Stock market dynamic monitoring and early warning system and method based on artificial intelligence

Abstract

The invention relates to the field of stock market early warning analysis, and discloses a stock market dynamic monitoring and early warning system and method based on artificial intelligence, and the method comprises the steps: collecting heterogeneous data, high-frequency order data and media data of a stock market in real time, and carrying out the integration processing, and obtaining fusion data; the method comprises the following steps: collecting historical data of a stock market, performing market state division on the stock market to obtain a defined market state, querying current news information, and performing impact event prediction on the stock market to obtain an event impact chain; acquiring investment data and risk index analysis requirements of stock users corresponding to the stock market, calculating investment risk values of the stock users, acquiring market mobility indexes of the stock market, and analyzing mobility pressure of the stock market; a combined early warning system is constructed by defining market states, event impact chains, investment risk values and mobility pressure, and real-time early warning is carried out on stock users. According to the invention, the effectiveness of dynamic monitoring and early warning of the stock market can be improved.

Description

Technical Field

[0001] The present invention relates to the field of stock market warning analysis, and particularly to a dynamic monitoring and warning system for the stock market based on artificial intelligence and its method. Background Art

[0002] The stock market, as a key venue for corporate financing and investors to trade equities, is an important component of the economic system. The stock market not only reflects the operating conditions of enterprises but also serves as a leading indicator of the macroeconomy, guiding the rational allocation of economic resources. Therefore, it is of great significance to conduct dynamic monitoring and warning of the stock market.

[0003] Currently, the dynamic monitoring and warning of the stock market mainly rely on technical analysis methods. By plotting charts based on historical stock prices, trading volumes and other data, and analyzing chart patterns and technical indicators to predict stock price trends. However, this method overly relies on historical data and is slow to respond to new market information. When major unexpected events or new trading patterns occur in the market, it is difficult to quickly capture market changes, resulting in prediction deviations. Especially in today's globalized and information-based market environment, multi-source heterogeneous data such as news, social media, and macroeconomic data are emerging continuously. Traditional methods are unable to handle these massive data effectively in terms of integration and analysis, and thus cannot timely and accurately grasp market dynamics, making it difficult to conduct effective warnings. Summary of the Invention

[0004] The present invention provides a dynamic monitoring and warning system for the stock market based on artificial intelligence and its method, whose main purpose is to improve the effectiveness of dynamic monitoring and warning of the stock market based on artificial intelligence.

[0005] To achieve the above objective, the dynamic monitoring and warning system for the stock market based on artificial intelligence and its method provided by the present invention include: a data fusion module, an impact event analysis module, a market pressure analysis module, and a fact warning module;

[0006] The data fusion module is used to collect heterogeneous data, high-frequency order data, and media data from different data sources of the stock market in real time, construct a knowledge graph of the stock market based on the heterogeneous data, identify data features of the high-frequency order data to determine order features of the stock market, conduct sentiment quantification processing on the stock market based on the media data to obtain a sentiment quantification index, and perform recursive feature elimination processing on the knowledge graph, the order features, and the sentiment quantification index to obtain fusion data;

[0007] The impact event analysis module is used to collect historical data of the stock market, divide the market state of the stock market based on the historical data to obtain a defined market state, query current news information of the stock market, and predict impact events on the stock market based on the current news information and the fusion data to obtain an event impact chain;

[0008] The market pressure analysis module is used to collect investment data of stock users corresponding to the stock market and risk index analysis requirements, calculate the investment risk value of the stock users based on the investment data and the risk index analysis requirements, collect market liquidity indicators of the stock market, and analyze the liquidity pressure of the stock market based on the market liquidity indicators;

[0009] The fact warning module is used to construct a joint warning system for the stock market by using the defined market state, the event impact chain, the investment risk value and the liquidity pressure, collect market data of the stock market in real time, and give a real-time warning to the stock users by using the joint warning system based on the market data.

[0010] Optionally, constructing the knowledge graph of the stock market based on the heterogeneous data includes:

[0011] Performing data preprocessing on the heterogeneous data to obtain preprocessed data;

[0012] Extracting entity information, relationship information and attribute information of the stock market from the preprocessed data to obtain knowledge data;

[0013] Performing entity alignment on the knowledge data to obtain entity alignment data;

[0014] Performing knowledge merging on the entity alignment data to obtain knowledge merging data

[0015] Using the knowledge merging data to construct the knowledge graph of the stock market.

[0016] Optionally, performing sentiment quantification processing on the stock market based on the media data to obtain sentiment quantification indicators, including:

[0017] Extracting user sentiment tags from the media data;

[0018] Using the user sentiment tags to construct a user influence factor for the stock market;

[0019] Querying the time axis information of the media data to construct a time decay coefficient matrix of the user influence factor;

[0020] Calculate the dynamic weights of the stock market by using the user influence factor and the time decay coefficient matrix;

[0021] Calculate the weighted sentiment index of the stock market by using the dynamic weights to obtain a sentiment quantification index.

[0022] Optionally, the recursive feature elimination process for the knowledge graph, the order features, and the sentiment quantification index to obtain the fusion data includes:

[0023] Extract multi-dimensional key features from the fusion data and construct an initial feature set of the fusion data by using the extracted key features;

[0024] After inputting the initial feature set into a pre-configured recursive feature elimination model, calculate the output contribution scores of the initial feature set by using the recursive feature elimination model;

[0025] Perform initial feature elimination on the initial feature set based on the output contribution scores to obtain preliminary feature processing data;

[0026] Use the preliminary feature processing data to train the recursive feature elimination model to obtain a trained model;

[0027] Use the trained model to iteratively eliminate the preliminary feature processing data to obtain target feature data;

[0028] Perform data fusion on the target feature data to obtain fusion data.

[0029] Optionally, the market state division of the stock market based on the historical data to obtain a defined market state includes:

[0030] Construct a market evaluation index for the stock market;

[0031] Calculate the index statistical value of the market evaluation index;

[0032] Analyze the market average level and market volatility of the stock market by using the index statistical value to determine the market static characteristics of the stock market;

[0033] Perform exponential smoothing on the historical data to analyze the market dynamic characteristics of the stock market;

[0034] Perform clustering analysis on the market static characteristics and the market dynamic characteristics;

[0035] Based on the analysis results of the clustering analysis on the

[0036] Optionally, based on the current affairs news information and the fusion data, predicting impact events on the stock market to obtain an event impact chain, including:

[0037] Extracting multi-source events from the current affairs news information and performing event structuring processing on the multi-source events to obtain structured events;

[0038] Constructing an event association graph of the structured events;

[0039] Collecting real-time market data of the stock market;

[0040] Using the real-time market data and the event association graph to analyze the event impact intensity of the stock market to obtain an event intensity matrix;

[0041] Using the event intensity matrix to simulate the impact propagation path of the stock market to obtain a simulated link;

[0042] Constructing a dynamic impact chain graph of the simulated link to obtain an event impact chain.

[0043] Optionally, the using the real-time market data and the event association graph to analyze the event impact intensity of the stock market to obtain an event intensity matrix includes:

[0044] Converting the event association graph into a Bayesian network;

[0045] Based on the Bayesian network, using the following formula to calculate the posterior probability of the impact event of the stock market:

[0046]

[0047] where P(E i |D m ) represents the posterior probability, n represents the number of nodes in the Bayesian network, E i represents the i-th node in the Bayesian network (i.e., the i-th event in the impact event), D m represents the m-th data in the real-time market data, P(E i ) represents the prior probability of E i , P(D m |E i ) represents the probability that the market data D m appears under the condition that E j occurs, E j represents the j-th node in the Bayesian network (i.e., the j-th event in the impact event), P(E j ) represents the prior probability of E m , P(D j |E j ) represents the probability that the market data Dj Under the occurring conditions, market data D m The probability of occurrence,

[0048] According to the posterior probability, use the following formula to calculate the event shock intensity of the shock event in the stock market:

[0049] S i = k × P(E i |D m ) × f(X)

[0050] Where S i represents the event shock intensity of the i-th shock event, k represents a constant, P(E i |D m ) represents the posterior probability, and f(X) represents an exponential function (the exponential change rate of the event shock intensity),

[0051] Use the event shock intensities of all events in the shock event as matrix elements to construct the event intensity matrix of the stock market.

[0052] Optionally, calculating the investment risk value of the stock user based on the investment data and the risk index analysis requirements includes:

[0053] Perform multi-dimensional preprocessing on the investment data to obtain a standardized data set;

[0054] Use the standardized data set to perform hierarchical quantization on the risk index to obtain a risk index;

[0055] Use the risk index analysis requirements to construct the risk preference label of the stock user;

[0056] Use the risk index and the risk preference label to construct a dynamic weight vector of the investment risk of the stock user;

[0057] Use the dynamic weight vector and the risk index to calculate the investment risk value of the stock user.

[0058] Optionally, constructing the joint warning system of the stock market using the defined market state, the event shock chain, the investment risk value, and the liquidity pressure includes:

[0059] Perform spatio-temporal alignment on the defined market state, the event shock chain, the investment risk value, and the liquidity pressure to obtain aligned data;

[0060] Perform state encoding on the aligned data to obtain a multi-dimensional state matrix;

[0061] Perceive the state of the stock market using the multi-dimensional state matrix to obtain a perceived state;

[0062] Based on the perceived state, construct a risk dynamic weight for the stock market;

[0063] Use the multi-dimensional state matrix and the risk dynamic weight to construct a multi-factor fusion early warning model for the stock market to obtain a joint early warning system.

[0064] A method for dynamic monitoring and early warning of the stock market based on artificial intelligence, characterized in that the method includes:

[0065] Collect heterogeneous data, high-frequency order data, and media data from different data sources in the stock market in real time. Based on the heterogeneous data, construct a knowledge graph of the stock market, identify the data characteristics of the high-frequency order data to determine the order characteristics of the stock market. Based on the media data, perform sentiment quantification processing on the stock market to obtain sentiment quantification indicators, and perform recursive feature elimination processing on the knowledge graph, the order characteristics, and the sentiment quantification indicators to obtain fusion data;

[0066] Collect historical data of the stock market, divide the market state of the stock market based on the historical data to obtain a defined market state, query the current news information of the stock market, and predict the impact events of the stock market based on the current news information and the fusion data to obtain an event impact chain;

[0067] Collect the investment data and risk indicator analysis requirements of stock users corresponding to the stock market. Based on the investment data and the risk indicator analysis requirements, calculate the investment risk value of the stock users. Collect the market liquidity indicators of the stock market and analyze the liquidity pressure of the stock market based on the market liquidity indicators;

[0068] Use the defined market state, the event impact chain, the investment risk value, and the liquidity pressure to construct a joint early warning system for the stock market. Collect market data of the stock market in real time, and use the joint early warning system to perform real-time early warning on the stock users based on the market data.

[0069] Compared with the prior art, the present invention first collects heterogeneous and high-frequency order and media data in real time to provide multi-faceted information, laying a data foundation for monitoring and early warning. Then, based on the heterogeneous data, a knowledge graph is constructed, the characteristics of high-frequency order data are identified to determine order characteristics, and the media data is subjected to sentiment quantification processing to obtain sentiment quantification indicators, and then data fusion is performed to comprehensively and refinedly describe market characteristics, providing high-quality data for subsequent analysis and prediction. Further, the present invention helps users understand the market development track and analyze the law and trend by collecting historical data, and then divides the market state based on the historical data, clearly defining the market state, providing a background reference for investment decisions and shock event predictions. By querying current affairs news information, instant information is provided for shock event predictions, and then shock events are predicted based on current affairs news and fused data to obtain an event shock chain, helping market participants understand the chain reaction, layout in advance to respond, reduce risks and grasp benefits. Further, the present invention collects user investment data and risk index analysis requirements, and can provide customized risk analysis and investment suggestions for users. Then, market liquidity indicators are collected to understand the market trading atmosphere and participation degree, and the liquidity pressure is analyzed based on the indicators to timely discover potential pressure and give early warnings to help investors adjust their strategies. Furthermore, the embodiments of the present invention construct a joint early warning system by using the defined market state, event shock chain, investment risk value and liquidity pressure, comprehensively considering market elements, and providing more accurate risk early warnings. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 FIG. is a functional module diagram of an artificial intelligence-based dynamic monitoring and early warning system and method for the stock market provided by an embodiment of the present invention;

[0071] Figure 2 FIG. is a flowchart of an artificial intelligence-based dynamic monitoring and early warning method for the stock market provided by an embodiment of the present invention;

[0072] The realization, functional features and advantages of the object of the present invention will be further described in conjunction with the embodiments with reference to the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0073] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0074] In addition, the step timing in the following method embodiments is only an example, not strictly limited.

[0075] In fact, the server devices deployed by the artificial intelligence-based stock market dynamic monitoring and early warning system may consist of one or more devices. The above-mentioned artificial intelligence-based stock market dynamic monitoring and early warning system can be implemented as: business instances, virtual machines, and hardware devices. For example, the artificial intelligence-based stock market dynamic monitoring and early warning system can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, the artificial intelligence-based stock market dynamic monitoring and early warning system can be understood as a software deployed on a cloud node, used to provide services for artificial intelligence-based stock market dynamic monitoring and early warning for each client. Alternatively, the artificial intelligence-based stock market dynamic monitoring and early warning system can also be implemented as a virtual machine deployed on one or more devices in a cloud node. An application software for managing each client is installed in the virtual machine. Alternatively, the artificial intelligence-based stock market dynamic monitoring and early warning system can also be implemented as a server composed of many identical or different types of hardware devices, and one or more hardware devices are set to provide services for artificial intelligence-based stock market dynamic monitoring and early warning for each client.

[0076] In terms of implementation form, the artificial intelligence-based stock market dynamic monitoring and early warning system and the client adapt to each other. That is, if the artificial intelligence-based stock market dynamic monitoring and early warning system is an application installed on a cloud service platform, then the client is a client that establishes a communication connection with this application; or if the artificial intelligence-based stock market dynamic monitoring and early warning system is implemented as a website, then the client is implemented as a web page; or if the artificial intelligence-based stock market dynamic monitoring and early warning system is implemented as a cloud service platform, then the client is implemented as a small program in an instant messaging application.

[0077] Refer to Figure 1 As shown, it is a functional module diagram of the artificial intelligence-based stock market dynamic monitoring and early warning system provided by an embodiment of the present invention.

[0078] The artificial intelligence-based stock market dynamic monitoring and early warning system 100 described in the present invention can be set in a cloud server. In terms of implementation form, it can be used as one or more service devices, or can be installed as an application on the cloud (such as a server for artificial intelligence-based stock market dynamic monitoring and early warning, a server cluster, etc.), or can also be developed as a website. According to the functions achieved, the artificial intelligence-based stock market dynamic monitoring and early warning system 100 includes a data fusion module 101, an impact event analysis module 102, a market pressure analysis module 103, and a fact early warning module 104.

[0079] In the embodiments of the present invention, in the tracking of dynamic monitoring and early warning of the stock market based on artificial intelligence, each of the above modules can be independently implemented and called with other modules. Here, the call can be understood as that a certain module can be connected to multiple modules of another type and provide corresponding services for the multiple modules it is connected to. In the dynamic monitoring and early warning system of the stock market based on artificial intelligence provided by the embodiments of the present invention, without modifying the program code, the applicable scope of the dynamic monitoring and early warning architecture of the stock market based on artificial intelligence can be adjusted by adding modules and directly calling, so as to achieve cluster-level horizontal expansion, so as to achieve the purpose of quickly and flexibly expanding the dynamic monitoring and early warning system of the stock market based on artificial intelligence. In practical applications, the above modules can be set in the same device or different devices, or can be set in virtual devices, such as service instances in cloud servers.

[0080] The following will respectively describe the various components and specific working processes of the dynamic monitoring and early warning system of the stock market based on artificial intelligence in combination with specific embodiments.

[0081] The data fusion module is used to collect heterogeneous data, high-frequency order data and media data from different data sources in the stock market in real time, construct a knowledge graph of the stock market based on the heterogeneous data, identify the data characteristics of the high-frequency order data to determine the order characteristics of the stock market, and perform sentiment quantification processing on the stock market based on the media data to obtain sentiment quantification indicators, and perform recursive feature elimination processing on the knowledge graph, the order characteristics and the sentiment quantification indicators to obtain fusion data.

[0082] In the embodiments of the present invention, by collecting heterogeneous data, high-frequency order data and media data from different data sources in the stock market in real time, data covering multiple aspects such as stock market transactions, company fundamentals, macroeconomics and the market can be obtained, providing strong data support for monitoring and warning the stock market.

[0083] Among them, the heterogeneous data refers to data from different data sources with different structures and formats, such as financial data, market data and company announcements, etc.; the high-frequency order data refers to a large amount of stock trading order information generated in a short time, such as order price, order volume and order flow, etc.; the media data refers to information data related to the stock market from various media channels, such as social media data, news media data and research reports, etc.

[0084] Optionally, the heterogeneous data, high-frequency order data and media data can be jointly obtained through professional financial data platforms, exchanges, financial institutions, web crawlers and social media platforms.

[0085] Furthermore, in the embodiment of the present invention, by constructing the knowledge graph of the stock market based on the heterogeneous data, entities (such as stocks, companies, industries, etc.) and their relationships (such as ownership relationships, association relationships, etc.) in the heterogeneous data can be sorted out and visually presented, and constructed into a structured knowledge network, thereby facilitating users to systematically understand and analyze the stock market.

[0086] Among them, the knowledge graph refers to a graph-based data structure, which consists of nodes and edges and is used to represent and store knowledge in a structured manner.

[0087] As an embodiment of the present invention, constructing the knowledge graph of the stock market based on the heterogeneous data includes: performing data preprocessing on the heterogeneous data to obtain preprocessed data, extracting entity information, relationship information, and attribute information of the stock market from the preprocessed data to obtain knowledge data, performing entity alignment on the knowledge data to obtain entity alignment data, performing knowledge merging on the entity alignment data to obtain knowledge merged data, and constructing the knowledge graph of the stock market by using the knowledge merged data.

[0088] Among them, the entity information refers to the basic elements in the knowledge graph, representing things or concepts with independent existence significance and distinguishability in the stock market, such as company types, person types, and transaction types. The relationship information refers to the description of the association and interaction between entities and is the bridge connecting various entities in the knowledge graph, such as the relationship between transactions and companies. The attribute information refers to the description of the characteristics and properties of entities, which is used to further enrich the information of entities and make the entities have more specific and detailed feature descriptions in the knowledge graph.

[0089] Optionally, the preprocessed data can be obtained by performing data cleaning, duplicate removal, and format normalization on the heterogeneous data. The knowledge data can be obtained by using named entity recognition technology to extract entity information of the stock market from the preprocessed data; by analyzing syntax and semantics through a relationship extraction algorithm to obtain relationship information; and by using a regular expression method to extract attribute information. The entity alignment data can be obtained by calculating the similarity of entities in the knowledge data. Based on features such as the names and attribute values of entities, a similarity calculation algorithm, such as an edit distance algorithm, can be used to match the same entities. The knowledge graph can be constructed based on the knowledge merged data by selecting a graph model such as RDF or an attribute graph, using entities as nodes and relationships as edges, and using graph database storage technology.

[0090] In the embodiment of the present invention, by identifying the data characteristics of the high-frequency order data to determine the order characteristics of the stock market, it can help users understand information such as the trading activity, capital flow, and investor trading behavior patterns in the stock market, thereby helping investment users grasp short-term market dynamics.

[0091] Optionally, the order features can use statistical analysis techniques to extract basic features such as the time interval, trading volume, and trading price of the order, then use clustering analysis techniques to cluster the basic features to discover different types of order patterns, and then use machine learning algorithms such as decision trees to build models for analysis and obtain.

[0092] Furthermore, in the embodiment of the present invention, by performing sentiment quantification processing on the stock market based on the media data, the obtained sentiment quantification index can help stock users understand the overall sentiment atmosphere of the stock market, so as to assist investment decisions and also assist in early warning.

[0093] As an embodiment of the present invention, performing sentiment quantification processing on the stock market based on the media data to obtain a sentiment quantification index includes: extracting user sentiment tags in the media data, using the user sentiment tags to construct a user influence factor of the stock market, querying the time axis information of the media data to construct a time decay coefficient matrix of the user influence factor, using the user influence factor and the time decay coefficient matrix to calculate the dynamic weight of the stock market, and using the dynamic weight to calculate the weighted sentiment index of the stock market to obtain the sentiment quantification index.

[0094] Among them, the user sentiment tag refers to the information transformed into a structured form through a sentiment analysis model, the user influence factor refers to a set of quantitative indicators of the degree of influence of user sentiment on the market, such as dissemination power, credibility, etc., and the time decay coefficient refers to acting on the influence factor to ensure that the model pays more attention to recent effective information.

[0095] Optionally, the user sentiment tag can be obtained by using a sentiment analysis model in the financial field (FinBERT) to perform sentiment classification on the media data to generate sentiment data including positive, neutral, negative tags and their confidence scores. The user influence factor can be obtained by calculating the correlation coefficient between the corresponding user historical sentiment tags in the user sentiment tags and the rise and fall of the corresponding stock within 3 days. The time decay coefficient matrix can be obtained by applying an exponential decay function to calculate the weight value according to the time difference between the media data releases (Δt = current time - posting time). The dynamic weight can be obtained by multiplying the user influence factor group and the time decay coefficient element by element. The weighted sentiment index can be calculated using the following formula:

[0096]

[0097] Among them, Q represents the weighted sentiment index.

[0098] Furthermore, in the embodiment of the present invention, through the recursive feature elimination processing of the knowledge graph, the order features, and the emotion quantification index, the obtained fusion data can more comprehensively and refinedly describe the characteristics of the stock market, providing a high-quality data basis for subsequent in-depth analysis and prediction.

Claims

1. An artificial intelligence-based dynamic monitoring and early warning system for the stock market, characterized in that, The system for dynamic monitoring and early warning of the stock market based on artificial intelligence includes: a data fusion module, an impact event analysis module, a market pressure analysis module, and a fact early warning module; The data fusion module is used to collect heterogeneous data, high-frequency order data, and media data from different data sources in the stock market in real time. Based on the heterogeneous data, construct a knowledge graph of the stock market, identify the data characteristics of the high-frequency order data to determine the order characteristics of the stock market, and based on the media data, perform sentiment quantification processing on the stock market to obtain a sentiment quantification index. Perform recursive feature elimination processing on the knowledge graph, the order characteristics, and the sentiment quantification index to obtain fusion data; The impact event analysis module is used to collect historical data of the stock market, divide the market state of the stock market based on the historical data to obtain a defined market state, query the current affairs news information of the stock market, and predict impact events on the stock market based on the current affairs news information and the fusion data to obtain an event impact chain; The market pressure analysis module is used to collect the investment data of stock users corresponding to the stock market and the requirements for risk index analysis. Based on the investment data and the requirements for risk index analysis, calculate the investment risk value of the stock users, collect the market liquidity index of the stock market, and analyze the liquidity pressure of the stock market based on the market liquidity index; The fact early warning module is used to utilize the defined market state, the event impact chain, the investment risk value, and the liquidity pressure to construct a joint early warning system for the stock market, collect market data of the stock market in real time, and use the joint early warning system to perform real-time early warning on the stock users based on the market data.

2. The artificial intelligence-based stock market dynamic monitoring and warning system according to claim 1, characterized in that, The construction of the knowledge graph of the stock market based on the heterogeneous data includes: Perform data preprocessing on the heterogeneous data to obtain preprocessed data; Extract entity information, relationship information, and attribute information of the stock market from the preprocessed data to obtain knowledge data; Perform entity alignment on the knowledge data to obtain entity alignment data; Perform knowledge merging on the entity alignment data to obtain knowledge merged data Use the knowledge merged data to construct the knowledge graph of the stock market.

3. The artificial intelligence-based stock market dynamic monitoring and early warning system according to claim 1, characterized in that, The sentiment quantification processing of the stock market based on the media data to obtain a sentiment quantification index includes: Extract user sentiment tags from the media data; Use the user sentiment tags to construct a user influence factor for the stock market; Query the time axis information of the media data to construct a time decay coefficient matrix of the user influence factor; Use the user influence factor and the time decay coefficient matrix to calculate the dynamic weight of the stock market; Use the dynamic weight to calculate the weighted sentiment index of the stock market to obtain a sentiment quantification index.

4. The artificial intelligence-based stock market dynamic monitoring and early warning system according to claim 1, characterized in that The recursive feature elimination processing of the knowledge graph, the order characteristics, and the sentiment quantification index to obtain fusion data includes: Extract multi-dimensional key features from the fused data, and construct an initial feature set of the fused data using the extracted key features; After inputting the initial feature set into a pre-configured recursive feature elimination model, use the recursive feature elimination model to calculate the output contribution scores of the initial feature set; Perform initial feature elimination on the initial feature set based on the output contribution scores to obtain preliminary feature processing data; Use the preliminary feature processing data to train the recursive feature elimination model to obtain a trained model; Use the trained model to iteratively eliminate the preliminary feature processing data to obtain target feature data; Perform data fusion on the target feature data to obtain fused data.

5. The artificial intelligence-based stock market dynamic monitoring and early warning system according to claim 1, characterized in that The market state of the stock market is divided based on the historical data to obtain a defined market state, including: Construct a market evaluation index for the stock market; Calculate the index statistical value of the market evaluation index; Use the index statistical value to analyze the market average level and market volatility of the stock market to determine the market static characteristics of the stock market; Perform exponential smoothing processing on the historical data to analyze the market dynamic characteristics of the stock market; Perform cluster analysis on the market static characteristics and the market dynamic characteristics; Divide the market state of the stock market based on the analysis results of the cluster analysis to obtain a defined market state.

6. The artificial intelligence-based stock market dynamic monitoring and early warning system according to claim 1, wherein, Based on the current affairs news information and the fused data, predict the impact events of the stock market to obtain an event impact chain, including: Extract multi-source events from the current affairs news information, and perform event structuring processing on the multi-source events to obtain structured events; Construct an event association graph of the structured events; Collect real-time market data of the stock market; Use the real-time market data and the event association graph to analyze the event impact intensity of the stock market to obtain an event intensity matrix; Use the event intensity matrix to simulate the impact propagation path of the stock market to obtain a simulated link; Construct a dynamic impact chain graph of the simulated link to obtain an event impact chain.

7. The artificial intelligence-based stock market dynamic monitoring and warning system according to claim 6, wherein The use of the real-time market data and the event association graph to analyze the event impact intensity of the stock market to obtain an event intensity matrix includes: Convert the event association graph into a Bayesian network; Based on the Bayesian network, use the following formula to calculate the posterior probability of the impact event of the stock market: Among them, P(E i |D m ) represents the posterior probability, n represents the number of nodes in the Bayesian network, E i represents the ith node in the Bayesian network (i.e., the ith event in the impact event), D m represents the mth data in the real-time market data, P(E i ) means E i The prior probability, P(D m |E i ) means E i Market data D m The probability of occurrence, E j represents the jth node in the Bayesian network (i.e., the jth event in the impact event), P(E j ) means E j The prior probability, P(D m |E j ) means E j Market data D m The probability of occurrence, According to the posterior probability, use the following formula to calculate the event impact intensity of the impact event of the stock market: S i = k × P(E i |D m ) × f(X) Among them, S i represents the event impact intensity of the i-th impact event, k represents a constant, P(E i |D m ) represents the posterior probability, and f(X) represents the exponential function (the exponential change rate of the event impact intensity). Use the event impact intensity of all events in the impact event as matrix elements to construct the event intensity matrix of the stock market.

8. The artificial intelligence-based stock market dynamic monitoring and early warning system according to claim 1, characterized in that, Based on the investment data and the risk indicator analysis requirements, calculate the investment risk value of the stock user, including: Perform multi-dimensional preprocessing on the investment data to obtain a standardized data set; Use the standardized data set to perform hierarchical quantization on the risk indicators to obtain risk indicators; Use the risk indicator analysis requirements to construct a risk preference label for the stock user; Construct a dynamic weight vector of the investment risk of the stock user by using the risk indicator and the risk preference label; Calculate the investment risk value of the stock user by using the dynamic weight vector and the risk indicator.

9. The artificial intelligence-based stock market dynamic monitoring and early warning system according to claim 1, characterized in that The construction of the joint warning system for the stock market by using the defined market state, the event impact chain, the investment risk value and the liquidity pressure includes: Perform spatio-temporal alignment on the defined market state, the event impact chain, the investment risk value and the liquidity pressure to obtain aligned data; Perform state encoding on the aligned data to obtain a multi-dimensional state matrix; Perform state perception on the stock market by using the multi-dimensional state matrix to obtain a perceived state; Construct the risk dynamic weight of the stock market based on the perceived state; Construct a multi-factor fusion warning model for the stock market by using the multi-dimensional state matrix and the risk dynamic weight to obtain a joint warning system.

10. A method for dynamic monitoring and early warning of the stock market based on artificial intelligence, characterized in that, The method includes: Collect heterogeneous data, high-frequency order data and media data from different data sources of the stock market in real time. Based on the heterogeneous data, construct a knowledge graph of the stock market, identify the data characteristics of the high-frequency order data to determine the order characteristics of the stock market. Based on the media data, perform sentiment quantification processing on the stock market to obtain sentiment quantification indicators, and perform recursive feature elimination processing on the knowledge graph, the order characteristics and the sentiment quantification indicators to obtain fusion data; Collect the historical data of the stock market, divide the market state of the stock market based on the historical data to obtain a defined market state, query the current affairs news information of the stock market, and predict the impact events of the stock market based on the current affairs news information and the fusion data to obtain an event impact chain; Collect the investment data and risk indicator analysis requirements of the stock user corresponding to the stock market. Based on the investment data and the risk indicator analysis requirements, calculate the investment risk value of the stock user. Collect the market liquidity indicator of the stock market and analyze the liquidity pressure of the stock market based on the market liquidity indicator; Construct a joint warning system for the stock market by using the defined market state, the event impact chain, the investment risk value and the liquidity pressure. Collect the market data of the stock market in real time, and use the joint warning system to perform real-time warning on the stock user based on the market data.

Images