Emergency pulse size determination method and device, medium and program product

Abstract

The invention provides an emergency pulse size determination method and device, a medium and a program product, and relates to the field of artificial intelligence. The emergency pulse size determination method comprises the steps of obtaining financial data corresponding to a target emergency, and extracting features of the financial data to obtain feature data; and inputting the feature data into a pre-trained pulse size determination model, and determining the emergency pulse size of the target emergency according to the feature data by the pulse size determination model based on an integrated cue word project. The cue word project is used for guiding the pulse size determination model to determine the pulse size of the emergency, enhancing the prediction capability of the pulse size determination model and improving the precision and processing efficiency of the pulse size determination model, so that the accuracy and generation efficiency of the pulse size of the emergency are improved, and the resource consumption of the system is reduced. The generalization ability of the pulse size determination model obtained through pre-training is higher, the method adapts to different emergencies and market environments, and the prediction precision is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and in particular to a method, device, medium and program product for determining the size of a sudden event pulse. Background Art

[0002] In financial risk management, forward-looking analysis is an important tool for predicting future risks. Among them, the impulse response function is widely used to evaluate the impact of sudden events on time series data. The sudden event pulse refers to the short and strong changes caused by simulating sudden events in time series analysis. The sudden event pulse is usually manifested in the form of a sudden change in a certain variable (such as transaction frequency, number of new accounts opened, etc.) at a certain moment, and is used to study the subsequent impact of this mutation on the entire system or other variables. Therefore, how to determine the size of the sudden event pulse is crucial.

[0003] In the related art, the pulse size of the sudden event is manually set, or the pulse size of the sudden event is automatically determined according to simple rules. However, the above methods have the problem of inaccuracy. Summary of the invention

[0004] The present application provides a method, device, medium and program product for determining the size of a sudden event pulse, so as to solve the technical problem of inaccurate size of a sudden event pulse.

[0005] In a first aspect, the present application provides a method for determining the size of a burst pulse, comprising:

[0006] Obtain financial data corresponding to the target emergency event;

[0007] Extract the features of financial data to obtain feature data;

[0008] The feature data is input into a pre-trained pulse size determination model. The pulse size determination model is based on an integrated prompt word engineering (referred to as Prompt) and determines the sudden event pulse size of the target sudden event according to the feature data.

[0009] In a second aspect, the present application provides a device for determining the size of an emergency pulse, comprising:

[0010] An acquisition module, used to acquire financial data corresponding to a target emergency event;

[0011] An extraction module is used to extract the features of financial data to obtain feature data;

[0012] The processing module is used to input the characteristic data into a pre-trained pulse size determination model. The pulse size determination model is based on the integrated Prompt and obtains the sudden event pulse size of the target sudden event according to the characteristic data.

[0013] In a third aspect, the present application provides an electronic device, comprising: a processor, and a memory communicatively connected to the processor;

[0014] Memory stores computer-executable instructions;

[0015] The processor executes the computer-executable instructions stored in the memory to implement any method of the first aspect.

[0016] In a fourth aspect, the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed, they are used to implement any method as in the first aspect.

[0017] In a fifth aspect, the present application provides a computer program product, including a computer program, which implements the method of any one of the first aspects when executed.

[0018] The method, device, medium and program product for determining the pulse size of an emergency event provided by the present application obtains the financial data corresponding to the target emergency event, extracts the features of the financial data, and obtains feature data; the feature data is input into the pulse size determination model obtained by pre-training, and the pulse size determination model is based on the integrated Prompt, and determines the emergency pulse size of the target emergency event according to the feature data. Among them, Prompt is used to guide the pulse size determination model to determine the emergency pulse size, thereby enhancing the prediction ability of the pulse size determination model, improving the accuracy and processing efficiency of the pulse size determination model, thereby improving the accuracy and generation efficiency of the emergency pulse size, and reducing the resource consumption of the system. In addition, the pulse size determination model obtained through pre-training has a stronger generalization ability, adapts to different emergencies and market environments, and further improves the prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0020] Figure 1 A schematic diagram of a scenario of a method for determining the size of an emergency pulse provided in an embodiment of the present application;

[0021] Figure 2 A flow chart of a method for determining the size of an emergency pulse provided in an embodiment of the present application;

[0022] Figure 3 A technical flow chart of a method for determining the size of an emergency pulse provided in an embodiment of the present application;

[0023] Figure 4A schematic diagram of the structure of a device for determining the size of an emergency pulse provided in an embodiment of the present application;

[0024] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0025] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0026] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0027] 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, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with the relevant laws, regulations and standards of the relevant countries and regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0028] In addition, this application involves conducting big data analysis of user information (including but not limited to personal biometrics, identity data, consumption data, asset data, electronic terminal operation data, etc.), and using artificial intelligence technology to make automated decisions, and making technical solutions that have a significant impact on personal rights and interests based on the results of automated decisions. The application provides users with corresponding operation entrances for them to choose to agree or reject the results of automated decisions; if the user chooses to reject, the expert decision-making process will be entered.

[0029] It should be noted that the method, device, medium and program product for determining the size of a sudden event pulse provided in the present application can be used in the field of artificial intelligence, and can also be used in any field other than the field of artificial intelligence. The application field of the method, device, medium and program product for determining the size of a sudden event pulse in the present application is not limited.

[0030] In the related art, it is usually dependent on the artificially set pulse size of sudden events, which has a large workload, many reference contents, high labor costs, and strong subjectivity; or, the pulse size of sudden events is determined based on simple rules, which lacks flexibility and accuracy.

[0031] Based on the above technical problems, the present application provides a method, device, medium and program product for determining the pulse size of a sudden event, aiming to analyze the characteristic data of sudden events through an artificial intelligence model, and based on an integrated prompt, automatically generate a reasonable sudden event pulse size, thereby improving the accuracy and generation efficiency of the sudden event pulse size.

[0032] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0033] Figure 1 A schematic diagram of a scenario of a method for determining the size of an emergency pulse provided in an embodiment of the present application. Figure 1 The collection device 11 collects the financial data corresponding to the target emergency event, and extracts the features of the financial data through the data processing device 12 to obtain feature data, and then inputs the feature data into the computing device 13 deployed with the trained pulse size determination model. The computing device 13 analyzes and predicts the feature data through the pulse size determination model and the integrated Prompt, thereby outputting the emergency event pulse size of the target emergency event.

[0034] It should be noted that Figure 1 The application scenarios shown are only examples. Among them, any two or three of the acquisition device, data processing device and computing device can be the same device, and there is no limit on the number of various devices. For example, the financial data collected by the same acquisition device can be processed by multiple data processing devices to obtain the characteristic data of the target emergency event; or, for the same characteristic data, for example, there are multiple target emergency events, multiple computing devices can be used to analyze and predict, and the emergency event pulse size corresponding to the corresponding target emergency event can be output respectively, etc., which needs to be set according to actual needs.

[0035] in addition, Figure 1 The acquisition device 11, data processing device 12 and computing device 13 in the system can also be a smart phone, a wearable device, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a server cluster, and the like.

[0036] Figure 2A flow chart of a method for determining the size of an emergency pulse provided in an embodiment of the present application. Figure 2 The method for determining the size of the sudden event pulse comprises the following steps:

[0037] S201. Obtain financial data corresponding to a target emergency event.

[0038] For example, financial data corresponding to the target emergency event is collected from data sources such as financial databases, social media, forums, news websites or financial media. Financial data can reflect the state of the market, such as the reaction of the stock market, bond market, foreign exchange market, etc. Financial data can also reflect the impact of the event, such as the impact on economic indicators (interest rates, inflation rates, employment data, unemployment rates, etc.), the impact on trading volume, current market prices or volatility, etc.

[0039] The data types of financial data include, but are not limited to, upward rate, deterioration downward rate, GDP growth rate, unemployment rate, market prices (stock prices, commodity prices, etc.), transaction volume (number of transactions, frequency, average single transaction amount), number of new accounts opened, transaction cycle, news events (important news involving financial markets), loan default rate (an indicator reflecting loan defaults), etc.

[0040] Target emergencies are events that may affect the development trend of financial market data, such as changes in economic policies, financial news, interest rate changes, market crashes, etc.

[0041] S202: Extract features of financial data to obtain feature data.

[0042] For example, based on the preset feature rules, data analysis tools are used to extract features of financial data, for example, extracting the time features, market indicator features, macroeconomic indicator features, sentiment features, event features, etc. of financial data. Among them, market indicator features are, for example, current market prices, volatility, trading volume, etc., macroeconomic indicator features are, for example, the latest gross domestic product (GDP) growth rate, unemployment rate, interest rate changes, etc., sentiment features are, for example, the number of positive and negative news and news popularity corresponding to real-time news data, etc., and event features are, for example, descriptions and classifications of new events (policy changes, market crashes, etc.).

[0043] Optionally, feature rules can be selected based on specific target emergencies to ensure that the extracted features have practical significance and predictive power.

[0044] S203, inputting the characteristic data into a pre-trained pulse size determination model, the pulse size determination model is based on the integrated Prompt, and determines the emergency event pulse size of the target emergency event according to the characteristic data.

[0045] The pulse size determination model can be understood as a large model, for example, a large model suitable for time series analysis, which can effectively predict the pulse size of sudden events based on the input. The pulse size determination model is usually trained on large-scale datasets, can handle complex tasks and generate high-quality results.

[0046] Prompt can help (guide) the big model to better understand the characteristic data, better predict the pulse size of sudden events, and affect the output quality and relevance of the big model.

[0047] For example, Prompt includes historical emergencies and their corresponding emergencies pulse sizes. When a new emergencies is input into the pulse size determination model, the pulse size determination model can quickly generate the required emergencies pulse size based on the guidance of Prompt. As an example only, Prompt: "Historical events: Financial crisis 2008, features: market prices fell sharply, economic indicators fluctuated greatly, emergencies pulse size: 0.8". It should be noted that this is only an example and does not constitute a specific limitation on Prompt.

[0048] The pulse size of a sudden event can be the amplitude of price changes, changes in trading volume, increase in market volatility, etc.

[0049] In the embodiment of the present application, the pulse size of the emergency event is determined based on the prompt-guided pulse size determination model, thereby enhancing the prediction ability of the pulse size determination model, improving the accuracy and processing efficiency of the pulse size determination model, and improving the accuracy and generation efficiency of the pulse size of the emergency event. Since prompt accelerates the prediction speed of the pulse size determination model and reduces the number of iterations, the resource consumption of the system is reduced to a certain extent. In addition, the pulse size determination model obtained by pre-training has a stronger generalization ability, adapts to different emergencies and market environments, and further improves the prediction accuracy. Compared with manually setting the pulse size of the emergency event, the embodiment of the present application automatically generates the pulse size of the emergency event through a large model, reduces the labor cost, reduces the subjectivity of the manual setting, enhances the objectivity of the pulse size of the emergency event, and then improves the accuracy of the pulse size of the emergency event. Compared with the manual setting method, the processing efficiency is also improved; in addition, the pulse size determination model can quickly process and analyze the input data, generate the pulse size of the emergency event in real time, and provide timely response for subsequent risk management and investment decisions, which improves safety to a certain extent.

[0050] In some embodiments, the pulse size determination model is based on an integrated Prompt, and determines the sudden event pulse size of the target sudden event according to the characteristic data, including: the pulse size determination model performs feature mapping on the characteristic data to obtain historical sudden events associated with the characteristic data; determines a first pulse size based on the pre-learned correspondence between historical sudden events and sudden event pulse sizes, and the guidance information in the Prompt; and, based on a preset second pulse size, calibrates and optimizes the first pulse size to obtain the sudden event pulse size of the target sudden event.

[0051] For example, the pulse size determination model uses cluster analysis, dimensionality reduction technology, etc. to map the characteristic data of the target emergency event to the historical emergency events, and identify the historical emergency events similar to the target emergency event. Then, the pulse size of the target emergency event can be preliminarily determined by using the correspondence between the historical emergency events and the pulse size of the emergency event learned in advance. In the determination process, combined with the guidance information in the prompt, the pulse size determination model is helped to better understand and predict the pulse size of the emergency event. In other words, based on the pulse size determination model and the prompt information, the first pulse size of the target emergency event can be predicted.

[0052] In order to further improve the accuracy and rationality of the first pulse size, the first pulse size can be adjusted according to market experience, expert opinions, or the pulse size of an emergency event set by rules. For example, according to market experience or expert opinions or the pulse size of an emergency event set by rules, an expected second pulse size is set, the first pulse size is compared with the second pulse size, the difference is identified, and then the first pulse size obtained by preliminary prediction is adjusted through an optimization algorithm (such as Bayesian optimization, genetic algorithm), regression algorithm, etc., so that it is closer to the preset second pulse size.

[0053] Optionally, the parameters of the pulse size determination model may be adjusted according to the second pulse size so that the first pulse size outputted by the model is closer to the preset second pulse size.

[0054] In the embodiment of the present application, through feature mapping, the pulse size determination model can identify the similarities between the target emergency and the historical emergency, which helps to utilize the patterns and laws in the historical data to enhance the understanding of the current event. Through effective prompts, the accuracy and relevance of the pulse size determination model prediction are improved. The calibration and optimization mechanism helps to improve the reliability and accuracy of the prediction, making it more consistent with market expectations and empirical judgments. In addition, accurate prediction results can provide strong support for risk management and investment decisions in the financial market.

[0055] A well-designed prompt can help the large model better understand the task requirements and generate more accurate and useful results, so the determination of the prompt is crucial.

[0056] In some embodiments, Prompt is determined by a deep learning model, and / or Prompt includes expert rules for the correspondence between burst events and burst event pulse sizes.

[0057] In one example, Prompt is determined by a deep learning model.

[0058] By training the deep learning model, the deep learning model can generate prompts related to specific emergencies, ensuring that the prompts can capture the key features of the event, thereby accelerating the prediction speed of the pulse size determination model. Optionally, the deep learning model can also dynamically generate and adjust prompts based on real-time data and market changes.

[0059] In another example, Prompt includes expert rules for the correspondence between burst events and burst event pulse sizes.

[0060] Among them, expert rules may include market reactions to similar events in history, changing patterns of economic indicators, etc. They may also include the correspondence between emergencies and the size of emergency pulses formulated based on human experience.

[0061] To further enrich the guidance information of the prompt, in another example, the expert rules can be combined with the prompt generated by the deep learning model. Human judgment is introduced on the basis of deep learning model processing. Deep learning provides data-driven insights, while expert rules provide experience and judgment to improve the robustness and reliability of the system.

[0062] In the embodiment of the present application, the construction of Prompt is combined with deep learning models and / or expert rules, so that more accurate and reliable prediction of the pulse size of sudden events can be provided in a complex financial environment. In addition, due to the combination of expert experience, the prediction results are easier for users to understand and trust, thereby improving the adoption rate in practical applications.

[0063] In some embodiments, the first pulse size is calibrated and optimized based on a preset second pulse size to obtain the burst event pulse size of the target burst event, including: based on the preset second pulse size, adjusting the first pulse size according to a regression algorithm to obtain the burst event pulse size of the target burst event.

[0064] For example, the regression algorithm includes linear regression, ridge regression, decision tree regression or random forest regression, etc. The regression algorithm can predict the adjustment amount of the first pulse size according to the second pulse size, thereby obtaining the burst pulse size of the target burst event.

[0065] It should be noted that a suitable regression algorithm can be selected according to the complexity of the data feature (the data relationship between the second pulse size and the first pulse size). For example, when the data relationship between the first pulse size and the second pulse size is relatively simple and linear, a linear regression algorithm can be used.

[0066] In the embodiment of the present application, the first pulse size is automatically adjusted through a regression algorithm to better respond to the needs of the target emergency event, improve the accuracy of the emergency event pulse size, and reduce errors caused by human intervention.

[0067] In some embodiments, financial data includes financial market data and emergency event data, and features of the financial data are extracted to obtain feature data, including: extracting calendar effects and seasonal features from financial market data to obtain time features; extracting volatility, trading volume and market depth from financial market data to obtain market indicator features; extracting GDP growth rate, inflation rate and interest rate changes from financial market data to obtain macroeconomic indicator features; extracting market sentiment features from emergency event data to obtain sentiment features; and extracting event features from emergency event data to obtain event features.

[0068] Among them, calendar effects such as month-end effect, weekend effect, and holiday effect can be used to extract market performance at different weekends and month-ends, and market fluctuations before and after holidays. Seasonal characteristics, for example, market trends and changes in each quarter, or cyclical changes in the market in different years.

[0069] Volatility, such as implied volatility and historical volatility; volume, such as average volume and volume change rate; market depth, such as the gap between buy and sell orders and market liquidity, and the structure and changes of the order book.

[0070] GDP growth rate, such as quarterly or annual GDP growth data; inflation rate, such as consumer price index, producer price index; interest rate changes, such as central bank interest rate decisions, bond yield curve.

[0071] Extract market sentiment features such as the number of positive and / or negative news, news popularity, etc. from real-time news data (breaking event data).

[0072] Extract descriptions and classifications of emergencies from emergency data, such as financial crises, policy changes, market crashes, etc.

[0073] It should be noted that, in actual applications, when extracting multiple dimensional features from the acquired financial data, if a feature of a certain dimension does not exist, the value corresponding to the feature of that dimension may be assigned to empty.

[0074] The embodiment of the present application extracts the features of financial data from different dimensions to construct a more comprehensive feature set for input into the pulse size determination model, which can significantly enhance the prediction capability of the pulse size determination model. In addition, the integration of multi-dimensional feature data can improve the sensitivity and prediction accuracy of the pulse size determination model to emergencies.

[0075] The above method embodiment is an application of a pulse size determination model. Next, how to train and obtain the pulse size determination model is described through a specific embodiment.

[0076] In some embodiments, the pulse size determination model is obtained by:

[0077] Step 1.1: Obtain training samples, which include financial data corresponding to sample events and the pulse size of sudden events.

[0078] For example, the financial data and the pulse size of the emergency event corresponding to the sample event are obtained from financial databases, news sources, etc. The data types include but are not limited to the upward rate, the deterioration downward rate, economic indicators (such as GDP growth rate, unemployment rate, etc.), market prices (stock prices, commodity prices, etc.), transaction volume (number of transactions, frequency, average single transaction amount), number of new accounts, transaction cycle, news events (important news involving financial markets), loan default rate (indicator reflecting loan default). It should be noted that the selection of sample events can cover historical emergencies of different types and sizes to enhance the diversification of training samples.

[0079] Step 1.2: Preprocess the financial data, including data cleaning, denoising and standardization.

[0080] For example, data cleaning includes handling missing values ​​and outliers in financial data, denoising includes using filtering techniques (such as moving average, Kalman filtering) to remove noise from the data, and standardization includes standardizing the data to ensure that data with different characteristics are analyzed on the same scale.

[0081] Step 1.3: Extract features from the preprocessed data to obtain sample features.

[0082] Sample characteristics include time characteristics, market indicator characteristics, macroeconomic indicator characteristics, sentiment characteristics and event characteristics. Specifically, calendar effect and seasonal characteristics are extracted from preprocessed data to obtain time characteristics; volatility, trading volume and market depth are extracted from preprocessed data to obtain market indicator characteristics; GDP growth rate, inflation rate and interest rate changes are extracted from preprocessed data to obtain macroeconomic indicator characteristics; market sentiment characteristics are extracted from preprocessed data to obtain sentiment characteristics; event characteristics are extracted from preprocessed data to obtain event characteristics.

[0083] Step 1.4: Input the sample features into the pulse size determination model to obtain the predicted emergency pulse size corresponding to the financial data.

[0084] Among them, the pulse size determination model can be a model suitable for time series analysis, such as one built based on Transformer (which has advantages in parallel computing and long-term dependency processing), or a time series model, etc., and this application does not impose any restrictions on this.

[0085] The sample features obtained in step 1.3 are input into the pulse size determination model, and the pulse size determination model outputs the predicted burst event pulse size corresponding to each sample event.

[0086] Step 1.5: Based on the sudden event pulse size and the predicted sudden event pulse size, the model parameters of the pulse size determination model are adjusted through cross-validation method and parameter tuning method.

[0087] For example, the training sample set is divided into multiple subsets for training and testing, so as to obtain the stability and generalization ability of the pulse size determination model. The parameter tuning method refers to systematically adjusting the hyperparameters of the pulse size determination model to find the parameter combination that can provide the best prediction performance. The cross-validation method and the parameter tuning method are not described here.

[0088] Optionally, during the training process of the pulse size determination model, a prompt can also be introduced to guide the training of the pulse size determination model. The generated predicted emergency pulse size can also be compared with the pulse size set by human experience or rules, and the model parameters of the pulse size determination model can be further adjusted to enhance the reliability and robustness of the pulse size determination model.

[0089] The embodiment of the present application trains a pulse size determination model by using financial data corresponding to a large number of different historical emergencies and the pulse sizes of emergencies, and combines the cross-validation method and the parameter tuning method to effectively adjust the model parameters of the pulse size determination model, thereby improving the generalization ability and accuracy of the pulse size determination model and better adapting to different types of emergencies.

[0090] In financial risk management, impulse response functions are widely used to evaluate the impact of emergencies on time series data. The impulse size of the emergency event is applied to the impulse response function to simulate the impulse response. In addition, during the application stage of the impulse size of the emergency event, the impulse size determination model can also be optimized.

[0091] In some embodiments, after obtaining the burst pulse size of the target burst event, the following steps may also be included:

[0092] Step 2.1: Input the pulse size of the emergency event into the financial market time series model to obtain the prediction result of the impact of the target emergency event on the future trend of financial market data. The financial market time series model is constructed based on the impulse response function.

[0093] Among them, the impulse response function can analyze the impact of the pulse size of sudden events on market data (such as price and trading volume), and the financial market time series model is used to analyze and predict the trends, cycles and fluctuations of financial market data.

[0094] The financial market time series model can be constructed through the vector autoregression model VAR or the vector error correction model VECM, and the impulse response function, that is, the impulse response function is combined with VAR, or the impulse response function is combined with VECM to complete the construction of the financial market time series model.

[0095] It can be understood that step 2.1 is to apply the obtained emergency event pulse size of the target emergency event to the time series data to perform impulse response analysis, so as to evaluate the short-term and long-term impact of the target emergency event on the market data of concern.

[0096] Step 2.2: Optimize the pulse size determination model based on the forecast results and actual financial market data.

[0097] Compare the prediction results with the actual financial market data to evaluate the accuracy of the sudden event pulse size generated by the pulse size determination model. If there is a significant inconsistency between the prediction results and the actual financial market data, it means that the sudden event pulse size generated by the pulse size determination model is inaccurate. The pulse size determination model is optimized through optimization algorithms, such as error correction mechanisms.

[0098] The pulse size determination model is continuously iterated according to market changes and new data to ensure that it always remains efficient and accurate, significantly improving the predictive performance of the pulse size determination model and making it better adaptable to the complex dynamics of the financial market and the impact of unexpected events.

[0099] In the embodiment of the present application, by inputting the pulse size of the emergency event into the financial market time series model, it is helpful to accurately predict the impact of the target emergency event on the future trend of financial market data. In addition, if it is found that the prediction result deviates from the actual financial market data, the pulse size determination model is continuously optimized, so that the performance of the pulse size determination model prediction is continuously improved.

[0100] Next, a method for determining the size of an emergency pulse is further described by using a specific embodiment. Figure 3 This is a technical flow chart of the method for determining the size of an emergency pulse provided in an embodiment of the present application. Figure 3 As shown, the method for determining the size of the sudden event pulse includes:

[0101] Step 3.1. New emergencies.

[0102] Obtaining financial data corresponding to the new emergency event: For example, collecting financial data corresponding to the target emergency event from data sources such as financial databases, social media, forums, news websites, or financial media, see S201 for details.

[0103] Step 3.2: Feature extraction.

[0104] The characteristics of the financial data corresponding to the new emergency events are extracted to obtain characteristic data.

[0105] For example, financial data includes financial market data and emergency event data. Calendar effects and seasonal characteristics are extracted from financial market data to obtain time characteristics; volatility, trading volume and market depth are extracted from financial market data to obtain market indicator characteristics; GDP growth rate, inflation rate and interest rate changes are extracted from financial market data to obtain macroeconomic indicator characteristics; market sentiment characteristics are extracted from emergency event data to obtain sentiment characteristics; event characteristics are extracted from emergency event data to obtain event characteristics.

[0106] Step 3.3, Prompt build.

[0107] For example, based on historical emergencies, a Prompt is constructed, which includes a corresponding relationship between historical emergencies and the emergencies pulse size of the historical emergencies.

[0108] Step 3.4: New pulse size.

[0109] The large model framework determines the size of the newborn pulse based on the prompt and feature data.

[0110] The large model framework can be regarded as a pre-trained pulse size determination model. For details on determining the size of the new pulse, see S203.

[0111] Step 3.5: Calibration and optimization.

[0112] Calibrate and optimize the size of the newborn pulse.

[0113] For example, the size of the new pulse may be adjusted according to market experience, expert opinion, or the size of the emergency pulse set by rules.

[0114] Step 3.6. Financial market data.

[0115] Apply the calibrated and optimized nascent impulse size to financial market data.

[0116] Apply the obtained new pulse size to the time series data to perform impulse response analysis, thereby evaluating the short-term and long-term impact of the target sudden event on the market data of interest. For details, see step 2.1.

[0117] In summary, this application has at least the following advantages:

[0118] 1. Based on the prompt-guided pulse size determination model, the pulse size of the emergency event is determined, thereby enhancing the prediction ability of the pulse size determination model, improving the accuracy and processing efficiency of the pulse size determination model, and improving the accuracy and generation efficiency of the pulse size of the emergency event. Since prompt accelerates the prediction speed of the pulse size determination model and reduces the number of iterations, the resource consumption of the system is reduced to a certain extent. In addition, the pulse size determination model obtained by pre-training has a stronger generalization ability, adapts to different emergencies and market environments, and further improves the prediction accuracy. Compared with manually setting the pulse size of the emergency event, the embodiment of the present application automatically generates the pulse size of the emergency event through a large model, reduces labor costs, reduces the subjectivity of manual setting, enhances the objectivity of the pulse size of the emergency event, and thus improves the accuracy of the pulse size of the emergency event. Compared with the manual setting method, the processing efficiency is also improved; in addition, the pulse size determination model can quickly process and analyze the input data, generate the pulse size of the emergency event in real time, and provide timely response for subsequent risk management and investment decisions, and improve safety to a certain extent.

[0119] Second, calibrating and optimizing the pulse size of emergencies can help improve the reliability and accuracy of the pulse size determination model prediction, making it more consistent with market expectations and empirical judgments. In addition, accurate prediction results can provide strong support for risk management and investment decisions in the financial market.

[0120] Third, the construction of Prompt combines deep learning models and / or expert rules to provide more accurate and reliable predictions of the size of sudden events in complex financial environments. In addition, due to the combination of expert experience, the prediction results are easier for users to understand and trust, thereby increasing the adoption rate in practical applications.

[0121] Fourth, by extracting the features of financial data from different dimensions and constructing a more comprehensive feature set for the input of the pulse size determination model, the prediction ability of the pulse size determination model can be significantly enhanced; in addition, the integration of multi-dimensional feature data can improve the sensitivity and prediction accuracy of the pulse size determination model to emergencies.

[0122] 5. The pulse size determination model is trained by using financial data corresponding to a large number of different historical emergencies and the pulse sizes of emergencies, and combined with cross-validation and parameter tuning methods, the model parameters of the pulse size determination model can be effectively adjusted, the generalization ability and accuracy of the pulse size determination model can be improved, and it can better adapt to different types of emergencies.

[0123] 6. By inputting the pulse size of the emergency event into the financial market time series model, it is helpful to accurately predict the impact of the target emergency event on the future trend of financial market data. In addition, if it is found that the prediction result deviates from the actual financial market data, the pulse size determination model will continue to be optimized, so that the performance of the pulse size determination model prediction can be continuously improved.

[0124] The following are device embodiments of the present application, which can be used to execute the method embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the present application.

[0125] Figure 4 This is a schematic diagram of the structure of the device for determining the size of an emergency pulse provided in an embodiment of the present application. Figure 4 As shown, the device 400 for determining the size of an emergency pulse includes an acquisition module 401, an extraction module 402 and a processing module 403. Among them:

[0126] An acquisition module 401 is used to acquire financial data corresponding to a target emergency event;

[0127] An extraction module 402 is used to extract features of financial data to obtain feature data;

[0128] The processing module 403 is used to input the characteristic data into a pre-trained pulse size determination model. The pulse size determination model is based on the integrated Prompt and obtains the burst pulse size of the target burst according to the characteristic data.

[0129] In one possible implementation, the processing module 403 is specifically used to: perform feature mapping on the characteristic data using a pulse size determination model to obtain historical emergency events associated with the characteristic data; determine a first pulse size based on the pre-learned correspondence between historical emergency events and emergency event pulse sizes, and guidance information in the Prompt; and, based on a preset second pulse size, calibrate and optimize the first pulse size to obtain the emergency event pulse size of the target emergency event.

[0130] In a possible implementation, Prompt is determined by a deep learning model, and / or Prompt includes expert rules for the correspondence between burst events and burst event pulse sizes.

[0131] In a possible implementation, the processing module 403 is also used to: after obtaining the emergency event pulse size of the target emergency event, input the emergency event pulse size into the financial market time series model to obtain a prediction result of the impact of the target emergency event on the future trend of financial market data, the financial market time series model is constructed based on the pulse response function; based on the prediction results and actual financial market data, optimize the pulse size determination model.

[0132] In a possible implementation, financial data includes financial market data and emergency event data, and the extraction module 402 is specifically used to: extract calendar effects and seasonal characteristics from financial market data to obtain time characteristics; extract volatility, trading volume and market depth from financial market data to obtain market indicator characteristics; extract GDP growth rate, inflation rate and interest rate changes from financial market data to obtain macroeconomic indicator characteristics; extract market sentiment characteristics from emergency event data to obtain sentiment characteristics; extract event characteristics from emergency event data to obtain event characteristics.

[0133] In one possible implementation, the pulse size determination model is obtained in the following manner: obtaining training samples, the training samples include financial data and sudden event pulse sizes corresponding to sample events; preprocessing the financial data, the preprocessing including data cleaning, denoising and standardization; extracting features from the preprocessed data to obtain sample features; inputting the sample features into the pulse size determination model to obtain the predicted sudden event pulse size corresponding to the financial data; based on the sudden event pulse size and the predicted sudden event pulse size, adjusting the model parameters of the pulse size determination model through cross-validation and parameter tuning.

[0134] The device for determining the size of an emergency pulse provided in the embodiment of the present application can execute the technical solution shown in the above method embodiment, and its implementation principle and beneficial effects are similar, which will not be repeated here.

[0135] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application, such as Figure 5 As shown, the electronic device 50 includes: at least one processor 51 and a memory 52. ​​The memory 52 is used to store instructions, and the processor 51 is used to call the instructions in the memory to execute the method steps provided in the above embodiment. The specific implementation method and technical effect are similar and will not be repeated here.

[0136] Optionally, the memory 52 may be independent or integrated with the processor 51 .

[0137] The memory 52 may include a random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0138] The processor 51 may be a general-purpose processor, including a central processing unit, a network processor (NP), etc.; a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components.

[0139] Optionally, the electronic device 50 may further include a communication interface 53. In a specific implementation, if the communication interface 53, the memory 52 and the processor 51 are implemented independently, the communication interface 53, the memory 52 and the processor 51 may be interconnected through a bus and communicate with each other. The system bus may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The system bus may be divided into an address bus, a data bus, a control bus, etc., but it does not mean that there is only one bus or one type of bus.

[0140] Optionally, in a specific implementation, if the communication interface 53, the memory 52 and the processor 51 are integrated on a chip, the communication interface 53, the memory 52 and the processor 51 can communicate through an internal interface.

[0141] The implementation principle and technical effects of the electronic device provided in this embodiment can be found in the aforementioned embodiments and will not be described in detail here.

[0142] The present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed, the method for determining the size of the sudden event pulse of any of the aforementioned embodiments is implemented.

[0143] The present application also provides a computer program product, including a computer program, which, when executed, implements the method for determining the size of a burst pulse of any of the aforementioned embodiments.

[0144] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present application.

[0145] It should be further noted that, although the various steps in the flowchart are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowchart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0146] It should be understood that the above-mentioned device embodiments are only illustrative, and the device of the present application can also be implemented in other ways. For example, the division of units / modules in the above-mentioned embodiments is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units, modules or components can be combined, or can be integrated into another system, or some features can be ignored or not executed.

[0147] In addition, unless otherwise specified, each functional unit / module in each embodiment of the present application may be integrated into one unit / module, each unit / module may exist physically separately, or two or more units / modules may be integrated together. The above-mentioned integrated unit / module may be implemented in the form of hardware or in the form of a software program module.

[0148] If the integrated unit / module is implemented in the form of hardware, the hardware may be a digital circuit, an analog circuit, etc. The physical implementation of the hardware structure includes but is not limited to transistors, memristors, etc. Unless otherwise specified, the processor may be any appropriate hardware processor, such as a CPU, a GPU, an FPGA, a DSP, an ASIC, etc. Unless otherwise specified, the storage unit may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory (RRAM), a dynamic random access memory (DRAM), a static random access memory (SRAM), an enhanced dynamic random access memory (EDRAM), a high-bandwidth memory (HBM), a hybrid memory cube (HMC), etc.

[0149] If the integrated unit / module is implemented in the form of a software program module and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk and other media that can store program codes.

[0150] In the above embodiments, the description of each embodiment has its own emphasis. For the part not described in detail in a certain embodiment, please refer to the relevant description of other embodiments. The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0151] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0152] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A method for determining the size of an emergency pulse, characterized in that: include: Obtain financial data corresponding to the target emergency event; Extracting features of the financial data to obtain feature data; The characteristic data is input into a pre-trained pulse size determination model, and the pulse size determination model is based on an integrated prompt word engineering and determines the sudden event pulse size of the target sudden event according to the characteristic data.

2. The method according to claim 1, characterized in that The pulse size determination model is based on integrated prompt word engineering and determines the sudden event pulse size of the target sudden event according to the characteristic data, including: The pulse size determination model performs feature mapping on the feature data to obtain historical emergencies associated with the feature data; determines a first pulse size based on the pre-learned correspondence between historical emergencies and emergencies pulse sizes, and guidance information in the prompt word project; and, based on a preset second pulse size, calibrates and optimizes the first pulse size to obtain the emergencies pulse size of the target emergencies.

3. The method according to claim 2, characterized in that The cue word engineering is determined by a deep learning model, and / or the cue word engineering includes expert rules for the correspondence between sudden events and sudden event pulse sizes.

4. The method according to claim 2, characterized in that: The method of calibrating and optimizing the first pulse size based on the preset second pulse size to obtain the sudden event pulse size of the target sudden event includes: adjusting the first pulse size according to a regression algorithm based on the preset second pulse size to obtain the sudden event pulse size of the target sudden event.

5. The method according to any one of claims 1 to 4, characterized in that After obtaining the burst pulse size of the target burst event, the method further includes: Inputting the pulse size of the emergency event into a financial market time series model to obtain a prediction result of the impact of the target emergency event on the future trend of financial market data, wherein the financial market time series model is constructed based on an impulse response function; Based on the prediction results and actual financial market data, the pulse size determination model is optimized.

6. The method according to any one of claims 1 to 4, characterized in that The financial data includes financial market data and emergency event data, and the feature extraction of the financial data to obtain feature data includes: Extracting calendar effects and seasonal characteristics from the financial market data to obtain time characteristics; Extracting volatility, trading volume and market depth from the financial market data to obtain market indicator characteristics; Extracting the GDP growth rate, inflation rate and interest rate changes from the financial market data to obtain macroeconomic indicator characteristics; Extracting market sentiment features from the emergency event data to obtain sentiment features; Event features are extracted from the emergency event data to obtain event features.

7. The method according to any one of claims 1 to 4, characterized in that The pulse size determination model is obtained by: Acquire a training sample, wherein the training sample includes financial data corresponding to a sample event and a pulse size of an emergency event; Preprocessing the financial data, wherein the preprocessing includes data cleaning, denoising and standardization; Perform feature extraction on the preprocessed data to obtain sample features; Inputting the sample features into a pulse size determination model to obtain a predicted emergency pulse size corresponding to the financial data; Based on the sudden event pulse size and the predicted sudden event pulse size, the model parameters of the pulse size determination model are adjusted through a cross-validation method and a parameter tuning method.

8. A device for determining the size of an emergency pulse, comprising: An acquisition module, used to acquire financial data corresponding to a target emergency event; An extraction module, used to extract features of the financial data to obtain feature data; A processing module is used to input the feature data into a pre-trained pulse size determination model, wherein the pulse size determination model is based on an integrated prompt word engineering and obtains the sudden event pulse size of the target sudden event according to the feature data.

9. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed.

Images