CN118628113A - A financial transaction risk assessment method, device and medium based on machine learning - Google Patents

Abstract

The embodiments of this specification disclose a financial transaction risk assessment method, device and medium based on machine learning, which relate to the field of financial technology. The method includes: determining the model prediction risk index corresponding to the financial transaction to be predicted through multi-dimensional historical transaction information corresponding to the financial transaction to be predicted and a pre-built financial risk assessment model, using web crawler technology to collect real-time text data of market participants, and based on the real-time text data of market participants, conducting market sentiment assessment on the previous financial market to determine the current market sentiment index corresponding to the current financial market; determining the predicted financial event corresponding to the current financial market according to the current market sentiment index, and correcting the model prediction risk index through the predicted financial event to determine the current predicted risk index of the financial transaction to be predicted; and determining the risk level of the financial transaction to be predicted based on the current predicted risk index of the financial transaction to be predicted.

Description

A financial transaction risk assessment method, device and medium based on machine learning

Technical Field

This specification relates to the field of financial technology, and in particular to a financial transaction risk assessment method, device and medium based on machine learning.

Background Art

As the global financial market continues to expand and become more complex, investors and financial institutions have an increasing need to identify and manage transaction risks. The accuracy and real-time nature of financial transaction risk assessment are of great significance for ensuring market stability, optimizing investment strategies, and ensuring asset security. Traditional financial transaction risk assessment methods mainly rely on manual analysis, expert experience, and simple mathematical models. They have great limitations when processing large-scale, high-dimensional financial data and are difficult to meet the risk assessment needs of modern financial markets. In addition, traditional cross-border transaction risk assessment relies on historical data that has already occurred and lacks real-time dimensional data. Therefore, traditional financial transaction risk assessment methods have great limitations when processing large-scale, high-dimensional financial data, and lack real-time dimensional data, resulting in the inability of the accuracy and reliability of risk assessment to meet assessment needs.

Summary of the invention

One or more embodiments of the present specification provide a financial transaction risk assessment method, device and medium based on machine learning, which are used to solve the following technical problems: Traditional financial transaction risk assessment methods have great limitations when processing large-scale, high-dimensional financial data, and lack real-time dimensional data, resulting in the accuracy and reliability of risk assessment being unable to meet assessment requirements.

One or more embodiments of this specification adopt the following technical solutions:

One or more embodiments of the present specification provide a financial transaction risk assessment method based on machine learning, the method comprising: determining a model prediction risk index corresponding to the financial transaction to be predicted through multi-dimensional historical transaction information corresponding to the financial transaction to be predicted and a pre-built financial risk assessment model, wherein the multi-dimensional historical transaction information includes historical transaction data, user behavior data and market trend data; using web crawler technology to collect real-time text data of market participants in the current financial market corresponding to the financial transaction to be predicted, so as to conduct a market sentiment assessment on the current financial market based on the real-time text data of the market participants, and determine a current market sentiment index corresponding to the current financial market; determining a predicted financial event corresponding to the current financial market based on the current market sentiment index, so as to correct the model prediction risk index through the predicted financial event and determine the current predicted risk index of the financial transaction to be predicted; and conducting a risk assessment on the financial transaction to be predicted based on the current predicted risk index of the financial transaction to be predicted, and determining the risk level of the financial transaction to be predicted.

Furthermore, the model prediction risk index corresponding to the financial transaction to be predicted is determined through the multi-dimensional historical transaction information corresponding to the financial transaction to be predicted and the pre-constructed financial risk assessment model, which specifically includes: collecting multi-dimensional historical transaction information related to the financial transaction to be predicted to perform feature extraction on the multi-dimensional historical transaction information to determine the corresponding transaction risk feature set, wherein the transaction risk feature set includes transaction price and volatility characteristics, transaction volume characteristics, user behavior characteristics and market trend indicators; using a preset machine learning algorithm to construct a financial risk assessment model, inputting the transaction risk feature set into the financial risk assessment model, and determining the model prediction risk index corresponding to the financial transaction to be predicted.

Furthermore, based on the real-time text data of the market participating users, a market sentiment assessment is performed on the current financial market to determine the current market sentiment index corresponding to the current financial market, specifically including: preprocessing the real-time text data of the market participating users to extract a real-time text emotion feature set in the real-time text data of the market participating users; determining the current market sentiment index corresponding to the current financial market based on the real-time text emotion feature set and a pre-constructed market sentiment prediction model.

Furthermore, the real-time text data of the market participating users are preprocessed to extract a real-time text emotion feature set from the real-time text data of the market participating users, specifically including: preprocessing the real-time text data of the market participating users to obtain standard text data; performing word segmentation processing on the standard text data to divide the standard text data into multiple text words to generate a text vocabulary set; using natural language processing technology to perform word form restoration on multiple question words in the text vocabulary set to determine the real-time text emotion feature set.

Further, according to the current market sentiment index, the predicted financial event corresponding to the current financial market is determined, specifically including: obtaining multiple historical market sentiment information, wherein the historical market sentiment information includes historical market sentiment indicators and historical timestamps; classifying the multiple historical market sentiment information according to the historical market sentiment indicators to generate a sentiment indicator data set corresponding to each sentiment indicator; based on the historical timestamps in each of the sentiment indicator data sets, obtaining at least one historical financial event within a specified time period corresponding to each of the historical timestamps, wherein the specified time period is a time interval after the historical timestamp; performing statistical analysis on at least one historical financial event within a specified time period corresponding to each of the historical timestamps in the sentiment indicator data set to determine the largest number of specified historical financial events in each of the sentiment indicator data sets; establishing a correspondence between each sentiment indicator and a financial event based on the largest number of specified historical financial events in each of the sentiment indicator data sets; matching the current market sentiment indicators based on the correspondence between each of the sentiment indicators and financial events to determine the predicted financial event.

Furthermore, through the predicted financial event, the model prediction risk indicator is corrected to determine the current prediction risk indicator of the financial transaction to be predicted, which specifically includes: obtaining the historical transaction impact parameters corresponding to the predicted financial event, wherein the historical transaction impact parameters include market volatility parameters and transaction completion parameters; through the historical transaction impact parameters, the transaction impact risk of the predicted financial event is evaluated to determine the corresponding impact risk indicator, wherein the impact risk indicator includes any one of a positive impact indicator and a negative impact indicator; based on the impact risk indicator and the model prediction risk indicator, the current prediction risk indicator of the financial transaction to be predicted is generated.

Furthermore, after performing a risk assessment on the financial transaction to be predicted based on the current predicted risk indicator of the financial transaction to be predicted and determining the risk level of the financial transaction to be predicted, the method further includes: setting a risk level threshold to generate early warning information based on the risk level of the financial transaction to be predicted and the risk level threshold; and determining a corresponding risk management strategy based on the risk level to generate a risk assessment report based on the early warning information and the risk management strategy.

Furthermore, feature extraction is performed on the multi-dimensional historical transaction information to determine a corresponding set of transaction risk features, specifically including: obtaining historical transaction data, user behavior data and market trend data in the multi-dimensional historical transaction information; analyzing historical transaction prices in the historical transaction data to determine the transaction price change rate and transaction price volatility; analyzing historical transaction volumes in the historical transaction data to generate the transaction volume change rate and the influence relationship between the transaction volume and the transaction price to determine the transaction volume characteristics; determining the user's transaction frequency, position holding time and transaction preference data based on the user behavior data to determine the user behavior characteristics; determining the market index through the market trend data to determine the market trend indicator.

One or more embodiments of this specification provide a financial transaction risk assessment device based on machine learning, including:

at least one processor; and,

a memory communicatively connected to the at least one processor; wherein,

The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

The model prediction risk index corresponding to the financial transaction to be predicted is determined by using the multi-dimensional historical transaction information corresponding to the financial transaction to be predicted and a pre-built financial risk assessment model, wherein the multi-dimensional historical transaction information includes historical transaction data, user behavior data and market trend data; the real-time text data of market participants in the current financial market corresponding to the financial transaction to be predicted is collected by using web crawler technology, so as to conduct a market sentiment assessment on the current financial market based on the real-time text data of the market participants, and determine the current market sentiment index corresponding to the current financial market; the predicted financial event corresponding to the current financial market is determined based on the current market sentiment index, so as to correct the model prediction risk index through the predicted financial event and determine the current predicted risk index of the financial transaction to be predicted; based on the current predicted risk index of the financial transaction to be predicted, the risk of the financial transaction to be predicted is assessed to determine the risk level of the financial transaction to be predicted.

One or more embodiments of this specification provide a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured to:

The model prediction risk index corresponding to the financial transaction to be predicted is determined by using the multi-dimensional historical transaction information corresponding to the financial transaction to be predicted and a pre-built financial risk assessment model, wherein the multi-dimensional historical transaction information includes historical transaction data, user behavior data and market trend data; the real-time text data of market participants in the current financial market corresponding to the financial transaction to be predicted is collected by using web crawler technology, so as to conduct a market sentiment assessment on the current financial market based on the real-time text data of the market participants, and determine the current market sentiment index corresponding to the current financial market; the predicted financial event corresponding to the current financial market is determined based on the current market sentiment index, so as to correct the model prediction risk index through the predicted financial event and determine the current predicted risk index of the financial transaction to be predicted; based on the current predicted risk index of the financial transaction to be predicted, the risk of the financial transaction to be predicted is assessed to determine the risk level of the financial transaction to be predicted.

At least one of the above technical solutions adopted in the embodiments of this specification can achieve the following beneficial effects: through the above technical solutions, multi-dimensional historical transaction information (including historical transaction data, user behavior data and market trend data) provides rich background information and data support for risk assessment, making the assessment results more comprehensive. The pre-built financial risk assessment model is trained based on a large amount of historical data, which can identify and quantify different types of risks, thereby improving the accuracy of risk assessment; the use of web crawler technology to collect text data of market participants in real time can quickly capture changes in market sentiment and provide the latest market feedback for risk assessment; real-time market sentiment assessment makes risk assessment dynamic and can adjust investment strategies and risk management measures in a timely manner; the predicted financial events determined based on the current market sentiment indicators provide an important basis for revising the model's predicted risk indicators, which helps to more accurately reflect the actual market situation and improve the pertinence and effectiveness of risk assessment; the determination of risk levels enables investors to more intuitively understand the risk level of the financial transactions to be predicted, providing an intuitive reference for investment decisions; through timely monitoring and assessment of market risks, financial institutions can promptly discover potential risks and take corresponding risk management measures to reduce potential losses.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following briefly introduces the drawings required for the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings:

FIG1 is a flow chart of a financial transaction risk assessment method based on machine learning provided in an embodiment of this specification;

FIG2 is a schematic diagram of the structure of a financial transaction risk assessment device based on machine learning provided in an embodiment of this specification.

DETAILED DESCRIPTION

In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

As the global financial market continues to expand and become more complex, investors and financial institutions have an increasing need to identify and manage transaction risks. The accuracy and real-time nature of financial transaction risk assessment are of great significance for ensuring market stability, optimizing investment strategies, and ensuring asset security. Traditional financial transaction risk assessment methods mainly rely on manual analysis, expert experience, and simple mathematical models. They have great limitations when processing large-scale, high-dimensional financial data and are difficult to meet the risk assessment needs of modern financial markets. In addition, traditional cross-border transaction risk assessment relies on historical data that has already occurred and lacks real-time dimensional data. Therefore, traditional financial transaction risk assessment methods have great limitations when processing large-scale, high-dimensional financial data, and lack real-time dimensional data, resulting in the inability of the accuracy and reliability of risk assessment to meet assessment needs.

The embodiment of this specification provides a financial transaction risk assessment method based on machine learning. It should be noted that the execution subject in the embodiment of this specification can be a server or any device with data processing capabilities. Figure 1 is a flow chart of a financial transaction risk assessment method based on machine learning provided by the embodiment of this specification. As shown in Figure 1, it mainly includes the following steps:

Step S101, determining the model prediction risk index corresponding to the financial transaction to be predicted through the multi-dimensional historical transaction information corresponding to the financial transaction to be predicted and the pre-built financial risk assessment model.

Among them, the multi-dimensional historical transaction information includes historical transaction data, user behavior data and market trend data.

The model prediction risk index corresponding to the financial transaction to be predicted is determined by using the multi-dimensional historical transaction information corresponding to the financial transaction to be predicted and a pre-constructed financial risk assessment model, specifically including: collecting the multi-dimensional historical transaction information related to the financial transaction to be predicted, performing feature extraction on the multi-dimensional historical transaction information, and determining the corresponding transaction risk feature set, wherein the transaction risk feature set includes transaction price and volatility features, transaction volume features, user behavior features, and market trend indicators; using a preset machine learning algorithm to construct a financial risk assessment model, inputting the transaction risk feature set into the financial risk assessment model, and determining the model prediction risk index corresponding to the financial transaction to be predicted.

In one embodiment of the present specification, multi-dimensional historical transaction information corresponding to the financial transaction to be predicted is collected, wherein the multi-dimensional historical transaction information includes historical transaction data, user behavior data and market trend data. Collect historical transaction records related to the financial transaction to be predicted, including transaction time, transaction price, transaction volume, etc.; analyze the transaction behavior of users in historical transactions, such as transaction frequency, position holding time, transaction preference, etc.; collect market trend data, such as macroeconomic data, industry data, policy changes, etc., to understand the market background. Clean, integrate and standardize the collected multi-dimensional historical transaction information to ensure the quality and consistency of the data, and perform operations such as missing value filling, outlier processing, and data normalization. Extract the characteristics of transaction price and volatility, transaction volume, user behavior and market trend indicators from the multi-dimensional historical transaction information to determine the corresponding transaction risk feature set. Load a pre-built financial risk assessment model, which is built based on machine learning, deep learning or traditional statistical methods and is used to predict the risks of financial transactions. The extracted features are used as input and passed to the financial risk assessment model. The model calculates based on the input features and outputs the model prediction risk index corresponding to the financial transaction to be predicted. The index is usually a number or score that represents the risk level of the transaction.

Feature extraction is performed on the multi-dimensional historical transaction information to determine the corresponding transaction risk feature set, specifically including: obtaining historical transaction data, user behavior data and market trend data in the multi-dimensional historical transaction information; analyzing the historical transaction prices in the historical transaction data to determine the transaction price change rate and transaction price volatility; analyzing the historical transaction volume in the historical transaction data to generate the transaction volume change rate and the influence relationship between the transaction volume and the transaction price to determine the transaction volume characteristics; determining the user's transaction frequency, position holding time and transaction preference data based on the user behavior data to determine the user behavior characteristics; determining the market index through the market trend data to determine the market trend indicator.

In one embodiment of the present specification, extracting features related to financial transaction risks from multi-dimensional historical transaction information is a crucial step in financial risk assessment. Features can reveal the dynamics of financial markets, trader behavior patterns, and potential risk factors. Obtain historical transaction data, user behavior data, and market trend data in multi-dimensional historical transaction information. Analyze the historical transaction prices in the historical transaction data, calculate the historical change rate of the transaction price, such as daily increase or decrease, weekly increase or decrease, etc., use standard deviation, variance or volatility index (such as VIX) to measure the degree of price fluctuation, and determine the transaction price change rate and transaction price volatility. Analyze the historical transaction volume in the historical transaction data, use daily or hourly transaction volume data, calculate the change rate of transaction volume, so as to understand the changes in market activity, analyze the performance of transaction volume when prices rise or fall, such as "volume and price rise together" or "volume shrinkage and price drop", generate the transaction volume change rate and the influence relationship between transaction volume and transaction price, and determine the transaction volume characteristics. Based on the user behavior data, the number of transactions or transaction frequency of the user is calculated, the length of time the user holds a certain asset is analyzed, and whether the user tends to buy or sell a certain asset, or whether the user prefers a certain trading strategy, is identified to determine the user's behavior characteristics. Through the market trend data, the market index is determined, and market indices such as the Dow Jones and S&P 500 are used to measure the performance of the overall market to determine the market trend indicators.

Through the above technical solutions, multi-dimensional historical transaction information contains a wealth of key data such as market volatility parameters and transaction parameters, which can provide a comprehensive and detailed transaction background for the financial risk assessment model. The pre-built financial risk assessment model is based on a large amount of historical data and statistical analysis, and can identify and quantify different types of risks, such as credit risk, market risk, and operational risk. By combining multi-dimensional historical transaction information and risk assessment models, the risk level of the financial transaction to be predicted can be more accurately predicted, improving the accuracy and reliability of risk assessment.

Step S102, using web crawler technology to collect real-time text data of market participants in the current financial market corresponding to the financial transaction to be predicted, so as to evaluate the market sentiment of the current financial market based on the real-time text data of market participants, and determine the current market sentiment index corresponding to the current financial market.

In one embodiment of the present specification, web crawler technology is used to collect real-time text data of market participants in the current financial market corresponding to the financial transaction to be predicted. The market participants here include various trading identities such as investment users, financial analysts, and trading users. The real-time text data may include text data on social platforms such as social media and financial forums, and may also be text data such as news reports related to the financial market. The data usually includes their views, predictions and emotional expressions on the market.

Based on the real-time text data of the market participating users, a market sentiment assessment is conducted on the current financial market to determine the current market sentiment index corresponding to the current financial market, specifically including: preprocessing the real-time text data of the market participating users to extract a real-time text sentiment feature set in the real-time text data of the market participating users; and determining the current market sentiment index corresponding to the current financial market based on the real-time text sentiment feature set and a pre-built market sentiment prediction model.

The real-time text data of the market participating users is preprocessed to extract a real-time text emotion feature set in the real-time text data of the market participating users, specifically including: preprocessing the real-time text data of the market participating users to obtain standard text data; performing word segmentation processing on the standard text data to divide the standard text data into multiple text words to generate a text vocabulary set; using natural language processing technology to perform word form restoration on multiple question words in the text vocabulary set to determine the real-time text emotion feature set.

In one embodiment of the present specification, the collected text data is cleaned to remove duplicate, irrelevant and noise data. Text segmentation is performed to divide the text data into independent words or phrases. Stop words, such as common prepositions, conjunctions, etc., are removed to reduce data sparsity. Stem extraction or morphological restoration may also be required to unify different forms of vocabulary into the same stem or root. The emotional tendency in the text data is extracted, which can be achieved by a sentiment analysis algorithm (such as a rule-based method or a machine learning-based method). The sentiment analysis algorithm can identify positive, negative or neutral emotions in the text and extract topics or keywords in the text, which helps to understand the main content of the user discussion. Based on the extracted features, a market sentiment model is constructed. The model can be based on statistics (such as calculating the ratio of positive and negative emotions) or based on machine learning (such as using a classifier to predict market sentiment categories). If there is historical market sentiment data and corresponding financial market data, these data can be used to train the model to improve the model's predictive ability. For example, multiple historical market sentiment text data are obtained, and the historical market sentiment text data are annotated to construct an annotated data set, wherein the annotated data set includes the historical market sentiment text data and corresponding sentiment category labels; the annotated data set is divided into a training set, a validation set, and a test set, and the market sentiment prediction model is trained using the training set data, the model parameters are adjusted, and the market sentiment prediction model is determined.

Input the text data collected in real time into the market sentiment model to obtain the prediction results of the current market sentiment, which can be a numerical value (such as sentiment score or sentiment index) or a category (such as optimistic, pessimistic, neutral). According to the prediction results output by the model, determine the current market sentiment index corresponding to the current financial market. This indicator can be directly based on the sentiment score or sentiment index output by the model, or it can be further calculated or converted based on the model prediction results. For example, when the investor sentiment index is high, it indicates that investors are generally optimistic, which may indicate an upward trend in the market, especially when the index continues to rise and exceeds the historical average; a low index may mean that the market is pessimistic and there is an opportunity for underestimation.

Through the above technical solutions, web crawler technology can automatically and quickly capture real-time text data on the Internet, such as news reports, social media comments, forum discussions, etc. These data often directly reflect the immediate emotions and opinions of market participants. By analyzing these real-time data, financial institutions and investors can quickly obtain information on changes in market sentiment, thereby adjusting investment strategies in a timely manner and reducing investment risks. Web crawler technology can collect a large amount of comprehensive real-time text data from market participants. The richness of this data provides more information sources and reference bases for market sentiment assessment; combined with advanced technologies such as natural language processing (NLP), these text data can be deeply analyzed and mined to extract the key features of market sentiment, thereby more accurately assessing market sentiment.

Step S103, determining the predicted financial event corresponding to the current financial market according to the current market sentiment index, so as to correct the model prediction risk index by predicting the financial event and determine the current prediction risk index of the financial transaction to be predicted.

According to the current market sentiment index, the predicted financial event corresponding to the current financial market is determined, specifically including: obtaining multiple historical market sentiment information, wherein the historical market sentiment information includes historical market sentiment indicators and historical timestamps; classifying the multiple historical market sentiment information according to the historical market sentiment indicators to generate a sentiment indicator data set corresponding to each sentiment indicator; based on the historical timestamps in each of the sentiment indicator data sets, obtaining at least one historical financial event within a specified time period corresponding to each of the historical timestamps, wherein the specified time period is a time interval after the historical timestamp; performing statistical analysis on at least one historical financial event within a specified time period corresponding to each of the historical timestamps in the sentiment indicator data set to determine the largest number of specified historical financial events in each of the sentiment indicator data sets; establishing a correspondence between each sentiment indicator and a financial event based on the largest number of specified historical financial events in each of the sentiment indicator data sets; matching the current market sentiment indicators based on the correspondence between each of the sentiment indicators and financial events to determine the predicted financial event.

In one embodiment of the present specification, a plurality of historical market sentiment information is obtained, and the historical market sentiment information includes historical market sentiment indicators and historical timestamps. According to the historical market sentiment indicators, the plurality of historical market sentiment information is classified, and a sentiment indicator data set corresponding to each sentiment indicator is generated to obtain a plurality of sentiment indicator sets. According to the historical timestamps in each of the sentiment indicator data sets, at least one historical financial event in a specified time period corresponding to each of the historical timestamps is obtained, wherein the specified time period is a time interval after the historical timestamp, that is, a historical financial event that occurs after the historical timestamp. A statistical analysis is performed on at least one historical financial event in a specified time period corresponding to each of the historical timestamps in the sentiment indicator data set, and the designated historical financial event with the most occurrences in each of the sentiment indicator data sets is determined. According to the designated historical financial event with the most occurrences in each of the sentiment indicator data sets, a corresponding relationship between each sentiment indicator and a financial event is established; according to the corresponding relationship between each sentiment indicator and a financial event, the current market sentiment indicator is matched to determine a predicted financial event. The predicted financial events herein include a variety of financial events such as market rise, market fall, and market turmoil. For example, when the investor sentiment index is high and continues to rise, market rising events are generally prone to occur. When the investor sentiment index is low and continues to decline, market falling events generally occur. Market turbulence generally refers to fluctuations in the investor sentiment index.

Through the predicted financial event, the model prediction risk index is corrected to determine the current prediction risk index of the financial transaction to be predicted, specifically including: obtaining the historical transaction impact parameters corresponding to the predicted financial event, wherein the historical transaction impact parameters include market volatility parameters and transaction completion parameters; through the historical transaction impact parameters, the transaction impact risk of the predicted financial event is evaluated to determine the corresponding impact risk index, wherein the impact risk index includes any one of a positive impact index and a negative impact index; based on the impact risk index and the model prediction risk index, the current prediction risk index of the financial transaction to be predicted is generated.

In one embodiment of the present specification, historical transaction impact parameters corresponding to the predicted financial event are obtained, and the historical transaction impact parameters include market volatility parameters and transaction completion parameters. The historical transaction impact parameters here generally refer to the transaction parameters affected when the financial event occurs in the historical state; the market data of the financial events that have occurred in history and their corresponding time points are collected to ensure that the data contains market volatility parameters (such as price volatility, transaction volume change rate, etc.) and transaction completion parameters (such as transaction price, transaction volume, buy-sell ratio, etc.). After cleaning the data, removing missing values, outliers or erroneous data, the data is standardized or normalized for subsequent analysis. The market volatility parameters and transaction completion parameters related to the financial event are extracted from the historical data as features, and time series analysis, event study method and other methods can be considered to further refine the features. Use statistical models (such as linear regression, logistic regression, random forest, etc.) or machine learning models (such as neural networks, deep learning, etc.) to establish a relationship model between financial events and market volatility parameters and transaction completion parameters. By training the model, the possible market fluctuations and transaction completion situations under a given financial event can be predicted. For the financial events to be predicted, the established models are used to predict their possible market volatility parameters and transaction parameters. Based on the prediction results, the positive impact (such as increasing transaction volume, stabilizing prices, etc.) or negative impact (such as causing price plunges, shrinking transaction volumes, etc.) of the financial events on transactions are evaluated.

Based on the prediction results and risk assessment, the impact risk indicators are determined, including positive impact indicators (such as expected transaction volume growth rate, expected price stability, etc.) or negative impact indicators (such as expected price decline, expected transaction volume decline, etc.). Combined with the model prediction risk indicators of the financial event (such as the risk prediction value output by the model trained based on historical data) and the above-determined impact risk indicators, the current prediction risk indicator of the financial transaction to be predicted is generated. The prediction risk indicator can be a comprehensive value or score that reflects the comprehensive impact risk of the financial event on the transaction.

Through the above technical solution, by comprehensively considering the market volatility parameters and transaction completion parameters in historical transactions, the actual impact of financial events on transactions can be more comprehensively evaluated; the determination of risk indicators (including positive and negative impact indicators) makes risk management more targeted, and more refined risk management strategies can be formulated based on these indicators to effectively deal with the potential risks brought by financial events. By predicting the transaction impact risk of financial events, investors can better understand the performance of different assets under specific events, thereby optimizing the allocation of investment portfolios; real-time analysis of financial events and assessment of their potential impact on transactions enable investors and financial institutions to promptly identify risks and take measures to reduce losses.

Step S104: Based on the current prediction risk index of the financial transaction to be predicted, a risk assessment is performed on the financial transaction to be predicted to determine the risk level of the financial transaction to be predicted.

After performing a risk assessment on the financial transaction to be predicted based on the current predicted risk indicator of the financial transaction to be predicted and determining the risk level of the financial transaction to be predicted, the method further includes: setting a risk level threshold to generate early warning information based on the risk level of the financial transaction to be predicted and the risk level threshold; and determining a corresponding risk management strategy based on the risk level to generate a risk assessment report based on the early warning information and the risk management strategy.

In one embodiment of the present specification, a series of risk levels are set based on historical data and experience, such as low risk, medium risk, high risk, etc. Each risk level should correspond to a range or threshold of a risk indicator. The current predicted risk indicator of the financial transaction to be predicted is compared with the set risk level threshold. Based on the comparison result, the risk level of the financial transaction to be predicted is determined. A risk assessment report is prepared, detailing the risk level of the financial transaction to be predicted and the basis for the assessment. The report may include a detailed analysis of risk factors, a prediction of potential risks, and corresponding risk management recommendations.

When the assessment results show that the risk level exceeds the preset threshold, the system will automatically issue an early warning signal. The early warning signal will be notified to the relevant risk managers directly via email, SMS or system interface. It also includes a feedback mechanism that can adjust and optimize the risk assessment model based on actual financial events. This self-learning and adaptation mechanism will continuously improve the accuracy of the model. Provide a user-friendly interface that allows risk managers to view risk assessment results, adjust risk thresholds, view early warning history, etc.

Through risk assessment based on current forecast risk indicators, the risk level of the financial transaction to be forecasted can be determined more accurately. This helps investors and financial institutions to more accurately grasp the risk level of transactions and formulate more refined risk management strategies; the risk assessment results provide investors with important information about the risk level of the financial transaction to be forecasted, and investors can weigh risks and benefits according to the risk level and make more informed trading decisions; by setting clear risk levels and assessment standards, risk management can be standardized and normalized, which helps to improve the efficiency and consistency of risk management and reduce the impact of human factors on risk management.

Through the above technical solutions, multi-dimensional historical transaction information (including historical transaction data, user behavior data and market trend data) provides rich background information and data support for risk assessment, making the assessment results more comprehensive. The pre-built financial risk assessment model is trained based on a large amount of historical data, which can identify and quantify different types of risks and improve the accuracy of risk assessment. The use of web crawler technology to collect text data of market participants in real time can quickly capture changes in market sentiment and provide the latest market feedback for risk assessment. Real-time market sentiment assessment makes risk assessment dynamic and can adjust investment strategies and risk management measures in a timely manner. The predicted financial events determined according to the current market sentiment indicators provide an important basis for revising the model's predicted risk indicators, which helps to more accurately reflect the actual market situation and improve the pertinence and effectiveness of risk assessment. The determination of risk levels enables investors to more intuitively understand the risk level of the financial transactions to be predicted, providing an intuitive reference for investment decisions. By timely monitoring and evaluating market risks, financial institutions can promptly discover potential risks and take corresponding risk management measures to reduce potential losses.

The embodiment of the present specification also provides a financial transaction risk assessment device based on machine learning, as shown in Figure 2, the device includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can: determine the model prediction risk index corresponding to the financial transaction to be predicted through the multi-dimensional historical transaction information corresponding to the financial transaction to be predicted and the pre-built financial risk assessment model, wherein the multi-dimensional historical transaction information includes historical transaction data, user behavior data and market trend data; use web crawler technology to collect real-time text data of market participants in the current financial market corresponding to the financial transaction to be predicted, so as to conduct market sentiment assessment on the current financial market based on the real-time text data of the market participants, and determine the current market sentiment index corresponding to the current financial market; determine the predicted financial event corresponding to the current financial market according to the current market sentiment index, so as to correct the model prediction risk index through the predicted financial event and determine the current predicted risk index of the financial transaction to be predicted; conduct risk assessment on the financial transaction to be predicted based on the current predicted risk index of the financial transaction to be predicted, and determine the risk level of the financial transaction to be predicted.

The embodiments of the present specification also provide a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured to: determine the model prediction risk index corresponding to the financial transaction to be predicted through multi-dimensional historical transaction information corresponding to the financial transaction to be predicted and a pre-built financial risk assessment model, wherein the multi-dimensional historical transaction information includes historical transaction data, user behavior data, and market trend data; collect real-time text data of market participants in the current financial market corresponding to the financial transaction to be predicted by using web crawler technology, so as to conduct a market sentiment assessment on the current financial market based on the real-time text data of the market participants, and determine the current market sentiment index corresponding to the current financial market; determine the predicted financial event corresponding to the current financial market based on the current market sentiment index, so as to correct the model prediction risk index through the predicted financial event, and determine the current predicted risk index of the financial transaction to be predicted; conduct a risk assessment on the financial transaction to be predicted based on the current predicted risk index of the financial transaction to be predicted, and determine the risk level of the financial transaction to be predicted.

Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

The devices and media provided in the embodiments of this specification correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

This specification is described with reference to the flowchart and/or block diagram of the method, device (system), and computer program product according to the embodiment of this specification. It should be understood that each process and/or box in the flowchart and/or block diagram, as well as the combination of the process and/or box in the flowchart and/or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device for implementing the functions specified in one process or multiple processes in the flowchart and/or one box or multiple boxes in the block diagram.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in one or more processes in the flowchart and/or one or more boxes in the block diagram.

These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart and/or one or more boxes in the block diagram.

In a typical configuration, a computing device includes one or more processors (CPU), input/output interfaces, network interfaces, and memory.

The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and/or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.

Claims (10)

1. A machine learning based financial transaction risk assessment method, the method comprising:

Determining a model prediction risk index corresponding to the financial transaction to be predicted through multidimensional historical transaction information corresponding to the financial transaction to be predicted and a pre-constructed financial risk assessment model, wherein the multidimensional historical transaction information comprises historical transaction data, user behavior data and market trend data;

collecting real-time text data of market participant users of the current financial market corresponding to the financial transaction to be predicted by utilizing a web crawler technology, so as to evaluate market emotion of the current financial market based on the real-time text data of the market participant users, and determining a current market emotion index corresponding to the current financial market;

determining a predicted financial event corresponding to the current financial market according to the current market emotion index, correcting the model predicted risk index through the predicted financial event, and determining the current predicted risk index of the financial transaction to be predicted;

And carrying out risk assessment on the financial transaction to be predicted based on the current predicted risk index of the financial transaction to be predicted, and determining the risk level of the financial transaction to be predicted.

2. The machine learning-based financial transaction risk assessment method according to claim 1, wherein determining a model prediction risk indicator corresponding to a financial transaction to be predicted through multidimensional historical transaction information corresponding to the financial transaction to be predicted and a pre-constructed financial risk assessment model specifically comprises:

Collecting multi-dimensional historical transaction information related to the financial transaction to be predicted, so as to perform feature extraction on the multi-dimensional historical transaction information and determine a corresponding transaction risk feature set, wherein the transaction risk feature set comprises transaction price and volatility features, transaction quantity features, user behavior features and market trend indexes;

And constructing a financial risk assessment model by using a preset machine learning algorithm, inputting the transaction risk feature set into the financial risk assessment model, and determining a model prediction risk index corresponding to the financial transaction to be predicted.

3. The machine learning based financial transaction risk assessment method according to claim 1, wherein the market emotion assessment is performed on the current financial market based on real-time text data of the market participant, and determining the current market emotion index corresponding to the current financial market specifically comprises:

Preprocessing the real-time text data of the market participant users to extract a real-time text emotion feature set in the real-time text data of the market participant users;

And determining the current market emotion index corresponding to the current financial market according to the real-time text emotion feature set and a pre-constructed market emotion prediction model.

4. A machine learning based financial transaction risk assessment method according to claim 3, wherein the preprocessing of the real-time text data of the market participant user to extract a set of real-time text emotional characteristics in the real-time text data of the market participant user specifically comprises:

Preprocessing the real-time text data of the market participant users to obtain standard text data;

Word segmentation is carried out on the standard text data so as to segment the standard text data into a plurality of text words and generate a text vocabulary set;

and performing morphological reduction on a plurality of question words in the text vocabulary set by using a natural language processing technology so as to determine the real-time text emotion feature set.

5. The machine learning based financial transaction risk assessment method according to claim 1, wherein determining the predicted financial event corresponding to the current financial market according to the current market emotion index specifically comprises:

Acquiring a plurality of historical market emotion information, wherein the historical market emotion information comprises historical market emotion indexes and historical timestamps;

Classifying the historical market emotion information according to the historical market emotion indexes to generate emotion index data sets corresponding to each emotion index;

Acquiring at least one historical financial event in a specified time period corresponding to each historical timestamp based on the historical timestamp in each emotion index data set, wherein the specified time period is a time interval after the historical timestamp;

carrying out statistical analysis on at least one historical financial event in a specified time period corresponding to each historical timestamp in the emotion index data set, and determining the most specified historical financial event in each emotion index data set;

establishing a corresponding relation between each emotion index and financial event according to the most appointed historical financial event in each emotion index data set;

and matching the current market emotion indexes according to the corresponding relation between each emotion index and the financial event, and determining the predicted financial event.

6. The machine learning based financial transaction risk assessment method according to claim 1, wherein the model predicted risk indicator is modified by the predicted financial event to determine a current predicted risk indicator for the financial transaction to be predicted, specifically comprising:

acquiring historical transaction influence parameters corresponding to the predicted financial event, wherein the historical transaction influence parameters comprise market fluctuation parameters and transaction achievement parameters;

Evaluating the transaction impact risk of the predicted financial event through the historical transaction impact parameters to determine corresponding impact risk indicators, wherein the impact risk indicators comprise any one of positive impact indicators and negative impact indicators;

And generating the current predicted risk index of the financial transaction to be predicted according to the influence risk index and the model predicted risk index.

7. The machine learning based financial transaction risk assessment method of claim 1, wherein, based on the current predicted risk indicator of the financial transaction to be predicted, risk assessment is performed on the financial transaction to be predicted, and after determining the risk level of the financial transaction to be predicted, the method further comprises:

setting a risk level threshold to generate early warning information based on the risk level of the financial transaction to be predicted and the risk level threshold;

and determining a corresponding risk management strategy based on the risk level so as to generate a risk assessment report based on the early warning information and the risk management strategy.

8. The machine learning based financial transaction risk assessment method according to claim 2, wherein the feature extraction is performed on the multi-dimensional historical transaction information, and the corresponding transaction risk feature set is determined, and specifically comprises:

Acquiring historical transaction data, user behavior data and market trend data in the multi-dimensional historical transaction information;

Analyzing the historical transaction price in the historical transaction data to determine the transaction price change rate and the transaction price fluctuation;

analyzing the historical transaction amount in the historical transaction data, generating a transaction amount change rate and an influence relation between the transaction amount and the transaction price, and determining transaction amount characteristics;

According to the user behavior data, determining the transaction frequency, the holding time and the transaction preference data of the user so as to determine the user behavior characteristics;

And determining the market index according to the market trend data so as to determine the market trend index.

9. A machine learning based financial transaction risk assessment device, the device comprising:

at least one processor; and

A memory communicatively coupled to the at least one processor; wherein,

The memory stores instructions executable by the at least one processor, the instructions are executable by the at least one processor to enable the at least one processor to:

Determining a model prediction risk index corresponding to the financial transaction to be predicted through multidimensional historical transaction information corresponding to the financial transaction to be predicted and a pre-constructed financial risk assessment model, wherein the multidimensional historical transaction information comprises historical transaction data, user behavior data and market trend data;

collecting real-time text data of market participant users of the current financial market corresponding to the financial transaction to be predicted by utilizing a web crawler technology, so as to evaluate market emotion of the current financial market based on the real-time text data of the market participant users, and determining a current market emotion index corresponding to the current financial market;

determining a predicted financial event corresponding to the current financial market according to the current market emotion index, correcting the model predicted risk index through the predicted financial event, and determining the current predicted risk index of the financial transaction to be predicted;

And carrying out risk assessment on the financial transaction to be predicted based on the current predicted risk index of the financial transaction to be predicted, and determining the risk level of the financial transaction to be predicted.

10. A non-transitory computer storage medium storing computer-executable instructions, the computer-executable instructions configured to:

Determining a model prediction risk index corresponding to the financial transaction to be predicted through multidimensional historical transaction information corresponding to the financial transaction to be predicted and a pre-constructed financial risk assessment model, wherein the multidimensional historical transaction information comprises historical transaction data, user behavior data and market trend data;

collecting real-time text data of market participant users of the current financial market corresponding to the financial transaction to be predicted by utilizing a web crawler technology, so as to evaluate market emotion of the current financial market based on the real-time text data of the market participant users, and determining a current market emotion index corresponding to the current financial market;

determining a predicted financial event corresponding to the current financial market according to the current market emotion index, correcting the model predicted risk index through the predicted financial event, and determining the current predicted risk index of the financial transaction to be predicted;

And carrying out risk assessment on the financial transaction to be predicted based on the current predicted risk index of the financial transaction to be predicted, and determining the risk level of the financial transaction to be predicted.