Data management method and system for automobile trading platform

Abstract

The present invention relates to the technical field of data storage management, and in particular to a data management method and system for an automobile trading platform. The method comprises the following steps: obtaining a data sequence to be stored corresponding to the automobile trading platform and performing index pair classification extraction and storage space limit quantitative calculation to obtain the storage release memory space limit corresponding to each type of index field; performing sequence storage division on the corresponding data sequence to be stored to obtain the storage data sequence segment corresponding to each type of index field; obtaining the query access rate of the corresponding sequence segment and the storage interval proximity access degree and performing storage activity evaluation analysis and index storage priority measurement analysis to obtain the index storage priority measurement of each type of index field for each sequence segment; and performing priority storage management on the corresponding storage data sequence segment. The present invention can efficiently and accurately manage the massive data in the automobile trading platform, thereby optimizing the data storage process.

Description

Technical Field

[0001] The present invention relates to the technical field of data storage management, and in particular to a data management method and system for an automobile trading platform. Background Art

[0002] The automobile trading platform involves a large amount of user data, vehicle information, transaction records, and payment information. These data usually come from different channels and systems, and have different formats and structures. How to efficiently integrate, store, and manage this data has become a major problem in platform operations. At the same time, by quantifying the storage space requirements of each type of data and evaluating the storage priority based on the data, it is ensured that high-frequency query data can be stored in higher-performance storage areas first, and low-frequency data can be managed using lower-performance storage media, thereby significantly improving the flexibility and efficiency of storage management. However, in the prior art, traditional data management methods usually adopt simple storage methods, which are difficult to cope with complex and diverse data access requirements, resulting in low storage efficiency and waste of storage space in large-scale data storage and high-frequency query environments. Summary of the invention

[0003] Based on this, it is necessary for the present invention to provide a data management method and system for an automobile trading platform to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a data management method for an automobile trading platform includes the following steps:

[0005] Step S1: obtaining a data sequence to be stored corresponding to the automobile trading platform, and classifying and extracting index pairs of the data sequence to be stored to obtain each type of index pair field; performing quantitative calculation of the storage space limit according to each type of index pair field to obtain the storage release memory space limit degree corresponding to each type of index field; performing sequence storage division on the corresponding data sequence to be stored based on the storage release memory space limit degree corresponding to each type of index field to obtain the storage data sequence segment corresponding to each type of index field;

[0006] Step S2: according to the stored data sequence segments corresponding to each type of index field, the query access rate and the storage interval proximity access degree of the corresponding sequence segments are obtained, and based on the query access rate and the storage interval proximity access degree, the storage activity evaluation analysis is performed on the stored data sequence segments corresponding to each type of index field to obtain the storage activity factor of each type of index field for each sequence segment; according to the storage activity factor of each type of index field for each sequence segment, the corresponding stored data sequence segments are subjected to index storage priority metric analysis to obtain the index storage priority metric of each type of index field for each sequence segment;

[0007] Step S3: Prioritize storage management of corresponding stored data sequence segments according to the index storage priority metric of each type of index field for each sequence segment.

[0008] Further, step S1 includes the following steps:

[0009] Step S11: obtaining a data sequence to be stored corresponding to the automobile transaction platform, wherein the data sequence to be stored includes a user data subsequence, a vehicle data subsequence, and an automobile transaction data subsequence;

[0010] Step S12: classifying the hierarchical relationship index fields for each type of data sub-column corresponding to the data sequence to be stored, so as to obtain a multi-dimensional index field cluster corresponding to each type of data sub-column;

[0011] Step S13: performing semantic index pair refinement and optimization on the multi-dimensional index field cluster corresponding to each type of data sub-column to obtain each type of index pair field, wherein each type of index pair field includes user ID, geographic location, browsing history and transaction history user index pair fields, vehicle brand, model, age and price vehicle index pair fields and transaction time, transaction amount and transaction parties' automobile transaction index pair fields;

[0012] Step S14: performing quantitative calculation of storage space limit for each type of index field to obtain storage release memory space limit corresponding to each type of index field;

[0013] Step S15: By taking the storage release memory space limit corresponding to each type of index field as the corresponding sequence storage memory space size, and dividing the corresponding data sequence to be stored based on the sequence storage memory space size corresponding to each type of index field, the storage data sequence segment corresponding to each type of index field is obtained.

[0014] Further, step S13 includes the following steps:

[0015] Step S131: performing entity extraction processing on each index field in the multi-dimensional index field cluster corresponding to each type of data sub-column, to obtain index field entities corresponding to each type of data sub-column index field cluster;

[0016] Step S132: Based on the index field entity corresponding to each type of data sub-column index field cluster, a word vector embedding process is performed on each corresponding index field in the multi-dimensional index field cluster corresponding to each type of data sub-column, so as to obtain an index field embedding word vector corresponding to each type of data sub-column index field cluster;

[0017] Step S133: performing contextual semantic reasoning analysis on the index field embedding word vectors corresponding to each type of data sub-column index field cluster to obtain the semantic relationship between each index field in each type of data sub-column index field cluster;

[0018] Step S134: Based on the semantic relationship between each index field in each type of data sub-column index field cluster, semantic index pair refinement and optimization are performed on each corresponding index field in the multi-dimensional index field cluster corresponding to each type of data sub-column, and semantically repeated or irrelevant index fields are removed to obtain each type of index pair field, wherein each type of index pair field includes user ID, geographic location, browsing history and transaction history user index pair fields, vehicle brand, model, age and price vehicle index pair fields and transaction time, transaction amount and transaction parties' automobile transaction index pair fields.

[0019] Further, step S14 includes the following steps:

[0020] Step S141: Analyze the index type and distribution characteristics of each type of index field to obtain the index data type and index distribution characteristics corresponding to each type of index field;

[0021] Step S142: determining the field element size of the corresponding index field in each type of index pair field based on the index data type and index distribution characteristics corresponding to each type of index field, and obtaining the element distribution size corresponding to each type of index field; predicting the index space occupancy of the corresponding index field in each type of index pair field according to the element distribution size corresponding to each type of index field, and obtaining the space occupancy requirement corresponding to each type of index field;

[0022] Step S143: Analyze the field memory usage and access frequency of each type of index field to obtain the memory usage and index access frequency corresponding to each type of index field;

[0023] Step S144: based on the memory usage corresponding to each type of index field and the index access frequency, memory space recovery prediction is performed for the corresponding index field in each type of index pair field to obtain the memory space recovery and release prediction amount corresponding to each type of index field;

[0024] Step S145: Based on the space occupation demand corresponding to each type of index field and the predicted amount of memory space recovery and release, the storage space limit of the corresponding index field in each type of index field is quantitatively calculated to obtain the storage release memory space limit corresponding to each type of index field.

[0025] Further, step S144 includes the following steps:

[0026] Based on the memory usage corresponding to each type of index field, memory space consumption analysis is performed on the corresponding index fields in each type of index field to obtain the memory space consumption corresponding to each type of index field;

[0027] Based on the index access frequency corresponding to each type of index field, the access heat evaluation and analysis of the corresponding index field in each type of index field is performed to obtain the index access heat corresponding to each type of index field;

[0028] Obtain the access time period corresponding to each type of index field, and perform access heat decay analysis on the index access heat corresponding to each type of index field based on the access time period corresponding to each type of index field, so as to obtain the access heat decay rate corresponding to each type of index field;

[0029] Based on the access heat decay rate corresponding to each type of index field, the memory recycling timing of the corresponding index field in each type of index pair field is determined to obtain the optimal index memory recycling timing corresponding to each type of index field;

[0030] According to the memory space consumption corresponding to each type of index field and the optimal time for index memory recovery, memory space recovery prediction is performed on the corresponding index fields in each type of index field to obtain the predicted amount of memory space recovery and release corresponding to each type of index field.

[0031] Furthermore, the storage space limit quantitative calculation is performed by using a storage space limit calculation formula, wherein the storage space limit calculation formula is specifically:

[0032] ;

[0033] In the formula, For the The storage release memory space limit corresponding to the class index field, is the total number of index field types, The item index of the index field type. is the time range parameter, is the time variable parameter, For the The space required for the class index field. For the The predicted amount of memory space to be reclaimed and released corresponding to the class index field. For the The memory consumption decay rate corresponding to the class index field, For the The storage space reuse efficiency coefficient corresponding to the class index field, For the Class index field at time The amount of memory released at A correction factor for the extent to which memory space is freed for storage.

[0034] Further, step S2 includes the following steps:

[0035] Step S21: obtaining the historical query log of the automobile trading platform data, and determining the query time period and access frequency of the stored data sequence segment corresponding to each type of index field based on the historical query log of the automobile trading platform data, and obtaining the query access time period and query access frequency of the sequence segment corresponding to each type of index field;

[0036] Step S22: performing a quantitative calculation of the query access rate according to the query access time period and the query access frequency of the sequence segment corresponding to each type of index field, and obtaining the query access rate of the sequence segment corresponding to each type of index field;

[0037] Step S23: performing query access interval analysis on the corresponding stored data sequence segments based on the query access time period of the sequence segments corresponding to each type of index field, so as to obtain the query access time interval of the sequence segments corresponding to each type of index field; obtaining the storage time interval of the sequence segments corresponding to each type of index field, and performing storage proximity access degree calculation based on the query access time interval and storage time interval of the sequence segments corresponding to each type of index field, so as to obtain the storage interval proximity access degree of the sequence segments corresponding to each type of index field;

[0038] Step S24: performing storage activity evaluation analysis on the stored data sequence segments corresponding to each type of index field based on the query access rate and the storage interval proximity access degree, so as to obtain the storage activity factor of each type of index field for each sequence segment;

[0039] Step S25: performing index storage priority metric analysis on the corresponding stored data sequence segments according to the storage activity factor of each type of index field for each sequence segment, and obtaining the index storage priority metric of each type of index field for each sequence segment.

[0040] Furthermore, the calculation of the storage proximity access degree according to the query access time interval and the storage time interval of the sequence segment corresponding to each type of index field in step S23 includes the following steps:

[0041] Synchronously draw the query access time interval and storage time interval of the sequence segment corresponding to each type of index field on the time axis to generate a time axis distribution diagram between the query time interval and storage time interval corresponding to each index sequence segment;

[0042] Dividing the time axis distribution diagram between the query time interval and the storage time interval corresponding to each index sequence segment by the repetition period between query and storage, so as to obtain the repetition period between the query interval and the storage interval corresponding to each index sequence segment;

[0043] Based on the repetitive period between the query interval and the storage interval corresponding to each index sequence segment, query the query access time interval and the storage time interval of the corresponding sequence segment and determine the storage overflow period, and obtain the query overflow period and the storage overflow period corresponding to each index sequence segment;

[0044] The storage proximity access degree is calculated for the query overflow period and the storage overflow period corresponding to each index sequence segment according to the repeated period between the query interval and the storage interval corresponding to each index sequence segment, and the storage interval proximity access degree of the sequence segment corresponding to each type of index field is obtained.

[0045] Further, step S24 includes the following steps:

[0046] Step S241: performing query access heat distribution analysis on the stored data sequence segments corresponding to each type of index field based on the query access rate to generate a query access heat distribution graph corresponding to each type of index field;

[0047] Step S242: Obtain the corresponding query access behavior cycle according to the query access heat distribution diagram corresponding to each type of index field, and perform access fluctuation range analysis on the query access rate based on the query access behavior cycle to obtain the query access fluctuation range corresponding to each type of index field in each access behavior cycle;

[0048] Step S243: performing storage access density analysis on the storage data sequence segments corresponding to each type of index field based on the query access fluctuation amplitude corresponding to each type of index field in each access behavior cycle, and obtaining the storage query access density corresponding to each type of index field;

[0049] Step S244: Perform storage activity evaluation analysis on the storage data sequence segments corresponding to each type of index field based on the storage query access density and storage interval proximity access degree corresponding to each type of index field, so as to obtain the storage activity factor of each type of index field for each sequence segment.

[0050] Furthermore, the present invention also provides a data management system for an automobile trading platform, which is used to execute the data management method for the automobile trading platform as described above. The data management system for the automobile trading platform includes:

[0051] The storage partitioning module of the data sequence to be stored is used to obtain the data sequence to be stored corresponding to the automobile trading platform, and classify and extract the index pairs of the data sequence to be stored to obtain each type of index pair field; perform quantitative calculation of the storage space limit according to each type of index pair field to obtain the storage release memory space limit degree corresponding to each type of index field; perform sequence storage partitioning on the corresponding data sequence to be stored based on the storage release memory space limit degree corresponding to each type of index field, so as to obtain the storage data sequence segmentation corresponding to each type of index field;

[0052] A storage sequence segmentation priority measurement module is used to obtain the query access rate and storage interval proximity access degree of the corresponding sequence segment according to the storage data sequence segment corresponding to each type of index field, and perform storage activity evaluation analysis on the storage data sequence segment corresponding to each type of index field based on the query access rate and storage interval proximity access degree to obtain the storage activity factor of each type of index field for each sequence segment; perform index storage priority measurement analysis on the corresponding storage data sequence segment according to the storage activity factor of each type of index field for each sequence segment, so as to obtain the index storage priority measurement of each type of index field for each sequence segment;

[0053] The sequence segment priority storage management module is used to perform priority storage management on the corresponding stored data sequence segments according to the index storage priority metric of each type of index field for each sequence segment.

[0054] Beneficial effects of the present invention:

[0055] 1. Compared with the prior art, the data management method for the automobile trading platform proposed in the present invention has the beneficial effect of obtaining the data sequence to be stored corresponding to the automobile trading platform. The main goal of this step is to obtain the data to be stored from the automobile trading platform and organize the data according to different categories. The data to be stored includes user data, vehicle data and transaction data, which can lay the foundation for subsequent indexing and storage optimization. This step provides a clear data structure and classification for the effective processing of data, subsequent classification and storage operations, which helps to reduce redundancy and improve storage and retrieval efficiency. By indexing and classifying the data sequence to be stored, the original data can be effectively divided according to its hierarchical structure. For example, for user data subsequences, they can be indexed according to different dimensions such as region and user activity; for vehicle data subsequences, they can be classified according to brand, model, price range, etc., and semantic indexing and optimization can be performed on the index fields corresponding to each type of data sub-column. The main goal of this step is to form more accurate and efficient index fields by semantically refining and optimizing the index fields of various types of data. For example, in the processing of user data, the user's historical browsing records and purchasing behaviors can be further subdivided to form a more semantic index, rather than just a simple storage of the original fields; in vehicle data, by semantically refining data such as vehicle age, price, and brand, more targeted index fields can be generated to improve the relevance of data and the accuracy of retrieval, and further improve the query performance and storage utilization efficiency of the automobile trading platform. By quantitatively calculating the storage space limit of each type of index field, the memory space required for each type of data in the storage process can be reasonably estimated. The purpose of this calculation is to help the automobile trading platform to reasonably allocate memory resources, avoid data loss or storage failure caused by insufficient space during the storage process, and reduce unnecessary memory waste. At the same time, by taking the storage release memory space limit corresponding to each type of index field as the corresponding sequence storage memory space size, and dividing the corresponding data sequence to be stored based on the sequence storage memory space size corresponding to each type of index field, the storage space is reasonably divided, and the corresponding memory space is allocated to each type of data subsequence. This operation can effectively avoid excessive memory waste, and ensure that each type of data can obtain appropriate space when stored, and there will be no memory overflow or insufficient storage. It can ensure that the automobile trading platform still maintains efficient performance when dealing with large-scale data storage and high-frequency query environments, thereby avoiding storage space waste. Then, by obtaining the query access rate of the corresponding sequence segment according to the storage data sequence segment corresponding to each type of index field, the access rate of each type of index field in a specific time period is quantitatively calculated. The key to this process is that it can provide an accurate basis for the platform's data management and resource scheduling.By calculating the access rate, the platform can identify which fields are frequently queried and which fields are in a low-usage state during certain time periods. The access rate directly reflects the usage demand of certain data. Especially when facing large-scale data storage, it can accurately know which data is frequently accessed and can give priority to optimization processing, such as caching, index acceleration, etc., to further improve the overall performance and response speed of the automobile trading platform. The "storage access proximity", that is, the matching degree of data storage and query requirements, is also analyzed and calculated. If the storage time interval is too long and the query is frequent, the database cache will fail or the storage access will not be timely, affecting the query performance. On the contrary, if the storage and query time intervals are very close, it will lead to unnecessary frequent storage updates and increase the system burden. The analysis of this step can help the platform find the best balance between data storage and query, avoid resource waste and improve the system response speed. By performing storage activity evaluation analysis on the storage data sequence corresponding to each type of index field based on the query access rate and the storage interval access proximity, the "storage activity factor" is obtained. The quantification of this factor reflects the activity of data in actual applications, and then provides a basis for the platform to optimize the storage strategy. Through this evaluation, highly active and highly accessed fields or data sequences can be identified. Such data requires more storage resource support, optimized storage structure, and index acceleration. For low-activity fields, the platform can adopt resource-saving strategies, such as moderate compression storage or adjustment of its storage location to reduce unnecessary overhead. The evaluation of the storage activity factor can not only improve storage efficiency, but also improve the overall performance of the database. Next, by performing index storage priority measurement analysis on the corresponding stored data sequence segments according to the storage activity factor of each sequence segment for each type of index field, the index storage priority of each data sequence segment can be further quantified. Through this priority measurement, the platform can accurately adjust the index strategy to ensure that the most active data sequences obtain priority storage and index resources, reduce query latency, and improve system response speed. Priority measurement analysis helps the platform to dynamically allocate storage resources according to data activity, thereby better supporting large-scale data storage and efficient query. It can also reduce the storage pressure of the database, avoid unnecessary resource consumption, and improve data query performance. This step helps to intelligently manage storage resources, improve the overall storage efficiency of the automobile trading platform, and ensure that the platform can still maintain efficient operation when facing data of different scales and complexities.Finally, based on the previous evaluation of data storage activity, priority storage management can be performed according to the index storage priority metric of each sequence segment. The core goal of this process is to rationally plan the allocation of storage resources by comprehensively considering the activity, access frequency, storage requirements and hardware resource limitations of the data. Specifically, through the storage priority metric analysis of each type of index field, different storage strategies will be implemented for different data segments. For example, for those high-priority storage data, the automobile trading platform will choose a more efficient storage method (such as storing it in cache or SSD) to improve the query speed; for low-priority data, a more economical storage method (such as traditional hard disk or cheaper cloud storage service) will be selected. This can help the automobile trading platform to achieve dynamic management of storage resources and ensure that when storage capacity is tight, those frequently accessed or critical data can be retained first to avoid unnecessary data loss or performance bottlenecks. While achieving storage optimization, it can also reduce the maintenance cost of the trading platform and improve the overall storage efficiency.

[0056] 2. The data management system of the automobile trading platform proposed in the present invention is generally composed of a storage partition module for the data sequence to be stored, a storage sequence segment priority measurement module and a sequence segment priority storage management module. It can implement the data management method of any automobile trading platform described in the present invention, and is used to combine the operations between the computer programs running on each module to implement the data management method of the automobile trading platform. The internal structures of the system cooperate with each other, which can greatly reduce duplication of work and manpower investment, and can quickly and effectively provide a more accurate and efficient data management process for the automobile trading platform, thereby simplifying the operating procedures of the data management system of the automobile trading platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0058] Figure 1 A schematic diagram of the steps of the data management method of the automobile trading platform of the present invention;

[0059] Figure 2 for Figure 1 Detailed step flow diagram of step S1;

[0060] Figure 3 for Figure 2 Detailed step flow chart of step S13 in FIG. DETAILED DESCRIPTION

[0061] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0062] To achieve this, please refer to Figures 1 to 3 The present invention provides a data management method for an automobile trading platform, the method comprising the following steps:

[0063] Step S1: obtaining a data sequence to be stored corresponding to the automobile trading platform, and classifying and extracting index pairs of the data sequence to be stored to obtain each type of index pair field; performing quantitative calculation of the storage space limit according to each type of index pair field to obtain the storage release memory space limit degree corresponding to each type of index field; performing sequence storage division on the corresponding data sequence to be stored based on the storage release memory space limit degree corresponding to each type of index field to obtain the storage data sequence segment corresponding to each type of index field;

[0064] Step S2: according to the stored data sequence segments corresponding to each type of index field, the query access rate and the storage interval proximity access degree of the corresponding sequence segments are obtained, and based on the query access rate and the storage interval proximity access degree, the storage activity evaluation analysis is performed on the stored data sequence segments corresponding to each type of index field to obtain the storage activity factor of each type of index field for each sequence segment; according to the storage activity factor of each type of index field for each sequence segment, the corresponding stored data sequence segments are subjected to index storage priority metric analysis to obtain the index storage priority metric of each type of index field for each sequence segment;

[0065] Step S3: Prioritize storage management of corresponding stored data sequence segments according to the index storage priority metric of each type of index field for each sequence segment.

[0066] In the embodiment of the present invention, please refer to Figure 1 FIG. 1 is a schematic diagram of the steps of the data management method of the automobile trading platform of the present invention. In this example, the data management method of the automobile trading platform includes the following steps:

[0067] Step S1: obtaining a data sequence to be stored corresponding to the automobile trading platform, and classifying and extracting index pairs of the data sequence to be stored to obtain each type of index pair field; performing quantitative calculation of the storage space limit according to each type of index pair field to obtain the storage release memory space limit degree corresponding to each type of index field; performing sequence storage division on the corresponding data sequence to be stored based on the storage release memory space limit degree corresponding to each type of index field to obtain the storage data sequence segment corresponding to each type of index field;

[0068] In an embodiment of the present invention, by extracting user data, vehicle data and transaction data from the automobile trading platform, the user data subsequence contains basic information of each user, such as user ID, registration time, geographic location, browsing history and transaction history, etc.; the vehicle data subsequence contains basic information of each vehicle, such as vehicle brand, model, age, price, owner information, maintenance history, etc.; the transaction data subsequence contains all transaction records, including transaction time, transaction amount, transaction parties, payment method, etc. These data can be obtained by querying the platform database and filtering out the data to be stored according to a certain time range and conditions (such as user behavior, time period of transaction records), thereby obtaining the data sequence to be stored corresponding to the automobile trading platform. Each type of data subsequence in the previously extracted data sequence to be stored contains multi-dimensional information. First, the fields in these data subsequences need to be indexed. The index fields in the user data subsequence include user ID, geographic location, browsing history, transaction history, etc.; the index fields in the vehicle data subsequence include vehicle brand, model, age, price, etc.; the index fields in the transaction data subsequence include transaction time, transaction amount, transaction parties, etc., and semantic analysis and optimization are performed on the multi-dimensional index fields in each type of data subsequence to extract index pairs that are more efficient for query and storage. For example, in user data, by dividing Analyze the user's browsing history and transaction history to extract more representative indexes. For example, define the relationship between the user ID and the geographic location, browsing history, and transaction history based on the user's purchasing interests, activity level, and other characteristics; in vehicle data, optimize the query optimization index by brand, age, and other dimensions based on the brand, model, and price of the vehicle; the index fields in the transaction data generate more efficient index pairs based on the timeliness and amount characteristics of the transaction. At this time, these fields can be further semantically optimized by combining natural language processing and data mining technology to ensure that they can accurately match user queries and needs, thereby obtaining each type of index pair field. After completing the optimization of the index fields of each type of data, it is necessary to perform a quantitative calculation of the space requirements for each index field. The key to this step is to calculate the storage space required for each type of index field based on the size of the data and the actual storage requirements of each field. For example, the index fields of user ID and geographic location require smaller storage space, while fields such as transaction amount and transaction time require larger storage space. The calculation needs to consider factors such as the size of the index field, the amount of data, and the frequency of data changes. By quantifying the storage space required for these fields, storage resources can be reasonably planned to avoid wasting storage space, thereby obtaining the maximum degree of storage release memory space corresponding to each type of index field.Then, after obtaining the storage release memory space limit of each index field through previous analysis, the data to be stored is divided and stored according to these space limits. During the storage process, the storage space limit of each type of index field is matched with the size of the data sequence to be stored, and the storage space of each type of data is reasonably divided. For example, if the space requirement of the user data field is small, you can choose to store it in the memory or on a fast storage medium; while the storage requirement of transaction data or vehicle data is large, it needs to be stored in a storage device such as a disk or the cloud. When storing, it is necessary to optimize according to the limit capacity of the system memory to avoid system overload or uneven data storage, ensure efficient data storage and do not waste storage resources, and finally obtain the storage data sequence segmentation corresponding to each type of index field.

[0069] Step S2: according to the stored data sequence segments corresponding to each type of index field, the query access rate and the storage interval proximity access degree of the corresponding sequence segments are obtained, and based on the query access rate and the storage interval proximity access degree, the storage activity evaluation analysis is performed on the stored data sequence segments corresponding to each type of index field to obtain the storage activity factor of each type of index field for each sequence segment; according to the storage activity factor of each type of index field for each sequence segment, the corresponding stored data sequence segments are subjected to index storage priority metric analysis to obtain the index storage priority metric of each type of index field for each sequence segment;

[0070] In an embodiment of the present invention, by obtaining the historical query log of the automobile trading platform, the historical query log records various operations of the user on the platform, including query conditions, query time, query frequency and other information. By analyzing these query logs, the query access time period and query access frequency of the stored data sequence (such as automobile data of a specific brand or model) corresponding to each type of index field (for example, brand, model, price range, etc.) can be determined, and the query access rate can be quantitatively calculated. The query access rate can be defined as the ratio of the query frequency of a certain type of index field in a specific time period to the query frequency of the index field in all time periods. By analyzing the query access time period of each type of index field, the query access interval of the data sequence of this type is further calculated. The query access interval can be defined as the time difference between two query accesses. In the operation, firstly, the timestamp information in the historical query log is used to extract all query time points corresponding to each type of index field. Then, for each query time point, the time difference between it and the previous query time point is calculated to obtain the distribution of query access intervals. At the same time, by obtaining the storage time interval of the stored data sequence corresponding to each type of index field, the storage time interval refers to the update frequency of a data sequence in the storage system or the periodicity of data changes. Based on the query access time interval of each type of index field and the storage time interval, the storage access proximity is calculated. The storage access proximity can be defined as the proximity between the query access interval and the storage time interval. The higher the value, the more the query access matches the data storage cycle. Thus, the storage interval access proximity of the sequence segment corresponding to each type of index field is obtained. By combining the query access rate and the storage interval access proximity, the storage activity of the stored data sequence segment corresponding to each type of index field is evaluated. The storage activity reflects the frequency of data use and the importance of access, so as to calculate the activity factor of each data segment. The higher the value of the storage activity factor, the higher the storage activity of the data. Thus, the storage activity factor of each type of index field for each sequence segment is obtained. Then, by combining the storage activity factor of each sequence segment obtained from the previous analysis, the corresponding storage data sequence segment is subjected to an index storage priority metric analysis. The storage priority metric evaluates and sorts the storage priority of the storage data sequence according to the storage activity factor and other related factors (such as storage capacity, storage cost, etc.). The specific operation includes calculating the storage priority of each type of index field by combining the storage activity factor with other factors such as the cost and performance requirements of data storage, and finally obtaining the index storage priority metric of each type of index field for each sequence segment.

[0071] Step S3: Prioritize storage management of corresponding stored data sequence segments according to the index storage priority metric of each type of index field for each sequence segment.

[0072] In an embodiment of the present invention, data storage management is performed on the corresponding storage data sequence segments according to the index storage priority metric previously analyzed. For each sequence segment corresponding to each type of index field, it is first sorted according to its index storage priority metric, and the storage data segments with higher priority are stored in a storage device with higher performance, such as an SSD or a memory cache. These storage areas usually have higher read and write speeds to support frequent query access operations. Data segments with higher index storage priority metrics will obtain more storage resources to improve access efficiency; while for data segments with lower index storage priority metrics, they can be stored in a less efficient but larger capacity storage device, such as an HDD storage pool or cloud storage. These data segments can be stored and archived in batches to reduce the occupation of efficient storage resources, while also ensuring the security and integrity of these data. In the storage management process, it is also necessary to dynamically adjust the storage resource allocation. For example, if the query access rate of certain data segments changes, the location of the storage segment and the storage resource allocation will be automatically adjusted according to the real-time monitored data access mode. This process relies on the automated management tool of the storage system, combined with big data analysis and machine learning algorithms to achieve dynamic scheduling of storage resources.

[0073] Further, as an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 Detailed step flow diagram of step S1 in FIG. 1 , in this embodiment, step S1 includes the following steps:

[0074] Step S11: obtaining a data sequence to be stored corresponding to the automobile transaction platform, wherein the data sequence to be stored includes a user data subsequence, a vehicle data subsequence, and an automobile transaction data subsequence;

[0075] In an embodiment of the present invention, by extracting user data, vehicle data and transaction data from the automobile trading platform, the user data subsequence contains basic information of each user, such as user ID, registration time, geographic location, browsing history and transaction history, etc.; the vehicle data subsequence contains basic information of each vehicle, such as vehicle brand, model, age, price, owner information, maintenance history, etc.; the transaction data subsequence contains all transaction records, including transaction time, transaction amount, transaction parties, payment method, etc. These data can be filtered out by querying the platform database according to a certain time range and conditions (such as user behavior, time period of transaction records) to be stored. After these data are collected, they will be organized into a unified sequence sorted by time or other relevant conditions, and finally the data sequence to be stored corresponding to the automobile trading platform is obtained.

[0076] Step S12: classifying the hierarchical relationship index fields for each type of data sub-column corresponding to the data sequence to be stored, so as to obtain a multi-dimensional index field cluster corresponding to each type of data sub-column;

[0077] In the embodiment of the present invention, each type of data subsequence in the previously extracted data sequence to be stored contains multi-dimensional information. First, the fields in these data subsequences need to be indexed. The index fields in the user data subsequence include user ID, geographic location, browsing history, transaction history, etc.; the index fields in the vehicle data subsequence include vehicle brand, model, age, price, etc.; the index fields in the transaction data subsequence include transaction time, transaction amount, transaction parties, etc. When these index fields are hierarchically classified, hierarchical management can be performed based on business logic and data importance. For example, user data is indexed with user ID as the core and combined with geographic location. Vehicle data can be classified by brand, model, etc., and transaction data can be classified by transaction time, amount, etc. This step ensures that different types of data can be efficiently stored and quickly queried according to their hierarchical relationship, and finally a multi-dimensional index field cluster corresponding to each type of data sub-column is obtained.

[0078] Step S13: performing semantic index pair refinement and optimization on the multi-dimensional index field cluster corresponding to each type of data sub-column to obtain each type of index pair field, wherein each type of index pair field includes user ID, geographic location, browsing history and transaction history user index pair fields, vehicle brand, model, age and price vehicle index pair fields and transaction time, transaction amount and transaction parties' automobile transaction index pair fields;

[0079] In the embodiment of the present invention, by performing semantic analysis and optimization on the multi-dimensional index field clusters in each type of data subsequence, more efficient index pairs for query and storage are extracted. For example, in user data, more representative indexes are extracted by analyzing the user's browsing history and transaction history, such as defining the relationship between the user ID and the geographic location, browsing history, and transaction history based on the user's purchase interest, activity level, and other characteristics; in vehicle data, query optimization indexes based on dimensions such as brand and age are optimized based on the brand, model, and price of the vehicle; and the index fields in transaction data generate more efficient index pairs based on the timeliness and amount characteristics of the transaction. When searching for a specific vehicle, you can combine natural language processing and data mining technology to further semantically optimize these fields to ensure that they can accurately match user queries and needs. For example, there is a direct business correlation between transaction history and transaction amount, and the relationship between browsing history and vehicle model can be more accurately optimized through machine learning algorithms to improve query efficiency and accuracy. Ultimately, each type of index pair field is obtained, where each type of index pair field includes user ID, geographic location, browsing history, and transaction history user index pair fields, vehicle brand, model, age, and price vehicle index pair fields, and transaction time, transaction amount, and transaction index pair fields for both parties to the transaction.

[0080] Step S14: performing quantitative calculation of storage space limit for each type of index field to obtain storage release memory space limit corresponding to each type of index field;

[0081] In an embodiment of the present invention, after completing the optimization of the index fields of each type of data, it is necessary to perform a quantitative calculation of the space requirement limit for each index field. The key to this step is to calculate the storage space required for each type of index field based on the scale of the data and the actual storage requirements of each field. For example, the index fields of user ID and geographic location require a smaller storage space, while fields such as transaction amount and transaction time require a larger storage space. Factors such as the size of the index field, the amount of data, and the frequency of data changes need to be considered during the calculation. By quantifying the storage space required for these fields, storage resources can be reasonably planned to avoid wasting storage space, and the storage release memory space limit size of each type of data field can be obtained, and finally the storage release memory space limit corresponding to each type of index field can be obtained.

[0082] Step S15: By taking the storage release memory space limit corresponding to each type of index field as the corresponding sequence storage memory space size, and dividing the corresponding data sequence to be stored based on the sequence storage memory space size corresponding to each type of index field, the storage data sequence segment corresponding to each type of index field is obtained.

[0083] In an embodiment of the present invention, after obtaining the storage release memory space limit of each index field through previous analysis, the data to be stored is divided and stored according to these space limit degrees. During the storage process, the storage space limit degree of each type of index field is matched with the size of the data sequence to be stored, and the storage space of each type of data is reasonably divided. For example, the storage space requirements of user data, vehicle data and transaction data are different, so their storage locations should also be separated, and the storage unit is adjusted according to the data size. For example, if the space requirement of the user data field is small, it can be selected to be stored in the memory or on a fast storage medium; while the storage requirement of transaction data or vehicle data is large, it needs to be stored in a storage device such as a disk or the cloud. When storing, it is optimized according to the limit capacity of the system memory to avoid system overload or uneven data storage, ensure efficient data storage and do not waste storage resources. Different storage strategies will be adopted according to the access frequency of different data types. For example, for frequently queried user data, a faster cache mechanism can be used, and for historical transaction data, a more persistent storage solution can be used, and finally the storage data sequence segment corresponding to each type of index field is obtained.

[0084] Further, as an embodiment of the present invention, refer to Figure 3 As shown, Figure 2 FIG. 1 is a schematic diagram of a detailed step flow chart of step S13 in FIG. 1 . In this embodiment, step S13 includes the following steps:

[0085] Step S131: performing entity extraction processing on each index field in the multi-dimensional index field cluster corresponding to each type of data sub-column, to obtain index field entities corresponding to each type of data sub-column index field cluster;

[0086] In an embodiment of the present invention, entity extraction is performed on each index field in the multidimensional index field cluster corresponding to each type of data sub-column. Assuming that the data sub-column contains information related to automobile transactions, such as user information, vehicle information, transaction information, etc., each data field such as "user ID", "geographic location", and "vehicle brand" needs to extract the corresponding entity from the original data. Entity extraction can be achieved by using a named entity recognition (NER) model in natural language processing (NLP) technology. The model can automatically identify and extract entities in the data, such as extracting a specific user identifier from the "user ID" field, extracting a city name or region from the "geographic location" field, and extracting a brand name (such as "Toyota" or "BMW") from the "vehicle brand" field. At this time, the entity information of each field has been separately identified and extracted, and finally the index field entity corresponding to each type of data sub-column index field cluster is obtained.

[0087] Step S132: Based on the index field entity corresponding to each type of data sub-column index field cluster, a word vector embedding process is performed on each corresponding index field in the multi-dimensional index field cluster corresponding to each type of data sub-column, so as to obtain an index field embedding word vector corresponding to each type of data sub-column index field cluster;

[0088] In an embodiment of the present invention, word vector embedding processing is performed on each index field based on the index field entity corresponding to each type of data sub-column index field cluster. The core goal of this step is to convert each extracted entity into a vector form for subsequent machine learning and semantic analysis operations. To this end, a word embedding method based on deep learning is adopted, such as Word2Vec or GloVe model. Assuming that the entity "Toyota" is extracted from the "vehicle brand" field, "Toyota" is converted into a vector representation of a fixed dimension through a pre-trained word vector model. For example, Word2Vec will calculate the semantic representation vector of "Toyota" in a specific context based on the context information of the vocabulary. In the same way, word vector conversion is performed on fields such as "geographic location" and "transaction amount" to ensure that the entity information of each index field can be uniformly represented in vector form, and finally the index field embedding word vector corresponding to each type of data sub-column index field cluster is obtained.

[0089] Step S133: performing contextual semantic reasoning analysis on the index field embedding word vectors corresponding to each type of data sub-column index field cluster to obtain the semantic relationship between each index field in each type of data sub-column index field cluster;

[0090] In an embodiment of the present invention, contextual semantic reasoning analysis is performed on the index fields in each type of data sub-column index field cluster based on the obtained index field embedding word vectors. The purpose of this step is to identify the semantic relationship between different fields so as to reveal their potential correlation and mutual influence. To this end, advanced models such as deep neural networks (DNNs) or graph neural networks (GNNs) can be used for reasoning analysis. Specifically, the semantic similarity between the fields is first calculated based on the embedded word vectors of each index field, such as using metrics such as cosine similarity or Euclidean distance to evaluate the relationship between the two fields. For example, by analyzing the word vectors of the "user ID" and "transaction amount" fields, the association between the "user ID" and the "transaction amount" fields can be identified, and then the transaction patterns and behavioral characteristics of certain users at specific time points can be inferred. In this way, the semantic relationship between the fields can be deeply mined, and finally the semantic relationship between each index field in each type of data sub-column index field cluster can be obtained.

[0091] Step S134: Based on the semantic relationship between each index field in each type of data sub-column index field cluster, semantic index pair refinement and optimization are performed on each corresponding index field in the multi-dimensional index field cluster corresponding to each type of data sub-column, and semantically repeated or irrelevant index fields are removed to obtain each type of index pair field, wherein each type of index pair field includes user ID, geographic location, browsing history and transaction history user index pair fields, vehicle brand, model, age and price vehicle index pair fields and transaction time, transaction amount and transaction parties' automobile transaction index pair fields.

[0092] In an embodiment of the present invention, semantic index pairs are refined and optimized for index fields in a multidimensional index field cluster corresponding to each type of data sub-column based on the semantic relationship between each index field obtained in the aforementioned steps. The purpose of this step is to remove redundant or irrelevant index fields to ensure that the final index field cluster is more streamlined, efficient and meaningful. First, the redundant parts between each index field are identified through the semantic relationship obtained by semantic reasoning analysis. For example, if the semantic relationship between "user ID" and "transaction history" is too similar, there is no need to repeatedly process these two fields, or they can be merged into a composite field. At the same time, irrelevant fields that do not contribute significantly to transaction prediction or analysis are eliminated, such as some low-frequency or isolated fields, such as "registration time". ”, “Owner information”, etc. Each type of index pair field retained will include the following important dimensions: “User ID” related to user information, “Geographic location” field related to geographic location and behavior, “Vehicle brand, model, age, price” field related to vehicle information, and “Transaction time, transaction amount, transaction parties” field related to transaction information. The field cluster optimized by this step can better reflect the intrinsic connection between different data sub-columns, and finally obtain each type of index pair field, where each type of index pair field includes user ID, geographic location, browsing history and transaction history user index pair fields, vehicle brand, model, age and price vehicle index pair fields and transaction time, transaction amount and transaction parties automobile transaction index pair fields.

[0093] Further, step S14 includes the following steps:

[0094] Step S141: Analyze the index type and distribution characteristics of each type of index field to obtain the index data type and index distribution characteristics corresponding to each type of index field;

[0095] In the embodiment of the present invention, by performing a detailed analysis on each type of index field, the corresponding index type and data distribution characteristics are determined. In the database of the automobile trading platform, SQL query statements or database management tools (such as MySQL Workbench, Oracle SQL Developer) can be used to retrieve relevant information of all index fields. The specific operation includes obtaining the index type (such as B-tree, hash index, etc.) and data type (such as integer, string, date, etc.) of the field. For example, if the field is "license plate number", the data type is string (VARCHAR), and the corresponding index type is B-tree index. For the "release date" field, if it is a date type, its index is based on B-tree or bitmap index type. By analyzing the distribution characteristics of the index field, the distribution of the field value (such as whether there is a long tail effect, a large number of repeated values ​​or sparse distribution) can be further understood. This can be done by performing aggregate query and statistical analysis on the database (such as COUNT, DISTINCT and other functions in SQL), calculating the frequency distribution of the field value, and generating a distribution chart for visualization. The key to this step is to accurately obtain the index type and distribution characteristics of each field, and finally obtain the index data type and index distribution characteristics corresponding to each type of index field.

[0096] Step S142: determining the field element size of the corresponding index field in each type of index pair field based on the index data type and index distribution characteristics corresponding to each type of index field, and obtaining the element distribution size corresponding to each type of index field; predicting the index space occupancy of the corresponding index field in each type of index pair field according to the element distribution size corresponding to each type of index field, and obtaining the space occupancy requirement corresponding to each type of index field;

[0097] In an embodiment of the present invention, the size of each field element is further determined based on the specific data type and distribution characteristics of each index field. Taking the database management tool as an example, the element size is first estimated by querying the specific data type and maximum length of each field and combining the actual distribution of the field value. For example, in the automobile trading platform, if the "license plate number" field is VARCHAR(10), the element size of each "license plate number" is a storage space of 10 characters; if the field is an integer (INT), the size of each element is 4 bytes. For larger string or date fields, their maximum length can be further analyzed to accurately calculate the element size of each field, thereby obtaining the element distribution size corresponding to each type of index field. At the same time, the space requirement of each index field is further predicted based on the element distribution size obtained in the previous step. Assuming that the element size of a field has been determined and the index type of the field is known, the space prediction can be performed based on the database storage model. Taking MySQL as an example, for the B-tree index type, the space occupancy of each index node will be determined by the field element size, the index type (single-column or multi-column index), and the hierarchical relationship of the index. By analyzing the number of index nodes for each field, the actual space requirement of each index field in the database is estimated. For example, if the field "license plate number" contains 100 million records and each index element occupies 10 bytes of space, the total space occupancy of its index can be predicted by the formula: Space occupancy = 100 million * 10 bytes * number of layers of the index tree, and finally the space occupancy requirement corresponding to each type of index field is obtained.

[0098] Step S143: Analyze the field memory usage and access frequency of each type of index field to obtain the memory usage and index access frequency corresponding to each type of index field;

[0099] In an embodiment of the present invention, memory usage and access frequency analysis is performed on each index field corresponding to each type of index pair field. This can be accomplished by combining database logs, performance monitoring tools (such as Prometheus, Grafana) and SQL query statistics. First, the query log of the database (such as MySQL's slow_query_log) is used to track the access frequency of each index field, identify the fields with more access times, and further calculate their memory usage. The memory usage can be evaluated through the database's internal query cache or external memory monitoring tools. For example, if the access frequency of an index field is high, it means that the field is frequently cached, resulting in large memory consumption. In addition, based on the access pattern (such as frequently queried fields may be loaded into memory), the memory usage of the index can be analyzed in combination with a specific query execution plan (such as MySQL's EXPLAIN), and finally the memory usage corresponding to each type of index field and the index access frequency are obtained.

[0100] Step S144: based on the memory usage corresponding to each type of index field and the index access frequency, memory space recovery prediction is performed for the corresponding index field in each type of index pair field to obtain the memory space recovery and release prediction amount corresponding to each type of index field;

[0101] In an embodiment of the present invention, the memory recovery strategy of each index field is analyzed according to the previously obtained memory usage and index access frequency. For example, for frequently accessed index fields (such as the "vehicle brand" field), a longer cache expiration time can be set to avoid frequent loading into the memory. For fields with low access frequency, their residence time in the memory can be reduced, and the memory recovery mechanism is used to release space. At this time, the memory management strategy of the operating system and the cache management mechanism of the database (such as the InnoDB buffer pool size of MySQL) are combined to predict the memory recovery and release amount of each index field. By setting the cache expiration strategy, adjusting the query cache parameters and other methods, data that is no longer frequently accessed can be effectively recovered, and finally the predicted amount of memory space recovery and release corresponding to each type of index field is obtained.

[0102] Step S145: Based on the space occupation demand corresponding to each type of index field and the predicted amount of memory space recovery and release, the storage space limit of the corresponding index field in each type of index field is quantitatively calculated to obtain the storage release memory space limit corresponding to each type of index field.

[0103] In an embodiment of the present invention, the storage space limit of the corresponding index field is quantitatively calculated based on the space occupation demand and memory recovery and release prediction obtained in the previous steps. At this stage, it is necessary to comprehensively consider the use of memory, disk and cache to determine the maximum amount of data that each index field can accommodate in actual operation. For example, by calculating the maximum storage demand of the field "license plate number" index, combined with the predicted memory recovery and release amount, it can be determined how the system should release no longer active data when a certain storage limit is reached. In specific implementation, database performance analysis tools (such as MySQLTuner, pg_stat_statements) can be used to evaluate the current storage usage of the system, and storage management strategies (such as data partitioning, shard storage, etc.) can be used to control the storage volume of data and prevent the system from experiencing performance bottlenecks due to storage overload, thereby quantifying the calculated memory space limit and finally obtaining the storage release memory space limit corresponding to each type of index field.

[0104] Further, step S144 includes the following steps:

[0105] Based on the memory usage corresponding to each type of index field, memory space consumption analysis is performed on the corresponding index fields in each type of index field to obtain the memory space consumption corresponding to each type of index field;

[0106] In an embodiment of the present invention, by combining the memory usage of each type of index field obtained by previous analysis, the corresponding index fields are analyzed in detail to determine the memory space they occupy in the database. This analysis can be achieved through the performance monitoring tool or SQL execution plan of the database. By using a memory analysis module such as a memory usage tracker (Memory Profiler) or a performance monitoring tool provided by the database (such as the SHOW STATUS command of MySQL, or the V$SEGMENT_STATISTICS view of Oracle), the memory space used by each field is obtained by monitoring the storage requirements of each index field. The analysis result should include the field type, the amount of data and its specific occupancy in the memory. For example, for an index field that stores automobile brand information in the "Automobile Trading Platform", assuming that the field is a string type, the space occupied by the field in the memory will be calculated, including the memory requirements of the character set, the storage requirements of the index structure itself, etc., and finally the memory consumption of the field is determined, thereby obtaining the specific occupancy of each type of index field in the memory, and finally obtaining the memory space consumption corresponding to each type of index field.

[0107] Preferably, based on the index access frequency corresponding to each type of index field, an access heat evaluation analysis is performed on the corresponding index field in each type of index field to obtain the index access heat corresponding to each type of index field;

[0108] In an embodiment of the present invention, after completing the memory consumption analysis, the automobile trading platform will evaluate the access popularity of each type of index field by counting the access frequency of the database operation log. This step requires the use of a log analysis tool (such as ELK Stack or Splunk) to collect and analyze the query log to identify the usage frequency of each index field. For example, if an index field related to "car model" has been queried significantly more times in the past month, then the field is considered to be a field with high access popularity. By parsing the query statements in the log, the frequency of occurrence of each index field in the query can be counted, thereby obtaining the access popularity of each field. This analysis helps to determine which fields are accessed more frequently, and ultimately obtains the index access popularity corresponding to each type of index field.

[0109] Preferably, the access time period corresponding to each type of index field is obtained, and based on the access time period corresponding to each type of index field, an access heat decay analysis is performed on the index access heat corresponding to each type of index field to obtain the access heat decay rate corresponding to each type of index field;

[0110] In an embodiment of the present invention, by obtaining the access time period corresponding to each type of index field, and performing a decay analysis on the access heat of each type of index field based on the access time period, this analysis can be performed through a time series analysis method, using the timestamp information recorded in the access log of the database, combined with an access frequency decay model (such as an exponential decay model) to calculate the heat decay rate of each type of index field. In specific operations, a script can be written to regularly read the log data, and calculate the heat change trend of each field according to the access time period (such as grouping by hour, day or week). For example, within a time span of one month, if the access volume of a certain index field is relatively frequent in the previous week and then gradually decreases, then the heat decay rate of this field can be calculated by comparing the access volume in different time periods. In this way, the platform can obtain the access heat decay rate of each type of index field, and finally obtain the access heat decay rate corresponding to each type of index field.

[0111] Preferably, the memory recovery timing of the corresponding index fields in each type of index pair field is determined based on the access heat decay rate corresponding to each type of index field, so as to obtain the optimal index memory recovery timing corresponding to each type of index field;

[0112] In an embodiment of the present invention, after performing access heat decay analysis, the platform will determine the best time to recycle the memory of each type of index field based on the access heat decay rate of the index field. To this end, thresholds and recycling strategies can be set. For example, when the heat decay rate of a field exceeds a predetermined threshold, the memory space occupied by the field will be marked as recyclable. By implementing a memory recycling module (such as using Redis or memory pool technology), the access heat of the index field is regularly checked and it is determined whether memory release is required. For example, if the heat decay rate of a "car brand" index field is greater than the set threshold, the memory space of the field will be released in the next periodic recycling. In this way, the memory resources of the database can be effectively managed to avoid useless data occupying too much memory, and finally the best time to recycle the index memory corresponding to each type of index field is obtained.

[0113] Preferably, memory space recovery prediction is performed on the corresponding index fields in each type of index field according to the memory space occupied and consumed by each type of index field and the best time for index memory recovery, so as to obtain the predicted amount of memory space recovery and release corresponding to each type of index field.

[0114] In an embodiment of the present invention, the memory recovery and release amount can be predicted based on the memory consumption and recovery timing of each type of index field. This process involves combining the memory consumption of each field with its recovery timing, and estimating the memory space to be recovered through a prediction algorithm (such as a regression analysis model based on historical data). The platform will predict each index field and analyze their recovery timing and memory release amount. For example, if the memory consumption of an index field is 500MB and the memory recovery threshold will be reached in the next recovery cycle, the predicted recovery and release amount is 500MB. The platform can ensure that the recovered memory space meets the performance requirements through calculation and analysis, while avoiding excessive recovery that leads to system performance degradation. Through accurate prediction of the recovery and release amount, it can optimize resource allocation and memory management to improve the operating efficiency of the database and the overall performance of the platform, and finally obtain the predicted amount of memory space recovery and release corresponding to each type of index field.

[0115] Furthermore, the storage space limit quantitative calculation is performed by using a storage space limit calculation formula, wherein the storage space limit calculation formula is specifically:

[0116] ;

[0117] In the formula, For the The storage release memory space limit corresponding to the class index field, is the total number of index field types, The item index of the index field type. is the time range parameter, is the time variable parameter, For the The space required for the class index field. For the The predicted amount of memory space to be reclaimed and released corresponding to the class index field. For the The memory consumption decay rate corresponding to the class index field, For the The storage space reuse efficiency coefficient corresponding to the class index field, For the Class index field at time The amount of memory released at A correction factor for the extent to which memory space is freed for storage.

[0118] The present invention obtains a storage space limit calculation formula by using a specific mathematical model and verifying it, which is used to quantitatively calculate the storage space limit of the corresponding index field in each type of index pair field. The storage space limit calculation formula can comprehensively consider multiple factors such as the space occupation demand of the index field, the memory recovery release amount, the memory consumption decay rate, the storage space reuse efficiency, etc., so as to optimize the management of the memory resources of the corresponding index field. By using the space occupation demand, the physical space size occupied by a certain index field in the database is reflected. By accurately calculating the space occupation demand of each type of index field, it is possible to identify how much memory needs to be allocated to store the index, thereby providing a reasonable space reservation and resource allocation plan for the platform; the memory space recovery release prediction amount measures the memory release rate or efficiency of the index field during the memory recovery process, which can help predict the memory recovery process and its efficiency, and guide the database on how to reduce the waste of memory resources by optimizing the recovery mechanism. Efficient recovery can reduce the burden of memory occupied by the index field, thereby optimizing the overall memory usage; as the memory consumption decay rate increases over time, some index fields will no longer be frequently accessed or used, and their memory The consumption may gradually decrease. The decay rate of memory consumption can help predict the trend of memory usage and identify which index fields are no longer active, so that strategies can be taken to optimize or recycle them, which helps to better control the growth of memory usage and delay the resource shortage of the platform. The storage space reuse efficiency coefficient indicates whether the free space can be effectively reused after the memory is released. If the platform can efficiently reuse the released memory space, it can significantly improve the efficiency of memory use and avoid performance losses in the frequent memory allocation and release process. The time function memory release amount changes with time and shows a nonlinear relationship. Dynamically tracking and calculating the memory release amount enables the platform to more accurately adjust the allocation of memory resources in the time dimension, avoid excessive or insufficient memory resources in certain periods, and thus achieve more optimized memory management. The calculated storage release memory space limit, that is, the maximum memory space occupation of this type of index field that can be carried after considering all factors, this quantitative result can provide a clear goal for the allocation and recovery of database memory resources, ensuring that the storage space is maximized and the memory usage efficiency is optimized without exceeding the memory resource limit. In summary, this formula fully considers the first The limit of storage and release memory space corresponding to the class index field , the total number of index field types , item index of index field type , time range parameter , time variable parameter , No. The space required for the class index field , No. The predicted amount of memory space to be reclaimed and released corresponding to the class index field , No. Memory consumption decay rate corresponding to the class index field , No. Storage space reuse efficiency coefficient corresponding to the class index field , No. Class index field at time Amount of memory released at , the correction factor for storing the limit of memory space released , according to The limit of storage and release memory space corresponding to the class index field The correlation between the above parameters constitutes a functional relationship This formula can realize the quantitative calculation process of the storage space limit of the corresponding index field in each type of index pair field, and at the same time, release the correction coefficient of the storage space limit by storing The introduction of can be adjusted according to the errors that occur during the calculation process, thereby improving the accuracy and applicability of the storage space limitation calculation formula.

[0119] Further, step S2 includes the following steps:

[0120] Step S21: obtaining the historical query log of the automobile trading platform data, and determining the query time period and access frequency of the stored data sequence segment corresponding to each type of index field based on the historical query log of the automobile trading platform data, and obtaining the query access time period and query access frequency of the sequence segment corresponding to each type of index field;

[0121] In an embodiment of the present invention, by obtaining the historical query log of the automobile trading platform, the historical query log records various operations of the user on the platform, including query conditions, query time, query frequency and other information. By analyzing these query logs, the query access time period and query access frequency of the storage data sequence (such as automobile data of a specific brand or model) corresponding to each type of index field (for example, brand, model, price range, etc.) can be determined. The specific operation includes extracting the timestamp and query conditions of each query request from the log, and then aggregating and counting the query conditions of the same type to calculate the access frequency in each time period. The query time period can be set to units such as hours, days or months according to specific needs. By counting the access frequencies in different time periods, the periodic characteristics of the query access are obtained, and finally the query access time period and query access frequency of the sequence segment corresponding to each type of index field are obtained.

[0122] Step S22: performing a quantitative calculation of the query access rate according to the query access time period and the query access frequency of the sequence segment corresponding to each type of index field, and obtaining the query access rate of the sequence segment corresponding to each type of index field;

[0123] In an embodiment of the present invention, by quantitatively calculating the query access rate based on the query access time period and query access frequency of the sequence segment corresponding to each type of index field previously obtained, the query access rate can be defined as the ratio of the query frequency of a certain type of index field in a specific time period to the query frequency of the index field in all time periods. The specific implementation method includes firstly calculating the total query frequency of each type of index field in all time periods (i.e., the total access volume of the field), and then dividing the query frequency in each specific query time period to obtain the query access rate of each time period. The quantified result of the query access rate can reflect the query activity of users on a certain type of data in a specific time period, and finally obtain the query access rate of the sequence segment corresponding to each type of index field.

[0124] Step S23: performing query access interval analysis on the corresponding stored data sequence segments based on the query access time period of the sequence segments corresponding to each type of index field, so as to obtain the query access time interval of the sequence segments corresponding to each type of index field; obtaining the storage time interval of the sequence segments corresponding to each type of index field, and performing storage proximity access degree calculation based on the query access time interval and storage time interval of the sequence segments corresponding to each type of index field, so as to obtain the storage interval proximity access degree of the sequence segments corresponding to each type of index field;

[0125] In an embodiment of the present invention, by analyzing the query access time period of each type of index field, the query access interval of the data sequence of this type is further calculated. The query access interval can be defined as the time difference between two query accesses. The time interval can reflect the query periodicity of the data. If the query interval is short, it means that the data has a higher query frequency in a short time; if the query interval is long, it means that the query frequency of the data is low. In the specific operation, first, all query time points corresponding to each type of index field are extracted using the timestamp information in the historical query log, and then for each query time point, the time difference between it and the previous query time point is calculated to obtain the distribution of the query access interval, thereby obtaining the query access time interval of the sequence segment corresponding to each type of index field. At the same time, by obtaining the storage time interval of the storage data sequence corresponding to each type of index field, the storage time interval refers to the update frequency of a data sequence in the storage system or the periodicity of data changes. The method for obtaining the storage time interval is usually implemented through data update logs or access records of the storage system. The query access time interval and the storage time interval based on each type of index field are compared to calculate the storage access proximity. The storage access proximity can be defined as the proximity between the query access interval and the storage time interval. The higher the value, the more matched the query access is with the data storage cycle. This calculation method includes comparing the query access interval and the storage time interval of each type of index field, calculating their differences, and quantifying the degree of storage access proximity according to a predetermined threshold standard, and finally obtaining the storage interval access proximity of the sequence segment corresponding to each type of index field.

[0126] Step S24: performing storage activity evaluation analysis on the stored data sequence segments corresponding to each type of index field based on the query access rate and the storage interval proximity access degree, so as to obtain the storage activity factor of each type of index field for each sequence segment;

[0127] In an embodiment of the present invention, the storage activity of the data sequence segment corresponding to each type of index field is evaluated by combining the query access rate and the degree of proximity of the storage interval to the access. The storage activity reflects the frequency of use of data and the importance of access. The specific implementation method is first, comprehensively considering the query access rate and the degree of proximity of the storage to the access. If the query access rate of a certain type of index field is high, and the time interval of the query access is close to the storage time interval, it means that the storage activity of the data sequence is high. The storage activity evaluation analysis scores each data sequence segment, and calculates the activity factor of each data segment according to the query access frequency, the query time interval and the degree of proximity of the storage time interval. The higher the value of the storage activity factor, the higher the storage activity of the data. Finally, the storage activity factor of each type of index field for each sequence segment is obtained.

[0128] Step S25: performing index storage priority metric analysis on the corresponding stored data sequence segments according to the storage activity factor of each type of index field for each sequence segment, and obtaining the index storage priority metric of each type of index field for each sequence segment.

[0129] In an embodiment of the present invention, an index storage priority metric analysis is performed on the corresponding stored data sequence segments by combining the storage activity factor of each sequence segment obtained by previous analysis. The storage priority metric is to evaluate and sort the storage priority of the stored data sequence according to the storage activity factor and other related factors (such as storage capacity, storage cost, etc.). The specific operation includes calculating the storage priority of each type of index field by combining the storage activity factor with other factors such as the cost and performance requirements of data storage. Data sequences with higher priorities should be indexed first to improve query efficiency. The key to this step is to allocate storage resources to different data sequences according to the level of the activity factor to ensure that high-activity data can be efficiently stored and can quickly respond to query requests, and finally obtain the index storage priority metric of each type of index field for each sequence segment.

[0130] Furthermore, the calculation of the storage proximity access degree according to the query access time interval and the storage time interval of the sequence segment corresponding to each type of index field in step S23 includes the following steps:

[0131] Synchronously draw the query access time interval and storage time interval of the sequence segment corresponding to each type of index field on the time axis to generate a time axis distribution diagram between the query time interval and storage time interval corresponding to each index sequence segment;

[0132] In an embodiment of the present invention, the query access time interval and storage time interval of the sequence segment corresponding to each type of index field previously obtained are synchronously drawn on a time axis. Specifically, for each index field (such as vehicle brand, vehicle condition, price range, etc.), the corresponding query access time interval and storage time interval data are obtained by tracking the timestamps of historical query records and storage operations. These time interval data can be obtained through the log system, access record table or background data analysis module of the database, and a time axis drawing tool (such as a time series analysis library or a data visualization tool) can be used to synchronously display these query and storage time intervals to form a corresponding time axis distribution diagram. Each segment represents a specific time interval, reflecting the relative timing of query and storage operations. In this way, the distribution and intersection of query access and data storage activities within a specific time period can be clearly displayed, and finally a time axis distribution diagram between the query time interval and the storage time interval corresponding to each index sequence segment is generated.

[0133] Preferably, the time axis distribution diagram between the query time interval and the storage time interval corresponding to each index sequence segment is divided into repeated periods between query and storage to obtain the repeated period between the query interval and the storage interval corresponding to each index sequence segment;

[0134] In an embodiment of the present invention, a time axis distribution diagram between query time intervals and storage time intervals corresponding to each type of index sequence segment previously generated is further analyzed. Specifically, the overlapping area analysis of query and storage time periods is first performed through a time axis tool, and the time period where query access and storage operations overlap is identified through a time axis overlap determination method (such as an interval intersection algorithm). The overlapping portion of each query and storage time interval is a "repeated period", which represents the common activity of the query operation and the storage operation in the same time period. For each repeated period, its start and end time are accurately calibrated, and its duration is further calculated to ensure that the division of the repeated period is accurate and can reflect the intersection relationship between query and storage, and finally the repeated period between the query interval and the storage interval corresponding to each index sequence segment is obtained.

[0135] Preferably, based on the repetitive period between the query interval and the storage interval corresponding to each index sequence segment, the query access time interval and the storage time interval of the corresponding sequence segment are queried and the storage overflow period is determined to obtain the query overflow period and the storage overflow period corresponding to each index sequence segment;

[0136] In an embodiment of the present invention, the overflow period of the query access time interval and the storage time interval is further determined based on the repetitive period between the query interval and the storage interval corresponding to each index sequence segment. The specific operation is to identify and mark the query and storage time intervals outside the repetitive period to determine the query overflow period and the storage overflow period. For example, when a query time interval exceeds the storage time interval, the query part is the query overflow period; conversely, when the storage interval exceeds the query interval, the storage part is the storage overflow period. This process can be automatically identified using algorithms (such as time interval comparison and difference calculation). The establishment of the query overflow period and the storage overflow period requires relying on the detailed timestamp records of the query and storage operations in the database, as well as the time interval calculation method (such as interval operation), and finally obtains the query overflow period and the storage overflow period corresponding to each index sequence segment.

[0137] Preferably, the storage proximity access degree is calculated for the query overflow period and the storage overflow period corresponding to each index sequence segment according to the repetitive period between the query interval and the storage interval corresponding to each index sequence segment, so as to obtain the storage interval proximity access degree of the sequence segment corresponding to each type of index field.

[0138] In an embodiment of the present invention, the storage proximity access degree is calculated for the query overflow period and the storage overflow period according to the repetition period of each index sequence segment. The specific operation is to statistically calculate the proximity between the query and storage operations by using the repetition period divided by the sum of the query overflow period and the storage overflow period, thereby obtaining the indicator of "storage proximity access degree", and finally quantitatively calculating the storage interval proximity access degree of the sequence segment corresponding to each type of index field.

[0139] Further, step S24 includes the following steps:

[0140] Step S241: performing query access heat distribution analysis on the stored data sequence segments corresponding to each type of index field based on the query access rate to generate a query access heat distribution graph corresponding to each type of index field;

[0141] In an embodiment of the present invention, the query access rate obtained by the previous quantitative calculation is combined to perform a query access heat distribution analysis on the stored data sequence corresponding to each type of index field, so as to determine the stored data sequence of each type of index field (for example, car brand, model, year, price range, etc.) according to the historical user query behavior data of the platform. These stored data sequences are usually segmented according to a certain time granularity, such as by day or by hour. After obtaining the query access data of each type of index field, the query access frequency in different time periods is summarized by a statistical analysis method (such as frequency analysis, weighted average, etc.) to generate a query access heat distribution map. Specifically, a data analysis tool (such as pandas, numpy or Matplotlib library in Python) is used to display the access frequency in each segment in a graph to form a heat distribution map. The color scale of the heat map indicates the distinction between different access frequencies. Through these analyses, the query heat distribution of each field in different time periods can be clearly understood, which helps the platform understand the degree of user attention to different types of cars in different periods, and finally generates a query access heat distribution map corresponding to each type of index field.

[0142] Step S242: Obtain the corresponding query access behavior cycle according to the query access heat distribution diagram corresponding to each type of index field, and perform access fluctuation range analysis on the query access rate based on the query access behavior cycle to obtain the query access fluctuation range corresponding to each type of index field in each access behavior cycle;

[0143] In an embodiment of the present invention, by obtaining the corresponding query access behavior cycle according to the previously generated query access heat distribution map, it is first necessary to define the standard of the access behavior cycle. According to the actual situation of the automobile trading platform, the access behavior cycle may refer to the periodic changes within a month, a week, or even a day. Based on the query access heat distribution of each index field, periodic analysis (such as Fourier transform or autocorrelation analysis) is used to extract the behavior cycle characteristics. These periodic characteristics usually reflect the active cycle of users when querying a certain type of automobile, such as a surge in the number of queries for certain models in a specific month or around a specific holiday. Further, through time series analysis (such as moving average method or seasonal trend analysis), the fluctuation range of the query access rate is analyzed. The fluctuation range refers to the change range of the query access frequency within a complete cycle, showing an obvious fluctuation pattern (such as the difference between weekdays and weekends, the fluctuation of holidays, etc.). Through these methods, the query access fluctuation range of each index field in each cycle can be quantitatively analyzed, and finally the query access fluctuation range corresponding to each type of index field in each access behavior cycle is obtained.

[0144] Step S243: performing storage access density analysis on the storage data sequence segments corresponding to each type of index field based on the query access fluctuation amplitude corresponding to each type of index field in each access behavior cycle, and obtaining the storage query access density corresponding to each type of index field;

[0145] In an embodiment of the present invention, a statistical analysis of storage access density is performed on the corresponding storage data sequence segments by combining the query access fluctuation amplitude corresponding to each type of index field in each access behavior cycle obtained by previous statistical analysis, so as to further analyze the access frequency of each storage data sequence segment according to the obtained query access fluctuation amplitude, calculate the query density of each segment (i.e., the number of accesses per unit time), and analyze which data sequence segments have denser query densities in certain time periods by calculating the correlation between the access density and the fluctuation amplitude, and then derive the storage access density to indicate which data segments need to be access optimized more frequently, and finally obtain the storage query access density corresponding to each type of index field.

[0146] Step S244: Perform storage activity evaluation analysis on the storage data sequence segments corresponding to each type of index field based on the storage query access density and storage interval proximity access degree corresponding to each type of index field, so as to obtain the storage activity factor of each type of index field for each sequence segment.

[0147] In an embodiment of the present invention, a storage activity evaluation analysis is performed on the storage data sequence segment of each type of index field by combining the storage query access density and the storage interval access proximity obtained by the previous quantitative calculation. It is necessary to evaluate the storage activity of the data based on the previously obtained storage query access density and the storage interval access proximity. The storage interval access proximity refers to whether there is a short time interval between the query access of a certain data sequence within the same query behavior cycle, indicating that the data segment is in a high-frequency access state. For each storage data sequence segment, based on the storage access density and the time interval data, a threshold algorithm or a weighted average method is set to calculate its storage activity factor. The higher the value of the activity factor, the more frequent the query activity of the data sequence segment, and also indicates that the data segment will require higher storage and processing resources in the future. Specifically, a multi-dimensional evaluation model (such as a weighted scoring method) can be designed to combine the query frequency, storage interval and fluctuation amplitude to calculate the storage activity factor of each data segment. The factor value can be used as a reference for optimizing the storage structure, scheduling query requests and allocating computing resources, thereby helping the platform to perform more refined storage management and resource allocation in actual operation, and finally obtain the storage activity factor of each type of index field for each sequence segment.

[0148] Furthermore, the present invention also provides a data management system for an automobile trading platform, which is used to execute the data management method for the automobile trading platform as described above. The data management system for the automobile trading platform includes:

[0149] The storage partitioning module of the data sequence to be stored is used to obtain the data sequence to be stored corresponding to the automobile trading platform, and classify and extract the index pairs of the data sequence to be stored to obtain each type of index pair field; perform quantitative calculation of the storage space limit according to each type of index pair field to obtain the storage release memory space limit degree corresponding to each type of index field; perform sequence storage partitioning on the corresponding data sequence to be stored based on the storage release memory space limit degree corresponding to each type of index field, so as to obtain the storage data sequence segmentation corresponding to each type of index field;

[0150] A storage sequence segmentation priority measurement module is used to obtain the query access rate and storage interval proximity access degree of the corresponding sequence segment according to the storage data sequence segment corresponding to each type of index field, and perform storage activity evaluation analysis on the storage data sequence segment corresponding to each type of index field based on the query access rate and storage interval proximity access degree to obtain the storage activity factor of each type of index field for each sequence segment; perform index storage priority measurement analysis on the corresponding storage data sequence segment according to the storage activity factor of each type of index field for each sequence segment, so as to obtain the index storage priority measurement of each type of index field for each sequence segment;

[0151] The sequence segment priority storage management module is used to perform priority storage management on the corresponding stored data sequence segments according to the index storage priority metric of each type of index field for each sequence segment.

[0152] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A data management method for an automobile trading platform, characterized in that: The following steps are involved: Step S1: obtaining a data sequence to be stored corresponding to the automobile trading platform, and classifying and extracting index pairs of the data sequence to be stored to obtain each type of index pair field; performing quantitative calculation of the storage space limit according to each type of index pair field to obtain the storage release memory space limit degree corresponding to each type of index field; performing sequence storage division on the corresponding data sequence to be stored based on the storage release memory space limit degree corresponding to each type of index field to obtain the storage data sequence segment corresponding to each type of index field; Step S1 includes the following steps: Step S11: obtaining a data sequence to be stored corresponding to the automobile transaction platform, wherein the data sequence to be stored includes a user data subsequence, a vehicle data subsequence, and an automobile transaction data subsequence; Step S12: classifying the hierarchical relationship index fields for each type of data sub-column corresponding to the data sequence to be stored, so as to obtain a multi-dimensional index field cluster corresponding to each type of data sub-column; Step S13: performing semantic index pair refinement and optimization on the multi-dimensional index field cluster corresponding to each type of data sub-column to obtain each type of index pair field, wherein each type of index pair field includes user ID, geographic location, browsing history and transaction history user index pair fields, vehicle brand, model, age and price vehicle index pair fields and transaction time, transaction amount and transaction parties' automobile transaction index pair fields; Step S14: performing quantitative calculation of the storage space limit of the field according to each type of index to obtain the storage release memory space limit corresponding to each type of index field; Step S14 includes the following steps: Step S141: Analyze the index type and distribution characteristics of each type of index field to obtain the index data type and index distribution characteristics corresponding to each type of index field; Step S142: determining the field element size of the corresponding index field in each type of index pair field based on the index data type and index distribution characteristics corresponding to each type of index field, and obtaining the element distribution size corresponding to each type of index field; predicting the index space occupancy of the corresponding index field in each type of index pair field according to the element distribution size corresponding to each type of index field, and obtaining the space occupancy requirement corresponding to each type of index field; Step S143: Analyze the field memory usage and access frequency of each type of index field to obtain the memory usage and index access frequency corresponding to each type of index field; Step S144: based on the memory usage corresponding to each type of index field and the index access frequency, memory space recovery prediction is performed for the corresponding index field in each type of index pair field to obtain the memory space recovery and release prediction amount corresponding to each type of index field; Step S145: Based on the space occupation demand corresponding to each type of index field and the memory space recovery and release prediction, a storage space limit quantitative calculation is performed on the corresponding index field in each type of index field to obtain the storage release memory space limit corresponding to each type of index field; the storage space limit quantitative calculation is performed by a storage space limit calculation formula, wherein the storage space limit calculation formula is specifically: ； In the formula, For the The storage release memory space limit corresponding to the class index field, is the total number of index field types, The item index of the index field type. is the time range parameter, is the time variable parameter, For the The space required for the class index field. For the The predicted amount of memory space to be reclaimed and released corresponding to the class index field. For the The memory consumption decay rate corresponding to the class index field, For the The storage space reuse efficiency coefficient corresponding to the class index field, For the Class index field at time The amount of memory released at Correction factor for the limit of free memory space for storage; Step S15: taking the storage release memory space limit corresponding to each type of index field as the corresponding sequence storage memory space size, and dividing the corresponding data sequence to be stored based on the sequence storage memory space size corresponding to each type of index field, to obtain the storage data sequence segment corresponding to each type of index field; Step S2: according to the storage data sequence segments corresponding to each type of index field, the query access rate and the storage interval proximity access degree of the corresponding sequence segments are obtained, and based on the query access rate and the storage interval proximity access degree, the storage data sequence segments corresponding to each type of index field are evaluated for storage activity, so as to obtain the storage activity factor of each type of index field for each sequence segment; according to the storage activity factor of each type of index field for each sequence segment, the corresponding storage data sequence segments are analyzed for index storage priority, so as to obtain the index storage priority metric of each type of index field for each sequence segment; Step S2 includes the following steps: Step S21: obtaining the historical query log of the automobile trading platform data, and determining the query time period and access frequency of the stored data sequence segment corresponding to each type of index field based on the historical query log of the automobile trading platform data, and obtaining the query access time period and query access frequency of the sequence segment corresponding to each type of index field; Step S22: performing a quantitative calculation of the query access rate according to the query access time period and the query access frequency of the sequence segment corresponding to each type of index field, and obtaining the query access rate of the sequence segment corresponding to each type of index field; Step S23: performing query access interval analysis on the corresponding stored data sequence segments based on the query access time period of the sequence segments corresponding to each type of index field, so as to obtain the query access time interval of the sequence segments corresponding to each type of index field; obtaining the storage time interval of the sequence segments corresponding to each type of index field, and performing storage proximity access degree calculation based on the query access time interval and storage time interval of the sequence segments corresponding to each type of index field, so as to obtain the storage interval proximity access degree of the sequence segments corresponding to each type of index field; Step S24: performing storage activity evaluation analysis on the stored data sequence segments corresponding to each type of index field based on the query access rate and the storage interval proximity access degree, so as to obtain the storage activity factor of each type of index field for each sequence segment; Step S25: performing index storage priority metric analysis on the corresponding stored data sequence segments according to the storage activity factor of each type of index field for each sequence segment, and obtaining the index storage priority metric of each type of index field for each sequence segment; Step S3: Prioritize storage management of corresponding stored data sequence segments according to the index storage priority metric of each type of index field for each sequence segment.

2. The data management method of the automobile trading platform according to claim 1, characterized in that: Step S13 includes the following steps: Step S131: performing entity extraction processing on each index field in the multi-dimensional index field cluster corresponding to each type of data sub-column, to obtain index field entities corresponding to each type of data sub-column index field cluster; Step S132: Based on the index field entity corresponding to each type of data sub-column index field cluster, a word vector embedding process is performed on each corresponding index field in the multi-dimensional index field cluster corresponding to each type of data sub-column, so as to obtain an index field embedding word vector corresponding to each type of data sub-column index field cluster; Step S133: performing contextual semantic reasoning analysis on the index field embedding word vectors corresponding to each type of data sub-column index field cluster to obtain the semantic relationship between each index field in each type of data sub-column index field cluster; Step S134: Based on the semantic relationship between each index field in each type of data sub-column index field cluster, semantic index pair refinement and optimization are performed on each corresponding index field in the multi-dimensional index field cluster corresponding to each type of data sub-column, and semantically repeated or irrelevant index fields are removed to obtain each type of index pair field, wherein each type of index pair field includes user ID, geographic location, browsing history and transaction history user index pair fields, vehicle brand, model, age and price vehicle index pair fields and transaction time, transaction amount and transaction parties' automobile transaction index pair fields.

3. The data management method of the automobile trading platform according to claim 1, characterized in that: Step S144 includes the following steps: Based on the memory usage corresponding to each type of index field, memory space consumption analysis is performed on the corresponding index fields in each type of index field to obtain the memory space consumption corresponding to each type of index field; Based on the index access frequency corresponding to each type of index field, the access heat evaluation and analysis of the corresponding index field in each type of index field is performed to obtain the index access heat corresponding to each type of index field; Obtain the access time period corresponding to each type of index field, and perform access heat decay analysis on the index access heat corresponding to each type of index field based on the access time period corresponding to each type of index field, so as to obtain the access heat decay rate corresponding to each type of index field; Based on the access heat decay rate corresponding to each type of index field, the memory recycling timing of the corresponding index field in each type of index pair field is determined to obtain the optimal index memory recycling timing corresponding to each type of index field; According to the memory space consumption corresponding to each type of index field and the optimal time for index memory recovery, memory space recovery prediction is performed on the corresponding index fields in each type of index field to obtain the predicted amount of memory space recovery and release corresponding to each type of index field.

4. The data management method of the automobile trading platform according to claim 1, characterized in that: The calculation of the storage proximity access degree according to the query access time interval and the storage time interval of the sequence segment corresponding to each type of index field in step S23 includes the following steps: Synchronously draw the query access time interval and storage time interval of the sequence segment corresponding to each type of index field on the time axis to generate a time axis distribution diagram between the query time interval and storage time interval corresponding to each index sequence segment; Dividing the time axis distribution diagram between the query time interval and the storage time interval corresponding to each index sequence segment by the repeated time period between query and storage, so as to obtain the repeated time period between the query interval and the storage interval corresponding to each index sequence segment; Based on the repetitive period between the query interval and the storage interval corresponding to each index sequence segment, query the query access time interval and the storage time interval of the corresponding sequence segment and determine the storage overflow period, and obtain the query overflow period and the storage overflow period corresponding to each index sequence segment; The storage proximity access degree is calculated for the query overflow period and the storage overflow period corresponding to each index sequence segment according to the repeated period between the query interval and the storage interval corresponding to each index sequence segment, and the storage interval proximity access degree of the sequence segment corresponding to each type of index field is obtained.

5. The data management method of the automobile trading platform according to claim 1, characterized in that: Step S24 includes the following steps: Step S241: performing query access heat distribution analysis on the stored data sequence segments corresponding to each type of index field based on the query access rate to generate a query access heat distribution graph corresponding to each type of index field; Step S242: Obtain the corresponding query access behavior cycle according to the query access heat distribution diagram corresponding to each type of index field, and perform access fluctuation range analysis on the query access rate based on the query access behavior cycle to obtain the query access fluctuation range corresponding to each type of index field in each access behavior cycle; Step S243: performing storage access density analysis on the storage data sequence segments corresponding to each type of index field based on the query access fluctuation amplitude corresponding to each type of index field in each access behavior cycle, and obtaining the storage query access density corresponding to each type of index field; Step S244: Perform storage activity evaluation analysis on the storage data sequence segments corresponding to each type of index field based on the storage query access density and storage interval proximity access degree corresponding to each type of index field, so as to obtain the storage activity factor of each type of index field for each sequence segment.

6. A data management system for an automobile trading platform, characterized in that: Used to execute the data management method of the automobile trading platform according to claim 1, the data management system of the automobile trading platform comprises: The storage partitioning module of the data sequence to be stored is used to obtain the data sequence to be stored corresponding to the automobile trading platform, and classify and extract the index pairs of the data sequence to be stored to obtain each type of index pair field; perform quantitative calculation of the storage space limit according to each type of index pair field to obtain the storage release memory space limit degree corresponding to each type of index field; perform sequence storage partitioning on the corresponding data sequence to be stored based on the storage release memory space limit degree corresponding to each type of index field, so as to obtain the storage data sequence segmentation corresponding to each type of index field; A storage sequence segmentation priority measurement module is used to obtain the query access rate and storage interval proximity access degree of the corresponding sequence segment according to the storage data sequence segment corresponding to each type of index field, and perform storage activity evaluation analysis on the storage data sequence segment corresponding to each type of index field based on the query access rate and storage interval proximity access degree to obtain the storage activity factor of each type of index field for each sequence segment; perform index storage priority measurement analysis on the corresponding storage data sequence segment according to the storage activity factor of each type of index field for each sequence segment, so as to obtain the index storage priority measurement of each type of index field for each sequence segment; The sequence segment priority storage management module is used to perform priority storage management on the corresponding stored data sequence segments according to the index storage priority metric of each type of index field for each sequence segment.

Images