtain distributed clusters, where the distance metric is the absolute value of the difference in write demand between data clusters, and the preset K value is the number of servers.

[0109] Then, the sum of the write requirements of all data clusters in each distribution cluster is used as the storage requirement value of each distribution cluster, and all distribution clusters are sorted in descending order according to the storage requirement value to obtain a storage requirement sequence; at the same time, all servers are sorted in descending order according to storage capacity to obtain a storage capacity sequence; each distribution cluster in the storage requirement sequence is stored in the server at the same position in the storage capacity sequence.

[0110] Finally, the storage capacity of each server is updated, and the overall storage data is stored in the server with the largest storage capacity.

[0111] In summary, given that the frequency of collecting professional data varies between different events, and even between different schedules for the same event, storing it in a relational database can lead to null value issues. This null value problem is primarily due to data collection frequency mismatches. Therefore, for each sporting event, we first obtain a data set for each athlete, including time series data on the collection frequencies of various professional data and the storage space occupied by each type of professional data. Then, within each event data set, we perform a cluster analysis on the frequency values in the time series data for each type of professional data, generating frequency clusters that can identify the collection frequency characteristics of each type of professional data. Furthermore, within the frequency clusters of different professional data, we calculate a synergy index based on the similarity and moment quantity characteristics of the frequency clusters. We then perform cluster analysis on the professional data to generate data clusters. This accurately quantifies the correlation between professional data within each event data set, reduces the mixed storage of low-correlation data in subsequent processes, and optimizes the null value problem in relational database storage. Furthermore, in order to avoid the problem of excessive server pressure caused by the simultaneous writing of a large amount of data in a certain stage of a highly collaborative data set, the present invention combines the matching between the frequency clustering clusters of professional data with the storage space occupied by professional data in the data clustering cluster of the event data set, and determines the write demand of each data clustering cluster, which is used to avoid the problem of local storage overload in the subsequent storage process. Finally, the integration of all event data sets of all types of sports events is driven by the write demand of the data clustering cluster, and all event data sets are stored. The embodiment of the present invention clusters professional data by analyzing the synergy of the changes in the collection frequency of different types of professional data during the event, thereby reducing the null value problem in the storage process of the relational database, and analyzing the storage space occupied by professional data, effectively reducing the write pressure on the server.

[0112] The embodiment of the present invention also provides a data integration system for multiple sports events. Figure 3 , which shows a system block diagram, including a data acquisition module 301, used to implement step S1 in the above method embodiment; a clustering analysis module 302, used to implement step S2 in the above method embodiment; a write demand analysis module 303, used to implement step S3 in the above method embodiment; and an integrated storage module 304, used to implement step S4 in the above method embodiment.

[0113] It should be noted that the system provided in the above embodiment is merely exemplified by the division of the aforementioned functional modules. In actual applications, the aforementioned functions can be distributed among different functional modules as needed, i.e., the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. Furthermore, the data integration system for multi-category sports events provided in the above embodiment and the data integration method for multi-category sports events are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be further described here.

[0114] See also Figure 4 , which shows a system structure diagram of a data integration system for multi-category sports events provided by an embodiment of the present invention, including a processor 400, a memory 401, a bus 402 and a communication interface 403, wherein the processor 400, the communication interface 403 and the memory 401 are connected via the bus 402; wherein the memory 401 may include a high-speed random access memory, the bus 402 may be an ISA bus, a PCI bus or an EISA bus, etc., and the processor 400 may be an integrated circuit chip with signal processing capabilities; the memory 401 stores at least one instruction, at least one program, a code set or an instruction set, and when the at least one instruction, at least one program, a code set or an instruction set is loaded and executed by the processor, the steps in a data integration method for multi-category sports events are implemented.

[0115] The embodiment of the present invention also provides a computer-readable storage medium corresponding to the method provided in the above embodiment, see Figure 5 , the storage medium shown is a CD, on which a computer program (ie, a program product) is stored. When the computer program is run by a processor, it will execute the method provided by any of the aforementioned embodiments.

[0116] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), read-only memory (ROM), and other optical and magnetic storage media, which are not listed here one by one.

[0117] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0118] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A data integration method for multi-category sports events, characterized in that: The method comprises: In each sports event, obtain a match data set for each athlete, wherein the match data set includes time series data of the acquisition frequency of various professional data and the storage space occupied by each professional data; In each event data set, cluster analysis is performed on the frequency values at all moments in each collected frequency time series data to obtain frequency clusters. In the frequency clusters of different professional data, the similarities between frequency clusters and the quantitative characteristics of the moments in the frequency clusters are comprehensively considered to determine the synergy index between any two professional data in each event data set. This is used to perform cluster analysis on the professional data to obtain data clusters. In the data clusters of each event data set, the write demand of each data cluster is determined based on the matching between the frequency clusters of professional data and the storage space occupied by professional data; All event data sets of all types of sports events are integrated and stored based on the write demand of data clusters.

2. A data integration method for multi-category sports events according to claim 1, characterized in that: The method for obtaining the collaborative indicator includes: In each event data set, for any two professional data, the frequency clusters in the two professional data are combined in pairs to obtain all non-repeated combination clusters; In each combined cluster, the similarity of frequency distribution and time between frequency clusters is analyzed to determine the frequency overlap index between the two frequency clusters; In each event data set, one professional data is selected as the target data. Under the target data and any other professional data, among all the combination clusters corresponding to each frequency cluster in the target data, the maximum value of the frequency overlap index is used as the frequency domain overlap degree of each frequency cluster in the target data; Based on the number of moments in the frequency clusters in the target data, the frequency domain overlap corresponding to the frequency clusters is weightedly fused to obtain the similarity index between the target data and the professional data; For any two professional data, the normalized value of the mean of the similarity index between the two professional data is used as the synergy index between the two professional data.

3. The data integration method for multi-category sports events according to claim 2, characterized in that: The method for obtaining the frequency overlap index includes: In each combined cluster, the mean of all frequency values in each frequency cluster is used as the mean eigenvalue; The absolute value of the difference between the mean eigenvalues of the two frequency clusters in each combination cluster is negatively correlated and normalized, and is used as the first similarity factor between the two frequency clusters in each combination cluster; Compare the moments of the frequency values in the two frequency clusters in each combination cluster, and use the proportion of the same moments as the second similarity factor between the two frequency clusters in each combination cluster; The value obtained by normalizing the product of the first similarity factor and the second similarity factor is used as the frequency overlap index between the two frequency clustering clusters in each combined cluster.

4. The data integration method for multi-category sports events according to claim 2, characterized in that: The method for obtaining the write demand includes: In each data cluster, the product of the mean of all frequency values in each frequency cluster in each professional data and the storage space occupied by each professional data is used as the write factor of each frequency cluster in each professional data; In each data cluster, select one professional data as the test data, and select one frequency cluster in the test data as the test cluster; In each data cluster, between the data to be tested and any remaining professional data, the frequency cluster with the largest frequency overlap index with the cluster to be tested in the professional data is used as the matching cluster of the cluster to be tested; in the data to be tested and the professional data, the sum of the write factors of the cluster to be tested and the matching cluster is normalized by the product of the frequency overlap index between the cluster to be tested and the matching cluster, and the result is used as the write index between the cluster to be tested and the matching cluster; Under the conditions of the data to be tested and any other professional data, the maximum value of the write index of all frequency clusters in the data to be tested is used as the simultaneous write demand factor between the data to be tested and the professional data; In each data cluster, the mean of the simultaneous write demand factors between the data to be tested and all other professional data is used as the write demand indicator corresponding to the data to be tested; In each data cluster, the maximum value of all write demand indicators of all professional data is taken as the write demand of each data cluster.

5. The data integration method for multi-category sports events according to claim 1, characterized in that: The method for obtaining the frequency clusters includes: In the collection frequency time series data of each professional data in each event data set, the DBSCAN clustering algorithm is used to perform cluster analysis on the frequency values at all times to obtain frequency cluster clusters; among them, the neighborhood radius and the minimum number of points in the neighborhood are both preset values.

6. The data integration method for multi-category sports events according to claim 1, characterized in that: The method for obtaining the data clusters includes: In each event data set, the value after negative correlation mapping of the synergy index between any two professional data is used as the difference index between the two professional data; Based on the K-means clustering algorithm and the optimal K value, all professional data in each event data set are clustered to obtain data clusters; among them, the distance metric is the difference index between professional data.

7. The data integration method for multi-category sports events according to claim 1, characterized in that: In all event data sets of all kinds of sports events, all event data sets are integrated and stored based on the write demand of data clusters, including: Sort the data clusters in all data sets of all sports events in ascending order according to the write demand to obtain a sorted sequence; In the sorting sequence, first-order difference calculation is performed on the write demand of the data clusters to obtain a difference sequence, and the latter of the two data clusters corresponding to the maximum value in the difference sequence in the sorting sequence is used as a breakpoint; In the sorting sequence, the data clusters at and after the breakpoint are used as distributed storage data, and the data clusters before the breakpoint are used as overall storage data; Perform cluster analysis on the data clusters in the distributed storage data based on the K-means clustering algorithm and the preset K value to obtain distributed clusters, where the distance metric is the absolute value of the difference in write demand between data clusters, and the preset K value is the number of servers; The sum of the write requirements of all data clusters in each distributed cluster is used as the storage requirement value of each distributed cluster. All distributed clusters are sorted in descending order according to the storage requirement value to obtain a storage requirement sequence. All servers are sorted in descending order according to storage capacity to obtain a storage capacity sequence. Each distributed cluster in the storage requirement sequence is stored in the server at the same position in the storage capacity sequence. Update the storage capacity of each server and store the overall storage data in the server with the largest storage capacity.

8. The data integration method for multi-category sports events according to claim 6, characterized in that: The method for obtaining the optimal K value includes: The optimal K value is obtained based on the elbow method.

9. A data integration system for multi-category sports events, characterized by: The method comprises a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and when the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor, the steps of a data integration method for multi-category sports events as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the data integration method for multi-category sports events as claimed in any one of claims 1 to 8 are implemented.

Images