Multi-modal investment unified information standard model construction method and device

Abstract

The invention discloses a multi-modal investment unified information standard model construction method and device, and relates to the technical field of financial data processing, and the method comprises the steps: constructing a standardized data system through unifying the format, the standard and the coding specification of multi-source investment information data, and training a prediction model in combination with a targeted time sequence analysis method; according to the invention, the data quality and management efficiency are significantly improved, the influence of data errors on services and downstream systems is greatly reduced, the cost and error of manual data processing are effectively reduced, the prediction and risk early warning of investment market fluctuation are realized, and the effects of unifying industry data management standards, reducing investment risks and meeting supervision requirements are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of financial data processing, in particular to a multi-modal investment unified information standard model construction method and device. BACKGROUND

[0002] In the insurance industry, investment is one of the core tasks of insurance operation, which is not only an important source of insurance company's income, but also helps it to achieve asset appreciation through investment portfolio, enhance the ability to pay claims, and support the implementation of national strategies by investing in transportation, environmental protection, energy and other fields to serve the real economy.

[0003] Due to different business directions, each subsidiary of an insurance group faces different regulatory requirements, resulting in different investment data procurement and application modes for each subsidiary. At present, there is no unified investment information standard model in the industry, and each subsidiary needs to purchase and independently manage investment-related data, with data sources covering external information suppliers and in-house data. Different data sources have differences in data format, standard and definition, resulting in inconsistent data quality, which brings risks and costs to investment business; single data source lacks stability, and multi-source procurement is prone to increase costs and resource redundancy. At the same time, the group-level data is managed in a scattered manner, making it difficult to form a unified standard, increasing the difficulty of auditing, and making it impossible to carry out unified evaluation and build investment prediction models, and the multi-source data lacks an effective comparison mechanism, and data errors are difficult to discover in time, plus the daily business relies on manual data processing, with low efficiency and accuracy.

[0004] In summary, the insurance investment industry urgently needs to integrate and unify investment securities data to improve data quality and management efficiency, reduce risk costs, and meet the needs of business development. SUMMARY

[0005] To solve the above problems, the present application provides a multi-modal investment unified information standard model construction method and device, which includes the following contents:

[0006] In the first aspect, the present application provides a multi-modal investment unified information standard model construction method, which includes:

[0007] Obtaining multi-source investment information data, and performing preprocessing operations on the multi-source investment information data;

[0008] Integrating fields of the preprocessed multi-source investment information data by means of a preset tool library with Python programming language, and constructing a unified information standard model in combination with a preset unified coding specification and a pre-established model root library;

[0009] Based on the standard data output by the unified information standard model, a time series analysis method combining a historical volatility analysis model and a seasonal volatility processing model is used to train a prediction index model;

[0010] Setting a stop loss point according to the output result of the prediction index model, triggering a pre-warning when the actual index exceeds the stop loss point;

[0011] Effect evaluation and iterative optimization are carried out on the prediction index model, and the optimized prediction index model is deployed online to form an end-to-end application.

[0012] Optionally, the acquisition of multi-source investment information data comprises:

[0013] Selecting a time range, and acquiring external vendor information data and internal data corresponding to the same information according to the unique identification of securities and institutions;

[0014] The multi-source investment information data covers five categories of data commonly required in daily investment business, including stock, fund, bond, institution and general information, including securities basic information, valuation information, transaction information, issuance information and credit rating information.

[0015] Optionally, the pre-processing operation on the multi-source investment information data comprises:

[0016] Uniform the information data format from different sources, map the characteristic variables of different dimensions to the same dimension, and process repeated data. In the case that there are multiple data information for the same coded securities, the latest data is retained according to the data timestamp.

[0017] Optionally, the field integration of the multi-source investment information data after preprocessing by using Python programming language with a preset tool library, and the construction of a unified information standard model combined with a preset unified coding specification and a pre-established model root library comprises:

[0018] The preset tool library compares the field content of the multi-source information data model constructed by the multi-source investment information data, merges the fields with different field names but the same field content, supplements the transaction whole process according to the financial industry standard directory, sets the field length and field type; The preset tool library includes data processing, data analysis and machine learning related libraries, and the data processing, data analysis and machine learning related libraries are specifically pandas, matplotlib, numpy, seaborn, scikit-learn and statsmodels;

[0019] According to business needs and general investment industry guidelines, securities codes, institution codes and person codes are uniformly recoded, and a serial relationship is established with each data source data, wherein the securities code is confirmed by the securities code + transaction market + issuance date to confirm the unique value, the institution code is confirmed by the unified social credit code + institution name + establishment date to confirm the unique value, and the person code is confirmed by the name + ID card number to confirm the unique value;

[0020] The field of the uniform information standard model is translated from Chinese to English by using the Python programming language and related tool libraries for translation and syllable splitting. The English fields are abbreviated and the roots are extracted to form a uniform information standard model root library. When a new root is generated, a search is first performed in the uniform information standard model root library. If the new root is not searched, the root is included in the root library to complete the construction.

[0021] Optionally, the standard data output based on the uniform information standard model is trained by using a time series analysis method combining a historical volatility analysis model and a seasonal volatility processing model to form a prediction index model.

[0022] The standard data is preprocessed, and the preprocessing includes filling in missing values by using mean statistics, processing data affected by uncontrollable factors by using a moving average method, and processing basic data according to investment industry experience to obtain investment-related indicators, wherein the investment-related indicators include bond present values calculated based on a zero-coupon bond valuation formula and a general bond valuation formula.

[0023] The ARCH / GARCH model is used to analyze historical volatility to predict future volatility trends, and the SARIMA model is used to process seasonal volatility data. The historical volatility analysis model is specifically an ARCH / GARCH model, and the seasonal volatility processing model is specifically a SARIMA model.

[0024] The model is checked for overfitting or underfitting, and the model is optimized by adjusting parameters or using regularization methods.

[0025] Optionally, the effect evaluation and iterative optimization of the prediction index model are performed, and the optimized prediction index model is deployed online to form an end-to-end application.

[0026] The prediction index model is evaluated by using precision, recall, mean square error, mean absolute error, and business coverage rate. The parameters are adjusted and the model is iteratively optimized according to the evaluation results.

[0027] The optimized prediction index model is saved as a pkl format file, and the model is deployed by Python based on platform-integrated multi-source information data. The output interface is called by the application side.

[0028] In a second aspect, the present application provides a multi-modal investment uniform information standard model construction device, which comprises:

[0029] An acquisition unit is configured to acquire multi-source investment information data and perform a preprocessing operation on the multi-source investment information data.

[0030] The model building unit is used to integrate fields of the preprocessed multi-source investment information data using the Python programming language with the help of a preset tool library, and to build a unified information standard model by combining a preset unified coding standard with a pre-established model root library.

[0031] The prediction training unit is used to train the prediction index model based on the standard data output by the unified information standard model, using a time series analysis method that combines historical volatility analysis model and seasonal volatility processing model.

[0032] The risk warning unit is used to set a stop-loss point based on the output of the prediction indicator model, and to trigger a warning when the actual indicator exceeds the stop-loss point.

[0033] The model optimization and deployment unit is used to evaluate the effectiveness and iteratively optimize the prediction index model, and deploy the optimized prediction index model online to form an end-to-end application.

[0034] Optionally, the acquisition unit is specifically used for:

[0035] Select a time range and retrieve external vendor information data and internal data corresponding to the same information based on the unique identifiers of securities and institutions;

[0036] The multi-source investment information data covers five major categories: stocks, funds, bonds, and general information needed for daily investment business by institutions, including basic securities information, valuation information, transaction information, issuance information, and credit rating information.

[0037] Optionally, the acquisition unit is specifically used to unify the format of information data from different sources, map feature variables of different dimensions to the same dimension, and handle duplicate data by retaining the latest data according to the data timestamp when there are multiple data information for securities with the same code.

[0038] Optionally, the model building unit is specifically used to compare the field content of the multi-source information data model constructed from the multi-source investment information data using the preset tool library, merge fields with different field names but the same field content, supplement the entire transaction process with reference to the financial industry standard catalog, and set the field length and field type; the preset tool library includes data processing, data analysis and machine learning related libraries, specifically pandas, matplotlib, numpy, seaborn, scikit-learn and statsmodels;

[0039] In accordance with business needs and general investment industry standards, securities codes, institutional codes, and personal codes are uniformly recoded and linked with data from various data sources. The unique value of securities codes is confirmed by securities code + trading market + issuance date; the unique value of institutional codes is confirmed by unified social credit code + institution name + establishment date; and the unique value of personal codes is confirmed by name + ID number.

[0040] Using the Python programming language and with the help of translation and syllable splitting tool libraries, the fields of the Unified Information Standard Model are translated from Chinese to English. The English fields are abbreviated by syllable and word roots are extracted to form a Unified Information Standard Model word root library. When a new word root is generated, it is first searched in the Unified Information Standard Model word root library. If the new word root is not found, it is added to the word root library to complete the construction.

[0041] Optionally, the prediction training unit is specifically used to preprocess the standard data. The preprocessing includes filling in missing values ​​with mean statistics, processing data affected by uncontrollable factors with moving average method, and processing basic data based on investment industry experience to obtain investment-related indicators. The investment-related indicators include the present value of bonds calculated based on the zero-coupon bond valuation formula and the general bond valuation formula.

[0042] The ARCH / GARCH model is used to analyze historical volatility and predict future volatility trends, and the SARIMA model is used to process seasonal volatility data. Specifically, the historical volatility analysis model is the ARCH / GARCH model, and the seasonal volatility processing model is the SARIMA model.

[0043] Check if the model is overfitted or underfitted, and optimize the model by adjusting parameters or using regularization methods.

[0044] Optionally, the model optimization and deployment unit is specifically used to evaluate the performance of the prediction indicator model using precision, recall, mean squared error, mean absolute error, and business coverage, and to adjust the parameters and iteratively optimize the model based on the evaluation results.

[0045] The optimized predictive index model is saved as a pkl format file. Based on the multi-source information data integrated into the platform, the model is deployed using Python, and the output interface is provided for users to call.

[0046] Thirdly, this application provides an apparatus comprising a memory and a processor, the memory for storing instructions or code, and the processor for executing the instructions or code to cause the apparatus to perform the multimodal investment unified information standard model construction method described in any implementation of the first aspect.

[0047] Fourthly, this application provides a computer-readable storage medium storing code, wherein when the code is executed, a device running the code implements the multimodal investment unified information standard model construction method described in any of the implementations of the first aspect.

[0048] This application provides a method for constructing a unified information standard model for multimodal investment. When executing the method, firstly, multi-source investment information data is acquired, and preprocessing operations are performed on the multi-source investment information data. Then, using the Python programming language and a pre-defined tool library, fields of the preprocessed multi-source investment information data are integrated. A unified information standard model is constructed by combining a pre-defined unified coding standard with a pre-established model root library. Next, based on the standard data output by the unified information standard model, a predictive indicator model is trained using a time series analysis method combining a historical volatility analysis model and a seasonal volatility processing model. Finally, a stop-loss point is set based on the output results of the predictive indicator model. When the actual indicator exceeds the stop-loss point, an early warning is triggered. The predictive indicator model is then evaluated and iteratively optimized. Finally, the optimized predictive indicator model is deployed online to form an end-to-end application. By unifying the format, standards, and coding specifications of multi-source investment information data, a standardized data system is constructed. Combined with targeted time series analysis methods to train predictive models, this significantly improves data quality and management efficiency, greatly reduces the impact of data errors on business and downstream systems, effectively reduces the cost and error of manual data processing, and enables accurate prediction and risk warning of investment market fluctuations. This achieves the goals of unifying industry data management standards, reducing investment risks, and meeting regulatory requirements. Consequently, it provides insurance groups and other entities within the industry with a unified basis for investment data management and evaluation, supports the establishment of group-level investment prediction models and the efficient conduct of auditing work, promotes the standardization and dissemination of data standards in the insurance asset management industry, and provides strong technical support for the sound development of investment business. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in this embodiment or the prior art, the drawings used in the description of the embodiment or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 A flowchart illustrating a method for constructing a unified information standard model for multimodal investment, as provided in this application embodiment;

[0051] Figure 2 This is a schematic diagram of a multimodal investment unified information standard model construction device provided in an embodiment of this application. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0053] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0054] Figure 1 A flowchart illustrating a method for constructing a unified information standard model for multimodal investment, as provided in this application embodiment. (Combined with...) Figure 1 As shown in the embodiments of this application, the method for constructing a unified information standard model for multimodal investment may include:

[0055] S101. Obtain multi-source investment information data and perform preprocessing operations on the multi-source investment information data.

[0056] When acquiring multi-source investment information data, it is necessary to first select a specific time range, and then collect external vendor information data and internal data corresponding to the same information based on the unique identifiers of securities and institutions, to ensure the comprehensiveness and relevance of the data sources.

[0057] The multi-source investment information data must fully cover all business scenarios, specifically encompassing five major categories of data: stocks, funds, bonds, and general information required for daily institutional and investment businesses. It also includes core content involved in the entire transaction process, such as basic securities information, valuation information, transaction information, issuance information, and credit rating information, providing comprehensive data support for subsequent model building.

[0058] The preprocessing operations performed on the multi-source investment information data mainly include two parts: data cleaning and deduplication. During data cleaning, the formats of information data from different sources need to be standardized and unified, and feature variables of different dimensions need to be mapped to the same dimension to eliminate interference caused by differences in data format and dimension. Deduplication is handled in response to possible modifications to investment information data. The latest generated data is assumed to be the corrected data. For fields such as securities valuation information, single transaction information, and issuance information, only a single data entry should exist. If multiple data entries correspond to the same security code, the latest data entry will be retained according to the data timestamp to avoid the adverse effects of duplicate data on the subsequent unified information model design and time series analysis.

[0059] S102. Using the Python programming language and a pre-defined tool library, the pre-processed multi-source investment information data is integrated into fields, and a unified information standard model is constructed by combining the pre-defined unified coding standard and the pre-established model root library.

[0060] In the field integration stage, the preset tool library includes data processing, data analysis, and machine learning related libraries, specifically pandas, matplotlib, numpy, seaborn, scikit-learn, and statsmodels. By comparing the field content of the multi-source information data model constructed from the multi-source investment information data using the preset tool library, fields with different names but the same content are accurately identified and merged. At the same time, referring to the national financial industry standard catalog, the fields corresponding to the entire transaction process are supplemented and improved. The length and type of each field are reasonably set according to actual business needs to ensure the completeness and standardization of the model fields.

[0061] In terms of building a unified coding standard, in accordance with business needs and general investment industry standards, securities codes, institutional codes, and personnel codes are uniformly recoded and a stable connection is established with data from various data sources. Specifically, securities codes are uniquely identified by a combination of securities code + trading market + issuance date; institutional codes are uniquely identified by a combination of unified social credit code + institution name + establishment date; and personnel codes (mainly for fund managers or investment managers) are uniquely identified by a combination of name + ID number, thus achieving unified integration of multi-source data at the coding level.

[0062] In the process of constructing the model root word library, the Python programming language is used with the help of translation and syllable splitting related tool libraries to translate the Chinese fields of the Unified Information Standard Model into English. Then, the translated English fields are abbreviated according to syllables, and root words are extracted from the abbreviation results to form the Unified Information Standard Model root word library. In order to avoid root word duplication, when a new root word is generated, it must first be searched in the constructed Unified Information Standard Model root word library. Only when the new root word is not found is the root word word added to the root word library, thus completing the supplementary construction of the root word library. The whole process is executed automatically by Python tools, which effectively saves manual coding time and reduces the error rate caused by human operation.

[0063] S103. Based on the standard data output by the unified information standard model, a time series analysis method combining historical volatility analysis model and seasonal volatility processing model is used to train the predictive index model.

[0064] First, the standard data needs to be preprocessed twice. This preprocessing process includes handling null values, handling data affected by force majeure, and calculating investment-related indicators. The null value handling adopts the mean statistical imputation method, which takes the average of the data before and after the null value for the same security code to imput the null value. Compared with directly deleting the row data containing null values, this method can retain data information to the greatest extent and reduce the impact of null values ​​on the accuracy of prediction.

[0065] The data processing for the impact of force majeure events addresses the abnormal impact of uncontrollable and non-recurring events such as public health emergencies and natural disasters on securities prices. It uses the moving average method to construct a temporary model based on normal data before the impact and data after the impact to restore normal levels. The results generated by this model replace the abnormal data affected by force majeure events, thereby improving the accuracy of forecast indicators and the stability of the model.

[0066] The calculation of investment-related indicators is based on mature experience in the investment industry. The basic data is processed in a targeted manner to form reference indicators that are easy to use in business. For example, for zero-coupon bonds that do not pay interest and are redeemed at par value at maturity, the valuation formula V=M / (1+r) is used. n To calculate its present value, the general valuation formula for conventional bonds is V=C×[1-(1+r)]. -n ] / r+F×(1+r) -n Calculate its present value, where V represents the present value of the bond, M represents the face value of the zero-coupon bond, r represents the market interest rate or required rate of return, n represents the time to maturity (in years or periods), C represents the interest paid per period on a conventional bond, and F represents the face value of a conventional bond.

[0067] After data preprocessing, the ARCH / GARCH model is used as the historical volatility analysis model to predict future market volatility trends by analyzing historical volatility data. This model is more suitable for the heteroscedastic characteristics of financial markets compared to traditional time series analysis methods. Simultaneously, the SARIMA model is combined as a seasonal volatility processing model to specifically address seasonal volatility data in certain sectors of the financial market, further improving prediction accuracy. During model training, it is necessary to continuously check for overfitting or underfitting. By adjusting model parameters or using regularization methods, the interference of overfitting or underfitting on the predicted values ​​can be reduced, thus optimizing model performance.

[0068] S104. Set a stop-loss point based on the output of the prediction indicator model, and trigger an early warning when the actual indicator exceeds the stop-loss point.

[0069] The predictive indicator model trained through the aforementioned steps can output investment management indicators that reflect the fluctuation trend of the investment market. Based on these indicators, scientific and reasonable risk management strategies can be formulated, and corresponding stop-loss points can be set. During the actual investment process, changes in relevant investment indicators are monitored in real time. When the actual indicator exceeds the preset stop-loss point, a risk warning mechanism is automatically triggered, and a warning message is pushed to the investment manager, providing a basis for the investment manager to adjust the investment strategy in a timely manner and effectively avoid investment risks.

[0070] S105. Conduct performance evaluation and iterative optimization of the prediction index model, and deploy the optimized prediction index model online to form an end-to-end application.

[0071] During the performance evaluation process, precision, recall, mean squared error, mean absolute error, and business coverage are used as core evaluation indicators. The predicted values ​​output by the model are compared and analyzed with actual market data. The model parameters are precisely adjusted based on the evaluation results, and the model is iteratively optimized until a model that meets business needs, covers all business scenarios, and has the best prediction performance is generated.

[0072] During the model deployment phase, the optimized predictive indicator model is saved as a PKL format file. Based on the multi-source information data integrated into the designated platform, the model is deployed using the Python programming language to form an end-to-end application system. At the same time, corresponding interfaces are output for application parties such as investment master data systems, risk control performance platforms, and investment management systems to ensure that the model can be effectively implemented in actual business scenarios. When relevant indicators exceed the stop-loss point in actual production, the deployed system will issue timely warnings to assist investment managers in making reasonable investment adjustments and effectively ensure the steady development of investment business.

[0073] The above are some specific implementations of a method for constructing a unified information standard model for multimodal investment provided in this application. Based on this, this application also provides a corresponding apparatus. The apparatus provided in this application will be described below from the perspective of functional modularization.

[0074] Figure 2 This is a schematic diagram of a multimodal investment unified information standard model construction device provided in an embodiment of this application. (Combined with...) Figure 2 As shown in the embodiment of this application, the multimodal investment unified information standard model construction device 200 includes:

[0075] Acquisition unit 210 is used to acquire multi-source investment information data and perform preprocessing operations on the multi-source investment information data;

[0076] The model building unit 220 is used to integrate the fields of the preprocessed multi-source investment information data with the help of a preset tool library using the Python programming language, and to build a unified information standard model by combining the preset unified coding standard and the pre-established model root library.

[0077] The prediction training unit 230 is used to train the prediction index model based on the standard data output by the unified information standard model, using a time series analysis method that combines historical volatility analysis model and seasonal volatility processing model.

[0078] Risk warning unit 240 is used to set a stop loss point based on the output of the prediction indicator model, and to trigger a warning when the actual indicator exceeds the stop loss point;

[0079] The model optimization and deployment unit 250 is used to evaluate the effectiveness and iteratively optimize the prediction index model, and deploy the optimized prediction index model online to form an end-to-end application.

[0080] In one implementation of this application embodiment, the acquisition unit is specifically used for:

[0081] Select a time range and retrieve external vendor information data and internal data corresponding to the same information based on the unique identifiers of securities and institutions;

[0082] The multi-source investment information data covers five major categories: stocks, funds, bonds, and general information needed for daily investment business by institutions, including basic securities information, valuation information, transaction information, issuance information, and credit rating information.

[0083] In one implementation of this application, the acquisition unit is specifically used to unify the format of information data from different sources, map feature variables of different dimensions to the same dimension, and handle duplicate data by retaining the latest data according to the data timestamp when there are multiple data information for securities with the same code.

[0084] In one implementation of this application, the model building unit is specifically used to compare the field content of the multi-source information data model constructed from the multi-source investment information data using the preset tool library, merge fields with different field names but the same field content, supplement the entire transaction process with reference to the financial industry standard catalog, and set the field length and field type; the preset tool library includes data processing, data analysis and machine learning related libraries, specifically pandas, matplotlib, numpy, seaborn, scikit-learn and statsmodels;

[0085] In accordance with business needs and general investment industry standards, securities codes, institutional codes, and personal codes are uniformly recoded and linked with data from various data sources. The unique value of securities codes is confirmed by securities code + trading market + issuance date; the unique value of institutional codes is confirmed by unified social credit code + institution name + establishment date; and the unique value of personal codes is confirmed by name + ID number.

[0086] Using the Python programming language and with the help of translation and syllable splitting tool libraries, the fields of the Unified Information Standard Model are translated from Chinese to English. The English fields are abbreviated by syllable and word roots are extracted to form a Unified Information Standard Model word root library. When a new word root is generated, it is first searched in the Unified Information Standard Model word root library. If the new word root is not found, it is added to the word root library to complete the construction.

[0087] In one implementation of this application, the prediction training unit is specifically used to preprocess the standard data. The preprocessing includes filling in missing values ​​with mean statistics, processing data affected by uncontrollable factors with moving average method, and processing basic data based on investment industry experience to obtain investment-related indicators. The investment-related indicators include the present value of bonds calculated based on the zero-coupon bond valuation formula and the general bond valuation formula.

[0088] The ARCH / GARCH model is used to analyze historical volatility and predict future volatility trends, and the SARIMA model is used to process seasonal volatility data. Specifically, the historical volatility analysis model is the ARCH / GARCH model, and the seasonal volatility processing model is the SARIMA model.

[0089] Check if the model is overfitted or underfitted, and optimize the model by adjusting parameters or using regularization methods.

[0090] In one implementation of this application, the model optimization and deployment unit is specifically used to evaluate the performance of the prediction index model using precision, recall, mean squared error, mean absolute error, and business coverage, and to adjust the parameters and iteratively optimize the model based on the evaluation results.

[0091] The optimized predictive index model is saved as a pkl format file. Based on the multi-source information data integrated into the platform, the model is deployed using Python, and the output interface is provided for users to call.

[0092] This application also provides corresponding devices and computer storage media for implementing the solutions provided in this application.

[0093] The device includes a memory and a processor. The memory stores instructions or code, and the processor executes the instructions or code to cause the device to perform the method described in any embodiment of this application.

[0094] The computer storage medium stores code, and when the code is run, the device running the code implements the method described in any embodiment of this application.

[0095] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that all or part of the steps in the methods of the above embodiments can be implemented by means of software plus a general-purpose hardware platform. Based on this understanding, the technical solution of this application can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as a read-only memory (ROM) / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, a server, or a network communication device such as a router) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0096] It is understood that in the specific embodiments of this application, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved need to obtain user permission or consent when the above embodiments of this application are applied to specific products or technologies, and the collection, use and processing of related data need to comply with the relevant laws, regulations and standards of relevant countries and regions.

[0097] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0098] It should also be noted that the various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for the device and apparatus embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiments. The device and apparatus embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components indicated as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the solution in this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0099] The above description is merely one specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for constructing a unified information standard model for multimodal investment, characterized in that, The method includes: Acquire multi-source investment information data and perform preprocessing operations on the multi-source investment information data; The Python programming language is used to integrate the fields of the preprocessed multi-source investment information data with the help of a preset tool library, and a unified information standard model is constructed by combining the preset unified coding standard and the pre-established model root library. Based on the standard data output by the unified information standard model, a time series analysis method combining historical volatility analysis model and seasonal volatility processing model is used to train the predictive index model. A stop-loss point is set based on the output of the prediction indicator model. An early warning is triggered when the actual indicator exceeds the stop-loss point. The predictive index model is evaluated and iteratively optimized, and the optimized predictive index model is deployed online to form an end-to-end application.

2. The method according to claim 1, characterized in that, The acquisition of multi-source investment information data includes: Select a time range and retrieve external vendor information data and internal data corresponding to the same information based on the unique identifiers of securities and institutions; The multi-source investment information data covers five major categories: stocks, funds, bonds, and general information needed for daily investment business by institutions, including basic securities information, valuation information, transaction information, issuance information, and credit rating information.

3. The method according to claim 1, characterized in that, The preprocessing operation on the multi-source investment information data includes: Unify the format of information data from different sources and map feature variables of different dimensions to the same dimension; for duplicate data processing, when there are multiple data information for the same security with the same code, retain the latest data according to the data timestamp.

4. The method according to claim 1, characterized in that, The process of using the Python programming language with the aid of a pre-defined tool library to integrate fields of the pre-processed multi-source investment information data, and constructing a unified information standard model by combining a pre-defined unified coding standard and a pre-established model root library, includes: The preset tool library is used to compare the field content of the multi-source information data model constructed from the multi-source investment information data, merge fields with different field names but the same field content, supplement the entire transaction process with reference to the financial industry standard catalog, and set the field length and field type; the preset tool library includes data processing, data analysis and machine learning related libraries, specifically pandas, matplotlib, numpy, seaborn, scikit-learn and statsmodels; In accordance with business needs and general investment industry standards, securities codes, institutional codes, and personal codes are uniformly recoded and linked with data from various data sources. The unique value of securities codes is confirmed by securities code + trading market + issuance date; the unique value of institutional codes is confirmed by unified social credit code + institution name + establishment date; and the unique value of personal codes is confirmed by name + ID number. Using the Python programming language and with the help of translation and syllable splitting tool libraries, the fields of the Unified Information Standard Model are translated from Chinese to English. The English fields are abbreviated by syllable and word roots are extracted to form a Unified Information Standard Model word root library. When a new word root is generated, it is first searched in the Unified Information Standard Model word root library. If the new word root is not found, it is added to the word root library to complete the construction.

5. The method according to claim 1, characterized in that, The training of the predictive index model based on the standard data output by the unified information standard model, using a time series analysis method combining historical volatility analysis and seasonal volatility processing models, includes: The standard data is preprocessed, including filling missing values ​​with mean statistics, processing data affected by uncontrollable factors using the moving average method, and processing basic data based on investment industry experience to obtain investment-related indicators. The investment-related indicators include the present value of bonds calculated based on the zero-coupon bond valuation formula and the general bond valuation formula. The ARCH / GARCH model is used to analyze historical volatility and predict future volatility trends, and the SARIMA model is used to process seasonal volatility data. Specifically, the historical volatility analysis model is the ARCH / GARCH model, and the seasonal volatility processing model is the SARIMA model. Check if the model is overfitted or underfitted, and optimize the model by adjusting parameters or using regularization methods.

6. The method according to claim 1, characterized in that, The step of evaluating and iteratively optimizing the prediction index model, and then deploying the optimized prediction index model online to form an end-to-end application includes: The performance of the prediction index model is evaluated using precision, recall, mean squared error, mean absolute error, and business coverage. The parameters are adjusted and the model is iteratively optimized based on the evaluation results. The optimized predictive index model is saved as a pkl format file. Based on the multi-source information data integrated into the platform, the model is deployed using Python, and the output interface is provided for users to call.

7. A device for constructing a unified information standard model for multimodal investment, characterized in that, The device includes: The acquisition unit is used to acquire multi-source investment information data and perform preprocessing operations on the multi-source investment information data. The model building unit is used to integrate fields of the preprocessed multi-source investment information data using the Python programming language with the help of a preset tool library, and to build a unified information standard model by combining a preset unified coding standard with a pre-established model root library. The prediction training unit is used to train the prediction index model based on the standard data output by the unified information standard model, using a time series analysis method that combines historical volatility analysis model and seasonal volatility processing model. The risk warning unit is used to set a stop-loss point based on the output of the prediction indicator model, and to trigger a warning when the actual indicator exceeds the stop-loss point. The model optimization and deployment unit is used to evaluate the effectiveness and iteratively optimize the prediction index model, and deploy the optimized prediction index model online to form an end-to-end application.

8. The apparatus according to claim 7, characterized in that, The acquisition unit is specifically used for: Select a time range and retrieve external vendor information data and internal data corresponding to the same information based on the unique identifiers of securities and institutions; The multi-source investment information data covers five major categories: stocks, funds, bonds, and general information needed for daily investment business by institutions, including basic securities information, valuation information, transaction information, issuance information, and credit rating information.

9. A computing device, characterized in that, The computing device includes: a memory and a processor; The memory is used to store computer programs; The processor is configured to implement the steps of the method as described in any one of claims 1 to 6 when executing the computer program.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method as described in any one of claims 1 to 6.

Images