Financial big data analysis platform based on artificial intelligence

Abstract

The invention discloses a financial big data analysis platform based on artificial intelligence, and relates to the field of data analysis. The data processing module is used for sequentially carrying out multi-channel progressive analysis on the obtained financial data and constructing a multi-level financial database; the data storage module is used for carrying out feature analysis and natural language processing on the financial data in the multi-level financial database to obtain corresponding semantic tags and carrying out real-time storage; the data analysis module is used for performing statistical analysis according to the financial data of the corresponding semantic tag to obtain a real-time analysis result; the data visualization module is used for carrying out visualization processing on semantic tags corresponding to financial data in the multi-level financial database; the data retrieval module is used for obtaining financial demand data and demand semantic tags corresponding to the financial demand data, retrieving corresponding visual processing results and obtaining demand retrieval information; the financial data analysis efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a financial big data analysis platform based on artificial intelligence. Background Technology

[0002] With the development of the financial industry, financial data has become characterized by its massive volume, diversity, and high real-time requirements. As businesses expand, financial institutions accumulate huge amounts of data, including not only structured financial statements and transaction records, but also a large amount of unstructured news reports and social media comments. Furthermore, due to the rapid changes in the financial market, high-frequency trading has extremely high requirements for data real-time performance. However, traditional financial data analysis methods rely on manual operation and simple statistical tools, which are inefficient and error-prone when processing large-scale data, making them difficult to adapt to the complex and ever-changing financial market.

[0003] A search revealed Chinese invention patent CN111861748A, which discloses an AI-based financial big data analysis platform. This platform includes an identity information acquisition module, an identity verification module, an initial database, a financial data extraction module, a financial data classification module, a financial data analysis module, and a financial data analysis result storage module. After the identity verification module verifies the identity of the person initiating the financial big data analysis, the financial data classification module categorizes the financial data extracted by the financial data extraction module. The financial data analysis module analyzes the categorized financial data sets and stores the analysis results in the corresponding financial data analysis result storage unit within the financial data analysis result storage module. This AI-based financial big data analysis platform provides a relatively structured analysis process for financial big data. The analysis processes for each type of financial data are independent of each other, and the analysis results are also independent of each other, improving the efficiency and reliability of financial data analysis.

[0004] Compared with existing technologies, the Chinese invention patent with patent number CN111861748A can improve the efficiency and reliability of financial data analysis by analyzing and processing each type of financial data in a way that prevents interference between the analysis processes of each type of financial data.

[0005] However, in actual use, the aforementioned platforms process different types of financial data at intervals, making it difficult to verify the correlation with other financial data that may affect the corresponding types of financial data. Analyzing and processing only one type of financial data may result in lag and errors, leading to inefficiency and inaccuracy in the financial data analysis process. Summary of the Invention

[0006] The purpose of this invention is to address the problems of low efficiency and insufficient accuracy in existing technologies by proposing an artificial intelligence-based financial big data analysis platform.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: An artificial intelligence-based financial big data analysis platform includes: The data acquisition module is used to collect financial data, including financial customer data, financial transaction data, and financial market data. The module performs labeling processing on the acquired data information to obtain the labeling results of each sub-data in each type of financial data. The data processing module is used to perform multi-channel progressive analysis based on the labeling results of each sub-data in various types of financial data, to obtain the financial data and feature data corresponding to each level, and to build a multi-level financial database. The data storage module is used to perform statistics on the financial data and feature data obtained in the constructed multi-level financial database, set corresponding semantic tags based on the statistical results, and store the corresponding financial data in real time based on the corresponding semantic tags. The data analysis module is used to perform statistical analysis on the real-time stored results corresponding to the semantic tags of the corresponding financial data in the multi-level financial database. It performs statistical analysis on the corresponding semantic tags and obtains the real-time analysis results corresponding to the corresponding semantic tags in the multi-level financial database. The data visualization module is used to visualize the real-time analysis results corresponding to each semantic tag in the multi-level financial database, and to set the multi-level financial database as a financial statistics visualization database based on the visualization results. The data retrieval module is used to obtain relevant financial demand data, acquire relevant demand semantic tags based on the financial demand data, retrieve relevant financial statistical visualization databases based on the demand semantic tags, and obtain relevant demand retrieval information.

[0008] The above technical solution further includes: the data acquisition module includes: The data acquisition module includes a customer acquisition unit, a transaction acquisition unit, a market acquisition unit, and an acquisition management unit; The customer acquisition unit is used to acquire a financial customer dataset, wherein the sub-data in the financial customer dataset includes financial customer data from various data acquisition sources, and the data acquisition time of the financial customer data from each data acquisition source is acquired. The transaction acquisition unit is used to acquire a financial transaction dataset, wherein the sub-data in the financial transaction dataset includes financial transaction data from various data acquisition sources, and the data acquisition time of the financial transaction data from each data acquisition source is acquired. The market acquisition unit is used to acquire a financial market dataset, wherein the sub-data in the financial market dataset includes financial market data from various data acquisition sources, and the data acquisition time of the financial market data from each data acquisition source is acquired. The data acquisition and management unit is used to mark the data according to the data acquisition source and data acquisition time of each sub-data in each type of financial data, and to obtain the marking results of each sub-data in each type of financial data.

[0009] Furthermore, the data processing module includes: The data processing module includes a multi-channel progressive analysis unit and a multi-level model building unit; The multi-channel progressive analysis unit is used to obtain each sub-data and the labeling results of each sub-data in each type of financial dataset. According to the data collection source corresponding to the labeling results of each sub-data, corresponding financial level sub-channels are set. The financial level channel includes each financial level sub-channel corresponding to the corresponding type of financial data and the corresponding type of financial data sub-data set. Based on the data collection time of each sub-data in the corresponding sub-data within the financial level channel corresponding to each type of financial data, time-series comparison processing is performed on each sub-data in the corresponding sub-data within each financial level channel, and corresponding sub-sequences are set for the financial data obtained within the same data collection time in each financial level channel. The set subsequences are integrated into the same vertical line according to the corresponding data collection time, and the set subsequences in each financial level channel are horizontally integrated to obtain the corresponding financial level channel sequence. Based on the financial hierarchy channel sequence corresponding to the financial transaction data, feature extraction is performed on the financial customer data to obtain the corresponding transaction customer feature data, and the obtained transaction customer feature data is uploaded to the financial hierarchy channel sequence corresponding to the financial transaction data. Based on the financial hierarchy channel sequence corresponding to the financial market data, feature extraction is performed on the financial transaction data to obtain the corresponding market transaction feature data, and the obtained market transaction feature data is uploaded to the financial hierarchy channel sequence corresponding to the financial market data. The multi-level model construction unit is used to obtain the financial level channel sequence corresponding to the corresponding type of financial data, perform sequence connection processing on the financial data and feature data contained in the financial level channel sequence according to the corresponding financial level channel, and integrate them according to the sequence connection processing results to construct a multi-level financial database.

[0010] Furthermore, the data storage module includes: The semantic analysis unit is used to acquire the financial data and feature data corresponding to the financial level channel sequence in the multi-level financial database, extract hierarchical features from the acquired financial data and feature data, and acquire the corresponding customer feature data, transaction feature data and market feature data respectively. The acquired feature data is then processed based on natural language processing technology and assigned corresponding semantic labels. The real-time storage unit is used to perform statistics based on the semantic tags corresponding to the data information within the financial hierarchy channel sequence, adjust the position of financial data with the same semantic tag in the financial hierarchy channel sequence according to the corresponding data acquisition time, and store the financial data in the multi-level financial database in real time based on the adjustment results.

[0011] Furthermore, the data analysis module includes: The hierarchical statistical analysis unit is used to perform statistical analysis based on the real-time storage results of semantic tags of corresponding data information in the channel sequences of each financial level corresponding to various types of financial data in the multi-level financial database. The unit sets a statistical analysis period and performs quantitative statistics on the financial data of corresponding semantic tags in each channel and level according to the sequence corresponding to each unit time within the statistical analysis period. The unit obtains the quantitative statistical data corresponding to each semantic tag in the corresponding channel and level. The quantitative statistical data includes quantity, proportion and change rate. The unit obtains the real-time analysis results corresponding to each semantic tag in the multi-level financial database. The correlation statistical analysis unit is used to perform correlation analysis on quantitative statistical data of various channels and levels within a multi-level financial database to obtain cross-correlation relationships between channels at different levels.

[0012] Furthermore, the process by which the correlation statistical analysis unit obtains the hierarchical channel cross-correlation relationships includes: The historical analysis results obtained from the corresponding multi-level financial databases within each statistical analysis period are obtained. The historical analysis results are then combined with the quantitative statistical data of the semantic labels corresponding to each channel within the corresponding financial level channel sequence to obtain the corresponding channel semantic reference set. Correlation analysis was performed on the obtained channel semantic reference sets to obtain the correlation data corresponding to each channel semantic reference set; The correlation data corresponding to the semantic reference sets of each channel are analyzed and processed to determine whether there is a correlation between the combinations corresponding to the semantic reference sets of each channel. If there is a correlation, the corresponding combination is marked as related. The combined channels and semantic labels corresponding to each associated channel semantic reference set are marked in a multi-level financial database, and the hierarchical channel cross-association relationship is generated based on the marking results.

[0013] Furthermore, the data visualization module includes: The association detection and verification unit is used to obtain the real-time analysis results of the semantic tags corresponding to each financial data in the multi-level financial database. It detects and verifies the real-time analysis results obtained within the corresponding statistical analysis period according to the cross-association relationship of the corresponding hierarchical channels, determines whether there is an association between the associated channels and the corresponding semantic tags, and marks the corresponding judgment results. The data visualization unit is used to acquire quantitative statistical data and corresponding detection and verification results for each semantic label within a multi-level financial database. It sets corresponding visualization charts based on the type of the quantitative statistical data and performs visualization processing. Based on the detection and verification results from the correlation detection and verification unit, it sets visualization markers corresponding to the cross-correlation relationships of the corresponding levels of channels. Based on these visualization markers, it visualizes the quantitative statistical data between corresponding semantic labels within the channels corresponding to the cross-correlation relationships of each level of channel. The obtained visualization results are then set as the financial statistical visualization database corresponding to the corresponding multi-level financial database.

[0014] Furthermore, the data retrieval module includes: The demand acquisition unit is used to acquire relevant financial demand data, including customer demand data, transaction demand data and market demand data. The acquired financial demand data is analyzed and processed based on natural language processing algorithms to obtain demand semantic tags corresponding to the financial demand data. The demand retrieval unit is used to traverse and search the financial statistics visualization database based on the obtained demand semantic tags, obtain visualization processing results that are associated with the demand semantic tags, integrate the obtained visualization processing results, and generate demand retrieval information.

[0015] The present invention has the following beneficial effects: 0. In this invention, by performing multi-channel progressive analysis on the obtained financial data according to the corresponding labeling results, the feature data corresponding to the financial data obtained by each channel in the first level is used as the financial data obtained by the corresponding channel in the next level, thus constructing a multi-level financial database. This can improve the utilization rate of corresponding types of financial data in the financial big data analysis platform and the accuracy of the data analysis process.

[0016] 1. In this invention, by statistically analyzing the financial data and feature data obtained from multi-level financial databases, setting corresponding semantic tags based on the statistical results, and storing and statistically analyzing the changes in the corresponding semantic tags in real time, corresponding real-time analysis results are obtained. Correlation analysis is performed on the semantic tags corresponding to the corresponding channels in each multi-level financial database, and the corresponding semantic tags are detected and verified based on the correlation analysis results. Correlation analysis is then performed on the corresponding semantic tags, and corresponding visualization markers are set based on the detection and verification analysis results. This improves the range of corresponding visualization processing results in the financial statistical visualization database, thereby improving the efficiency and accuracy of the financial data analysis process. 2. In this invention, by analyzing and processing the obtained financial demand data, corresponding demand semantic tags are obtained. Based on the demand semantic tags, various visualization processing results corresponding to the related semantic tags in the financial statistical visualization database are obtained. The obtained visualization processing results are integrated to obtain corresponding demand retrieval information. This can improve the retrieval scope corresponding to the financial demand data to a certain extent. Based on the visualization processing results between other related semantic tags, more perspectives are provided for the financial data analysis process, thereby improving the accuracy and efficiency of the financial data analysis process. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the structure of a financial big data analysis platform based on artificial intelligence proposed in this invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 like Figure 1 As shown, the present invention proposes an artificial intelligence-based financial big data analysis platform, comprising: The data acquisition module is used to collect financial data, including financial customer data, financial transaction data, and financial market data. The module performs labeling processing on the acquired data information to obtain the labeling results of each sub-data in each type of financial data. The data processing module is used to perform multi-channel progressive analysis based on the labeling results of each sub-data in various types of financial data, to obtain the financial data and feature data corresponding to each level, and to build a multi-level financial database. The data storage module is used to perform statistics on the financial data and feature data obtained in the constructed multi-level financial database, set corresponding semantic tags based on the statistical results, and store the corresponding financial data in real time based on the corresponding semantic tags. The data analysis module is used to perform statistical analysis on the real-time stored results corresponding to the semantic tags of the corresponding financial data in the multi-level financial database. It performs statistical analysis on the corresponding semantic tags and obtains the real-time analysis results corresponding to the corresponding semantic tags in the multi-level financial database. The data visualization module is used to visualize the real-time analysis results corresponding to each semantic tag in the multi-level financial database, and to set the multi-level financial database as a financial statistics visualization database based on the visualization results. The data retrieval module is used to obtain relevant financial demand data, obtain relevant demand semantic tags based on the financial demand data, and retrieve relevant financial statistical visualization databases based on the demand semantic tags to obtain relevant demand retrieval information. In summary, with the continuous development of the financial industry, the requirements for financial data analysis and processing are constantly increasing. However, while different types of financial data may be related during data analysis, repeatedly analyzing different types of financial data each time results in significant resource waste, data redundancy, and inaccuracies. This invention improves the scope of different types of financial data acquisition and the efficiency of data analysis by performing hierarchical progressive analysis. Furthermore, it analyzes the financial data and feature data obtained from the multi-level financial database using natural language processing technology, sets corresponding semantic tags, performs statistical analysis based on changes in these semantic tags, and obtains corresponding real-time analysis results. It then performs correlation analysis on the semantic tags corresponding to different levels of channels for different real-time analysis results. The correlation analysis results are used to verify the real-time analysis results of the corresponding semantic tags for different types of data information, and the real-time analysis results and verification results are visualized, thereby improving the accuracy and efficiency of financial data analysis.

[0020] In specific implementation, the data acquisition module includes: A customer acquisition unit is used to acquire a financial customer dataset, wherein the sub-data in the financial customer dataset includes financial customer data from various data acquisition sources, and the data acquisition time of the financial customer data from each data acquisition source is acquired. The financial customer data includes basic identity data, financial status data, and risk behavior data, etc. A transaction acquisition unit is used to acquire a financial transaction dataset, wherein the sub-data in the financial transaction dataset includes financial transaction data from various data acquisition sources, and the data acquisition time of the financial transaction data from each data acquisition source is acquired. The financial transaction data includes personal transaction data, corporate transaction data, and transaction-related data, etc. The market acquisition unit is used to acquire a financial market dataset, wherein the sub-data in the financial market dataset includes financial market data from various data acquisition sources, and the data acquisition time of the financial market data from each data acquisition source is acquired. The financial market data includes product price data and market fluctuation data, etc. The data collection and management unit is used to mark the data according to the data collection source and data collection time of each sub-data in each type of financial data, and to obtain the marking results of each sub-data in each type of financial data. It should be further explained that, in the specific implementation process, the data collection units of the data collection module, when collecting the corresponding financial data, mark the data collection sources, including but not limited to the customer accounts of financial customers, channels of different transaction types, etc., and set the corresponding channels according to the corresponding data collection sources to facilitate subsequent statistics, thereby improving the efficiency of data analysis.

[0021] In specific implementation, the data processing module includes: The multi-channel progressive analysis unit is used to obtain the sub-data and the labeling results of each sub-data in various types of financial datasets. According to the data collection source corresponding to the labeling results of each sub-data, the corresponding financial level sub-channels are set. The financial level channels include the financial level sub-channels corresponding to the corresponding type of financial data and the sub-data sets of the corresponding type of financial data. It should be further explained that, in the process of setting up financial tier channels, taking financial customer data as an example, the corresponding financial tier channels are set up according to the financial customer data. The financial tier channels are divided into n financial tier sub-channels according to the data collection source. Each financial tier sub-channel has a unique corresponding data collection source. The financial tier sub-channels corresponding to the financial customer data are set up according to the corresponding customer account information. The financial customer data obtained in each financial tier sub-channel is marked as the subset of the financial customer data. Based on the data collection time of each sub-data in the corresponding sub-data within the financial level channel corresponding to each type of financial data, time-series comparison processing is performed on each sub-data in the corresponding sub-data within each financial level channel, and corresponding sub-sequences are set for the financial data obtained within the same data collection time in each financial level channel. The set subsequences are integrated into the same vertical line according to the corresponding data collection time, and the set subsequences in each financial level channel are horizontally integrated to obtain the corresponding financial level channel sequence. It should be further explained that, in the specific implementation process, if the corresponding financial sub-channel within each financial level channel does not collect the corresponding financial data within the corresponding data collection time, the corresponding sub-sequence will be marked as an empty sequence. In the process of horizontally integrating the various sub-sequences set within each financial level channel, the sub-sequences with or without financial data will be connected sequentially according to the data collection time, and the corresponding financial level channel sequence will be obtained based on the connection result. Based on the financial hierarchy channel sequence corresponding to the financial transaction data, feature extraction is performed on the financial customer data to obtain the corresponding transaction customer feature data, and the obtained transaction customer feature data is uploaded to the financial hierarchy channel sequence corresponding to the financial transaction data. Based on the financial hierarchy channel sequence corresponding to the financial market data, feature extraction is performed on the financial transaction data to obtain the corresponding market transaction feature data, and the obtained market transaction feature data is uploaded to the financial hierarchy channel sequence corresponding to the financial market data. Furthermore, the term "financial tier channel" is a collective term for the sub-channels within the corresponding data collection sources of each type of financial data. For ease of understanding, the following explanation is provided: the tiers corresponding to financial customer data, financial transaction data, and financial market data are respectively labeled as Tier 1, Tier 2, and Tier 3. Feature extraction is then performed on the financial data obtained within Tier 1, Tier 2, and Tier 3 to obtain corresponding feature data. For example, for financial customer data within Tier 1, feature extraction is performed to obtain customer feature data for each financial customer data involved in each sub-channel of Tier 1. Customer feature data includes identity feature data, financial feature data, and behavioral feature data, etc. For financial transaction data within Tier 2, feature extraction is performed to obtain transaction feature data for each financial transaction data involved in each financial transaction channel of Tier 2. Transaction feature data includes transaction attribute feature data, transaction amount feature data, transaction frequency feature data, and transaction time feature data, etc. For financial transaction data within Tier 3, feature extraction is performed to obtain the corresponding financial transaction data. Market data is used for feature extraction. Market feature data of financial market data involved in each financial level channel within the third level are obtained. Market feature data includes market trend feature data, volume and transaction feature data, and capital flow feature data. The corresponding feature data is analyzed progressively according to the corresponding level. Feature data related to the second level is obtained from the feature data corresponding to the first level. Feature data related to the third level is obtained from the feature data corresponding to the second level. The obtained feature data are set into corresponding progressive channels. The second level includes progressive channels corresponding to the relevant feature data in each channel of the corresponding financial data in the first level. The third level includes progressive channels corresponding to the relevant feature data in each channel of the corresponding financial data in the second level. The obtained progressive channels are also set as financial level sub-channels within the corresponding financial level channel. The feature data obtained from the corresponding progressive channels are also set as corresponding subsets. Each subset is uploaded to the corresponding financial level channel sequence. Set the corresponding financial data or feature data of each channel within the financial hierarchy channel according to the corresponding data collection time to set the corresponding financial hierarchy channel sequence; The multi-level model construction unit is used to obtain the financial level channel sequence corresponding to the corresponding type of financial data, perform sequence connection processing on the financial data and feature data in the financial level channel sequence according to the corresponding financial level channel, and construct a multi-level financial database based on the sequence connection processing result; It should be further explained that the constructed multi-level financial database is used for temporary storage of financial data from corresponding data collection sources. Furthermore, the process of obtaining the characteristic data corresponding to the corresponding type of financial data, taking financial customer data as an example: The financial customer data includes basic identity data, financial status data, and risk behavior data, etc. The basic identity data includes various sub-data, including education, occupation, fixed assets, etc. Obtain the basic identity data assessment level corresponding to various preset sub-data combinations, compare and match the various sub-data combinations in the basic identity data assessment level, and obtain the assessment level corresponding to the basic identity data in the financial customer data. By comparing and matching the above methods in sequence, the assessment levels of financial status data and risk behavior data are obtained in sequence. The assessment levels corresponding to the basic identity data, financial status data, and risk behavior data of financial customers are used as customer characteristic data for the corresponding financial customer data.

[0022] In specific implementation, the data storage module includes: The semantic analysis unit is used to acquire the corresponding financial data and feature data within the corresponding financial level channel sequence in a multi-level financial database. It performs hierarchical feature extraction on the corresponding financial data and feature data, obtaining corresponding customer feature data, transaction feature data, and market feature data. Finally, it assigns corresponding semantic labels to the obtained feature data based on natural language processing technology. The process of acquiring customer characteristic data involves extracting features based on dimensions such as identity stability, financial capacity, risk characteristics, and service needs. Based on the feature extraction results, corresponding feature intervals are set, and corresponding semantic labels are assigned based on natural language processing technology. The semantic labels include corresponding dimensions plus subdivision values, such as identity plus stability, identity plus risk, etc. The process of acquiring transaction feature data involves feature extraction based on dimensions such as transaction frequency, amount distribution, behavior assessment, and fund flow. Based on the feature extraction results, corresponding feature intervals are set, and corresponding semantic labels are assigned based on natural language processing technology according to the corresponding feature intervals. The semantic labels include corresponding dimensions + subdivision values, such as frequency + high frequency, frequency + low frequency, amount + large amount, etc. The process of acquiring market characteristic data involves feature extraction based on dimensions such as price fluctuations, energy changes, related indicators, and public opinion trends. Based on the feature extraction results, corresponding feature intervals are set, and corresponding semantic labels are assigned based on natural language processing technology. The semantic labels include corresponding dimensions + subdivision values, such as volatility + high, volatility + stable, energy + volume increase, etc. The real-time storage unit is used to perform statistics based on the semantic tags corresponding to the data information in the financial hierarchy channel sequence, adjust the position of financial data with the same semantic tags in the financial hierarchy channel sequence according to the corresponding data collection time, and store the financial data in the multi-level financial database in real time according to the adjustment results. It should be further explained that, in the specific implementation process, when the real-time storage unit performs statistics on the corresponding semantic tags, it sets the financial level channel sequence corresponding to the largest number of identical semantic tags in the corresponding type of financial data as adjacent, and then sorts and adjusts the corresponding channels in the corresponding level according to the number of identical semantic tags, which provides convenience for subsequent data analysis and processing.

[0023] In practical implementation, the data analysis module includes: The hierarchical statistical analysis unit is used to perform statistical analysis based on the real-time storage results of semantic tags of corresponding data information in the channel sequences of each financial level corresponding to various types of financial data in the multi-level financial database. The unit sets a statistical analysis period and performs quantitative statistics on the financial data of corresponding semantic tags in each channel and level according to the sequence corresponding to the corresponding unit time within the statistical analysis period. The unit obtains the quantitative statistical data corresponding to each semantic tag in the corresponding channel and level. The quantitative statistical data includes quantity, proportion and change rate. The unit obtains the real-time analysis results corresponding to the semantic tags in the multi-level financial database. The correlation statistical analysis unit is used to perform correlation analysis on quantitative statistical data within various channels and levels of a multi-level financial database to obtain cross-correlation relationships between channels at different levels. Its specific implementation process includes: The historical analysis results obtained from the corresponding multi-level financial databases within each statistical analysis period are obtained. The historical analysis results are then combined with the quantitative statistical data of the semantic labels corresponding to each channel within the corresponding financial level channel sequence to obtain the corresponding channel semantic reference set. Correlation analysis was performed on the obtained channel semantic reference sets to obtain the correlation data corresponding to each channel semantic reference set. Specifically, the historical analysis results corresponding to the financial data with the corresponding semantic labels within the respective channel semantic reference sets were marked as follows: , where i is the label value of the corresponding semantic label financial data in the channel semantic reference set, n is the number of samples, and the corresponding correlation data is labeled as R, where: , and This represents the average of historical analysis results. ; The correlation data corresponding to each channel semantic reference set is analyzed and processed to determine whether there is a correlation between the combinations corresponding to the respective channel semantic reference sets. A correlation threshold is set based on the correlation data of each channel semantic reference set. ,like Then the corresponding combination will be marked as associated; The combined channels and semantic labels corresponding to each associated channel semantic reference set are marked in a multi-level financial database, and the hierarchical channel cross-association relationship is generated based on the marking results.

[0024] In practical implementation, the data visualization module includes: The correlation detection and verification unit is used to obtain the real-time analysis results of the semantic tags corresponding to each financial data in the multi-level financial database. It detects and verifies the real-time analysis results obtained within the corresponding statistical analysis period according to the cross-correlation relationship of the corresponding hierarchical channels, determines whether there is a correlation between the channels and the corresponding semantic tags, obtains the correlation rate of the changes of the corresponding semantic tags within the statistical analysis period, and determines whether the cross-correlation relationship of the corresponding hierarchical channels is consistent with the changes of the corresponding semantic tags. It sets a correlation rate threshold. If the correlation rate is greater than or equal to the correlation rate threshold, the corresponding semantic tags in the corresponding channels are correlated; otherwise, there is no correlation. The data visualization unit is used to acquire quantitative statistical data and corresponding detection and verification results for each semantic label within a multi-level financial database. It sets corresponding visualization charts based on the type of the quantitative statistical data and performs visualization processing. Based on the detection and verification results from the correlation detection and verification unit, it sets visualization markers corresponding to the cross-correlation relationships of the corresponding levels and channels. Based on these visualization markers, it visualizes the quantitative statistical data between corresponding semantic labels within the channels corresponding to the cross-correlation relationships of each level and channel. The obtained visualization results are then set as the financial statistical visualization database corresponding to the multi-level financial database. It should be further explained that, in the specific implementation process, the corresponding hierarchical channel cross-association relationship is set according to the detection and verification results of the corresponding associated detection and verification units, and the corresponding visualization marks are set. This provides convenience for subsequent financial data analysis by retrieving information corresponding to the needs of the relevant customers.

[0025] In specific implementation, the data retrieval module includes: The demand acquisition unit is used to acquire relevant financial demand data, including customer demand data, transaction demand data and market demand data. The acquired financial demand data is analyzed and processed based on natural language processing algorithms to obtain demand semantic tags corresponding to the financial demand data. The demand retrieval unit is used to traverse and search the financial statistics visualization database according to the corresponding demand semantic tags, obtain visualization processing results that are associated with the demand semantic tags, and generate demand retrieval information from the obtained visualization processing results.

[0026] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A financial big data analysis platform based on artificial intelligence, characterized in that, include: The data acquisition module is used to collect financial data, including financial customer data, financial transaction data, and financial market data. The module performs labeling processing on the acquired data information to obtain the labeling results of each sub-data in each type of financial data. The data processing module is used to perform multi-channel progressive analysis based on the labeling results of each sub-data in various types of financial data, to obtain the financial data and feature data corresponding to each level, and to build a multi-level financial database. The data storage module is used to perform statistics on the financial data and feature data obtained in the constructed multi-level financial database, set corresponding semantic tags based on the statistical results, and store the corresponding financial data in real time based on the set semantic tags. The data analysis module is used to perform statistical analysis on the real-time stored results corresponding to the semantic tags of the corresponding financial data in the multi-level financial database. It performs statistical analysis on the corresponding semantic tags and obtains the real-time analysis results corresponding to the corresponding semantic tags in the multi-level financial database. The data visualization module is used to visualize the real-time analysis results corresponding to each semantic tag in the multi-level financial database, and to set the multi-level financial database as a financial statistics visualization database based on the visualization results. The data retrieval module is used to acquire financial demand data, obtain corresponding demand semantic tags based on the financial demand data, and retrieve relevant financial statistical visualization databases based on the demand semantic tags to obtain relevant demand retrieval information.

2. The artificial intelligence-based financial big data analysis platform according to claim 1, characterized in that, The data acquisition module includes: The data acquisition module includes a customer acquisition unit, a transaction acquisition unit, a market acquisition unit, and an acquisition management unit; The customer acquisition unit is used to acquire a financial customer dataset, wherein the sub-data in the financial customer dataset includes financial customer data from various data acquisition sources, and the data acquisition time of the financial customer data from each data acquisition source is acquired. The transaction acquisition unit is used to acquire a financial transaction dataset, wherein the sub-data in the financial transaction dataset includes financial transaction data from various data acquisition sources, and the data acquisition time of the financial transaction data from each data acquisition source is acquired. The market acquisition unit is used to acquire a financial market dataset, wherein the sub-data in the financial market dataset includes financial market data from various data acquisition sources, and the data acquisition time of the financial market data from each data acquisition source is acquired. The data acquisition and management unit is used to mark the data according to the data acquisition source and data acquisition time of each sub-data in each type of financial data, and to obtain the marking results of each sub-data in each type of financial data.

3. The financial big data analysis platform based on artificial intelligence according to claim 2, characterized in that, The data processing module includes: The data processing module includes a multi-channel progressive analysis unit and a multi-level model building unit; The multi-channel progressive analysis unit is used to obtain each sub-data and the labeling results of each sub-data in each type of financial dataset. According to the data collection source corresponding to the labeling results of each sub-data, corresponding financial level sub-channels are set. The financial level channel includes each financial level sub-channel corresponding to the corresponding type of financial data and the corresponding type of financial data sub-data set. Based on the data collection time of each sub-data in the corresponding sub-data within the financial level channel corresponding to each type of financial data, time-series comparison processing is performed on each sub-data in the corresponding sub-data within each financial level channel, and corresponding sub-sequences are set for the financial data obtained within the same data collection time in each financial level channel. The set subsequences are integrated into the same vertical line according to the corresponding data collection time, and the set subsequences in each financial level channel are horizontally integrated to obtain the corresponding financial level channel sequence. Based on the financial hierarchy channel sequence corresponding to the financial transaction data, feature extraction is performed on the financial customer data to obtain the corresponding transaction customer feature data, and the obtained transaction customer feature data is uploaded to the financial hierarchy channel sequence corresponding to the financial transaction data. Based on the financial hierarchy channel sequence corresponding to the financial market data, feature extraction is performed on the financial transaction data to obtain the corresponding market transaction feature data, and the obtained market transaction feature data is uploaded to the financial hierarchy channel sequence corresponding to the financial market data. The multi-level model construction unit is used to obtain the financial level channel sequence corresponding to the corresponding type of financial data, perform sequence connection processing on the financial data and feature data contained in the financial level channel sequence according to the corresponding financial level channel, and construct a multi-level financial database based on the sequence connection processing result.

4. The financial big data analysis platform based on artificial intelligence according to claim 3, characterized in that, The data storage module includes: The semantic analysis unit is used to acquire the financial data and feature data corresponding to the financial level channel sequence in the multi-level financial database, extract hierarchical features from the acquired financial data and feature data, and acquire the corresponding customer feature data, transaction feature data and market feature data respectively. The acquired feature data is then processed based on natural language processing technology and assigned corresponding semantic labels. The real-time storage unit is used to perform statistics based on the semantic tags corresponding to the data information within the financial hierarchy channel sequence, adjust the position of financial data with the same semantic tag in the financial hierarchy channel sequence according to the corresponding data acquisition time, and store the financial data in the multi-level financial database in real time based on the adjustment results.

5. The artificial intelligence-based financial big data analysis platform according to claim 4, characterized in that, The data analysis module includes: The hierarchical statistical analysis unit is used to perform statistical analysis based on the real-time storage results of semantic tags of corresponding data information in the channel sequences of each financial level corresponding to various types of financial data in the multi-level financial database. The unit sets a statistical analysis period and performs quantitative statistics on the financial data of corresponding semantic tags in each channel and level according to the sequence corresponding to each unit time within the statistical analysis period. The unit obtains the quantitative statistical data corresponding to each semantic tag in the corresponding channel and level. The quantitative statistical data includes quantity, proportion and change rate. The unit obtains the real-time analysis results corresponding to each semantic tag in the multi-level financial database. The correlation statistical analysis unit is used to perform correlation analysis on quantitative statistical data of various channels and levels within a multi-level financial database to obtain cross-correlation relationships between channels at different levels.

6. The financial big data analysis platform based on artificial intelligence according to claim 5, characterized in that, The process by which the correlation statistical analysis unit obtains the hierarchical channel cross-correlation relationships includes: The historical analysis results obtained from the corresponding multi-level financial databases within each statistical analysis period are obtained. The historical analysis results are then combined with the quantitative statistical data of the semantic labels corresponding to each channel within the corresponding financial level channel sequence to obtain the corresponding channel semantic reference set. Correlation analysis was performed on the obtained channel semantic reference sets to obtain the correlation data corresponding to each channel semantic reference set; The correlation data corresponding to the semantic reference sets of each channel are analyzed and processed to determine whether there is a correlation between the combinations corresponding to the semantic reference sets of each channel. If there is a correlation, the corresponding combination is marked as related. The combined channels and semantic labels corresponding to each associated channel semantic reference set are marked in a multi-level financial database, and the hierarchical channel cross-association relationship is generated based on the marking results.

7. A financial big data analysis platform based on artificial intelligence according to claim 6, characterized in that, The data visualization module includes: The association detection and verification unit is used to obtain the real-time analysis results of the semantic tags corresponding to each financial data in the multi-level financial database. It detects and verifies the real-time analysis results obtained within the corresponding statistical analysis period according to the cross-association relationship of the corresponding hierarchical channels, determines whether there is an association between the associated channels and the corresponding semantic tags, and marks the corresponding judgment results. The data visualization unit is used to acquire quantitative statistical data and corresponding detection and verification results for each semantic label within a multi-level financial database. It sets visualization charts based on the type of the quantitative statistical data and sets visualization markers for the cross-correlation relationships of corresponding levels based on the detection and verification results of the correlation detection and verification unit. Based on these visualization markers, it visualizes the quantitative statistical data between corresponding semantic labels within the channels corresponding to the cross-correlation relationships of each level. The obtained visualization results are then set as the financial statistical visualization database corresponding to the multi-level financial database.

8. The financial big data analysis platform based on artificial intelligence according to claim 7, characterized in that, The data retrieval module includes: The demand acquisition unit is used to acquire financial demand data, which includes customer demand data, transaction demand data and market demand data. The acquired financial demand data is analyzed and processed based on natural language processing algorithms to obtain the demand semantic tags corresponding to the financial demand data. The demand retrieval unit is used to traverse and search the financial statistics visualization database based on the obtained demand semantic tags, obtain visualization processing results that are associated with the demand semantic tags, and generate demand retrieval information from the obtained visualization processing results.

Images