Economic data intelligent analysis system and method based on big data

Abstract

The invention discloses an economic data intelligent analysis system and method based on big data, and the method comprises the steps: obtaining background data including public economic data and macrocurrency policy text description; a neural network model based on deep learning is adopted to carry out feature coding on the two features to enhance the understanding ability of the model for the deep meaning of background data, and public economic data item coding features and macrocurrency policy semantic coding features are obtained; performing collaborative analysis on the public economic data item coding characteristics and the macrocurrency policy semantic coding characteristics to mine deep semantic association between the public economic data item coding characteristics and the macrocurrency policy semantic coding characteristics, and capturing a dynamic relationship between the currency policy and an economic index based on the deep semantic association; and outputting an analysis result of the corresponding economic data analysis problem by using the large language model. Therefore, the system can provide more comprehensive and accurate economic data analysis results, and helps users to better understand economic phenomena and policy influences.

Description

Technical Field

[0001] The present application relates to the field of intelligent analysis, and more specifically, to an intelligent analysis system and method for economic data based on big data. Background Art

[0002] In today's complex and ever-changing economic environment, accurately interpreting and predicting economic development trends has become an important basis for government decision-making, corporate strategic planning, and personal investment. However, existing economic data analysis methods and technologies have many limitations and cannot meet the needs of modern society for efficient and accurate analysis.

[0003] Existing economic data analysis mainly relies on historical data and simple statistical models. Although these methods can reveal the basic laws of economic phenomena to a certain extent, they are often unable to cope with the dynamic impact of macroeconomic policies, especially monetary policies. Especially in the context of frequent adjustments in monetary policy, how to accurately capture its impact on different economic indicators has become a major challenge. In addition, unstructured public economic data and text descriptions of macro-monetary policies increase the complexity of data analysis, and existing tools and technologies often require a lot of manual intervention for data preprocessing, which is not only time-consuming and labor-intensive, but also prone to introduce subjective bias.

[0004] Therefore, we look forward to an optimized intelligent analysis system of economic data based on big data. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an intelligent analysis system and method for economic data based on big data, which obtains background data including public economic data and macro-monetary policy text descriptions, and adopts a neural network model based on deep learning to perform feature encoding on the two respectively to enhance the model's ability to understand the deep meaning of the background data, and obtains public economic data item coding features and macro-monetary policy semantic coding features. Further, the public economic data item coding features and the macro-monetary policy semantic coding features are collaboratively analyzed to dig out the deep semantic association between the two, and based on this, the dynamic relationship between monetary policy and economic indicators is captured. Finally, a large language model is used to output the analysis results of the corresponding economic data analysis problems. In this way, the system can provide more comprehensive and accurate economic data analysis results to help users better understand economic phenomena and policy impacts.

[0006] According to one aspect of the present application, there is provided an economic data intelligent analysis system based on big data, which includes:

[0007] A background data acquisition module, used to acquire background data, wherein the background data includes a text description of public economic data and macro-monetary policy;

[0008] An economic data analysis problem acquisition module for acquiring a natural language description of an economic data analysis problem input by a user;

[0009] A structured coding module for performing structured coding on the disclosed economic data to obtain a set of coded feature matrices of disclosed economic data items;

[0010] A semantic coding module for performing semantic coding on the text description of the macro monetary policy to obtain a macro monetary policy semantic coding vector;

[0011] A semantic depth collaborative analysis module for performing heterogeneous data collaborative parsing on the macro monetary policy semantic coding vector and the set of coded feature matrices of disclosed economic data items to obtain a background data depth semantic collaborative implicit coding vector;

[0012] An analysis result generation module for obtaining an economic data analysis result based on the background data depth semantic collaborative implicit coding vector.

[0013] According to another aspect of the present application, there is provided an intelligent economic data analysis method based on big data, which includes:

[0014] Acquiring background data, where the background data includes disclosed economic data and text descriptions of macro monetary policies;

[0015] Acquiring a natural language description of an economic data analysis problem input by a user;

[0016] Performing structured coding on the disclosed economic data to obtain a set of coded feature matrices of disclosed economic data items;

[0017] Performing semantic coding on the text description of the macro monetary policy to obtain a macro monetary policy semantic coding vector;

[0018] Performing heterogeneous data collaborative parsing on the macro monetary policy semantic coding vector and the set of coded feature matrices of disclosed economic data items to obtain a background data depth semantic collaborative implicit coding vector;

[0019] Obtaining an economic data analysis result based on the background data depth semantic collaborative implicit coding vector.

[0020] Compared with the prior art, the present application provides an intelligent analysis system and method for economic data based on big data, which obtains background data including public economic data and macro-monetary policy text descriptions, and uses a neural network model based on deep learning to perform feature encoding on the two respectively to enhance the model's ability to understand the deep meaning of the background data, thereby obtaining public economic data item encoding features and macro-monetary policy semantic encoding features. Furthermore, the public economic data item encoding features and the macro-monetary policy semantic encoding features are collaboratively analyzed to dig out the deep semantic association between the two, and based on this, the dynamic relationship between monetary policy and economic indicators is captured. Finally, a large language model is used to output the analysis results of the corresponding economic data analysis problems. In this way, the system can provide more comprehensive and accurate economic data analysis results, helping users to better understand economic phenomena and policy impacts. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0022] Figure 1 A block diagram of an economic data intelligent analysis system based on big data according to an embodiment of the present application;

[0023] Figure 2 A data flow diagram of an economic data intelligent analysis system based on big data according to an embodiment of the present application;

[0024] Figure 3 A block diagram of a semantic depth collaborative analysis module in an economic data intelligent analysis system based on big data according to an embodiment of the present application;

[0025] Figure 4 It is a flow chart of the intelligent analysis method of economic data based on big data according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0027] As shown in this application and the claims, unless the context clearly indicates otherwise, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. A method or device may also include other steps or elements.

[0028] Although this application makes various references to certain modules in the system according to the embodiments of this application, however, any number of different modules can be used and run on the user terminal and / or server. The modules are merely illustrative, and different aspects of the system and method can use different modules.

[0029] Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the operations before or below do not necessarily need to be executed precisely in order. On the contrary, according to the need, various steps can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0030] Next, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited by the exemplary embodiments described herein.

[0031] In the technical solution of this application, an intelligent economic data analysis system based on big data is proposed. Figure 1 FIG. is a block diagram of an intelligent economic data analysis system based on big data according to an embodiment of this application. Figure 2 FIG. is a schematic diagram of data flow of an intelligent economic data analysis system based on big data according to an embodiment of this application. As Figure 1 and Figure 2As shown, the big data-based intelligent economic data analysis system 300 according to an embodiment of the present application includes: a background data acquisition module 310 for acquiring background data, where the background data includes public economic data and text descriptions of macro monetary policies; an economic data analysis problem acquisition module 320 for acquiring a natural language description of an economic data analysis problem input by a user; a structured coding module 330 for performing structured coding on the public economic data to obtain a set of public economic data item coding feature matrices; a semantic coding module 340 for performing semantic coding on the text descriptions of the macro monetary policies to obtain macro monetary policy semantic coding vectors; a semantic depth collaborative analysis module 350 for performing heterogeneous data collaborative parsing on the macro monetary policy semantic coding vectors and the set of public economic data item coding feature matrices to obtain background data depth semantic collaborative implicit coding vectors; and an analysis result generation module 360 for obtaining an economic data analysis result based on the background data depth semantic collaborative implicit coding vectors.

[0032] Specifically, the background data acquisition module 310 is used to acquire background data, where the background data includes public economic data and text descriptions of macro monetary policies. Among them, public economic data usually contains data on key economic indicators such as GDP growth rate, unemployment rate, inflation rate, etc., which reflect the state of the national economic operation; while the text descriptions of macro monetary policies cover various policy statements, meeting minutes, interest rate resolutions, etc. issued by the central bank, and are important bases for understanding the trend of monetary policies. By acquiring these two types of data, a comprehensive and detailed background database can be constructed, laying a foundation for subsequent in-depth analysis.

[0033] Specifically, the economic data analysis problem acquisition module 320 is used to acquire a natural language description of an economic data analysis problem input by a user. Considering that in the actual scenario of economic data analysis, the needs of users are often diverse and dynamically changing, and natural language is the most direct way for users to express their needs and often contains rich semantic information, such as the focus of the problem, related economic indicators, and specific policies concerned by users. These information are crucial for improving the accuracy and depth of analysis. By acquiring the natural language description of the economic data analysis problem input by the user, the system can better focus on the economic data and policy trends most relevant to the user's concerns, thereby providing more targeted insights and suggestions, and improving the accuracy and relevance of the analysis.

[0034] In particular, the structured coding module 330 is used to perform structured coding on the public economic data to obtain a set of coding feature matrices of public economic data items. In the technical solution of the present application, first, the public economic data is split based on the data items to obtain a set of public economic data items; then, each public economic data item in the set of public economic data items is semantically embedded based on the BERT model to obtain a set of semantic embedding coding vectors of public economic data items; further, the set of semantic embedding coding vectors of public economic data items is input into the text convolutional neural network model to obtain a set of coding feature matrices of public economic data items. It should be understood that public economic data usually exists in a structured or semi-structured form, such as indicator data such as GDP growth rate, unemployment rate, and inflation rate. Although these data contain rich economic information, the economic laws and trends contained therein often need to be effectively captured through complex feature extraction. Traditional statistical methods or simple data processing techniques are difficult to cope with the complexity and diversity of data, and text convolutional neural networks, with their powerful feature extraction capabilities, can dig out deeper information from data. In order to extract more valuable feature information from the public economic data, the system uses a text convolutional neural network (TextCNN) model, which is a deep learning model specifically used to process text data and can effectively capture local features and global semantic information in the public economic data. Specifically, first, the input public economic data is converted into a numerical vector representation, such as mapping text data to a high-dimensional vector space through word embedding technology; then, the convolution layer uses multiple convolution kernels of different sizes to perform sliding window operations on the data to extract local features in the data, such as the trend of economic indicators in a certain period of time or the correlation pattern between different indicators; then, the pooling layer performs dimensionality reduction processing on the extracted features to retain the most important feature information; finally, a set of public economic data item encoding feature matrices reflecting the deep semantics of the data is generated. These features not only contain the specific numerical information of economic indicators, but also capture the complex relationship and dynamic change trend between various economic indicators, comprehensively reflect the deep semantic information of public economic data, and provide high-quality feature input for subsequent analysis.

[0035] In particular, the semantic encoding module 340 is used to perform semantic encoding on the text description of the macro monetary policy to obtain a macro monetary policy semantic encoding vector. Since the text description of the macro monetary policy usually contains complex language structures and professional terms, directly processing these texts is a huge challenge for machine learning models. The policy semantic information contained in the text description of the macro monetary policy is of great significance for understanding economic phenomena and predicting economic trends. For example, the monetary policy statement issued by the central bank may contain key information about interest rate adjustments, changes in the money supply, or credit policy adjustments, and the impact of this information on economic indicators is often complex and dynamic. In order to extract valuable semantic information from these texts, in the technical solution of this application, semantic encoding is performed on the text description of the macro monetary policy to obtain a macro monetary policy semantic encoding vector. Specifically, first, one-hot encoding is performed on the text description of the macro monetary policy to convert it into multiple one-hot encoding vectors. One-hot encoding is a common method for converting discrete features into numerical vectors, which can map each word in the text to a unique vector representation; then, these one-hot encoding vectors are input into the macro monetary policy context encoder based on a transformer, which captures the context relationships in the text through the self-attention mechanism to generate multiple macro monetary policy semantic feature vectors; finally, the system concatenates these semantic feature vectors to generate a macro monetary policy semantic encoding vector. In this way, more comprehensive and accurate semantic information can be extracted from the text description of the macro monetary policy, enhancing the system's ability to understand the deep meaning of the macro monetary policy and enabling the system to more accurately evaluate the actual impact of the monetary policy on the economic environment.

[0036] In particular, the semantic depth collaborative analysis module 350 is used to perform heterogeneous data collaborative parsing on the set of the macro monetary policy semantic encoding vectors and the public economic data item encoding feature matrices to obtain background data depth semantic collaborative implicit encoding vectors. It should be understood that the set of the public economic data item encoding feature matrices and the macro monetary policy semantic encoding vectors respectively contain important information about the economic operation state, and the deep semantic association between the macro monetary policy and the public economic data is of great significance for accurately understanding economic phenomena and predicting economic trends. However, due to the significant heterogeneity in data form, representation space, and information distribution between the two, traditional multimodal fusion methods often struggle to directly achieve effective interaction and joint encoding between them. For example, the "interest rate adjustment" in the macro monetary policy text description may have a deep association with the "inflation rate" in the public economic data, but this association is often hidden in the complex structure and dynamic changes of the data and is difficult to capture through simple feature splicing or explicit alignment methods. Therefore, in the technical solution of this application, heterogeneous data collaborative parsing is performed on the set of the macro monetary policy semantic encoding vectors and the public economic data item encoding feature matrices to more accurately evaluate the dynamic relationship between the monetary policy and various economic indicators, so as to obtain background data depth semantic collaborative implicit encoding vectors. In this process, in order to extract this deep semantic association from heterogeneous data, the system adopts cross-domain joint encoding based on the weaving of core clues between modalities to perform heterogeneous data collaborative parsing on the two. As a deep learning architecture for multimodal data processing, this engine realizes the joint encoding of cross-modal features and multi-level semantic alignment by mining and modeling the core information of each modality. By performing heterogeneous data collaborative parsing on the set of the macro monetary policy semantic encoding vectors and the public economic data item encoding feature matrices, the system can construct a highly optimized and information-rich background data depth semantic collaborative implicit encoding vector. This process not only enhances the depth of understanding of economic phenomena and policy impacts but also provides a solid foundation for subsequent intelligent analysis of economic data based on large language models. In particular, in a specific example of this application, such as Figure 3As shown, the semantic depth collaborative analysis module 350 includes: a key clue extraction unit 351, configured to extract the core semantic association clues of the macro monetary policy semantic coding vector and the core clue coding vector of the public economic data item to obtain the core clue coding vector of the macro monetary policy and the core clue coding vector of the public economic data item; a core feature weaving unit 352, configured to construct an inter-modal core clue weaving template between the core clue coding vector of the macro monetary policy and the core clue coding vector of the public economic data item to obtain a monetary policy-economic data core clue weaving template matrix; and a feature interaction coding unit 353, configured to perform template-driven interaction coding on the macro monetary policy semantic coding vector and the core clue coding vector of the public economic data item based on the monetary policy-economic data core clue weaving template matrix to obtain the background data depth semantic collaborative implicit coding vector.

[0037] Specifically, the key clue extraction unit 351 is used to extract the core semantic association clues of the macro monetary policy semantic coding vector and the core clue coding vector of the public economic data item to obtain the core clue coding vector of the macro monetary policy and the core clue coding vector of the public economic data item. In the embodiment of the present application, first, the core clue extraction based on point convolution coding is performed on the macro monetary policy semantic coding vector to obtain the core clue coding vector of the macro monetary policy; it should be understood that in the monetary policy statement, some keywords may imply changes in future interest rate trends or fiscal stimulus measures, and this information is crucial for predicting its impact on the stock market. Through the point convolution coding operation, the system can effectively extract the most core and representative information from the complex monetary policy text, while maintaining the integrity of high-dimensional information, strengthening the semantically significant features, and forming a highly abstract but highly expressive core clue coding vector of the macro monetary policy. Through the core clue coding vector of the macro monetary policy, the system can more accurately analyze the change trend of relevant economic indicators, and then provide users with more in-depth insights. Next, the core clue coding vector of the public economic data item is extracted from the set of the public economic data item coding feature matrices; it should be understood that although the original public economic data contains rich information, it is also mixed with a large amount of background noise and non-critical data. For example, when analyzing the growth trend of a certain industry, some specific economic indicators (such as the employment rate of this industry or the price index of related products) may be decisive, while other data are relatively less important. Through the core clue extraction, the system can focus on the key factors that really affect economic phenomena, effectively distinguish important features from redundant features, eliminate possible information conflicts, and at the same time better capture the dynamic relationship between different economic variables and its impact on the target problem (such as the development trend of a specific industry). In an example, the sequence of significant region attention coefficients can be obtained by performing attention-based significant region attention analysis on each public economic data item coding feature matrix in the set of the public economic data item coding feature matrices; then, based on the sequence of the significant region attention coefficients, the core clue search is performed on the set of the public economic data item coding feature matrices to obtain the core clue coding vector of the public economic data item. Specifically, the specific steps for extracting the core clue coding vector of the public economic data item from the set of the public economic data item coding feature matrices are as follows: perform attention-based significant information attention analysis on each public economic data item coding feature matrix in the set of the public economic data item coding feature matrices to obtain the sequence of significant region attention coefficients; and based on the sequence of the significant region attention coefficients, perform the core clue search on the set of the public economic data item coding feature matrices to obtain the core clue coding vector of the public economic data item.More specifically, in a specific example of the present application, the core semantic association clue extraction formula is used to extract the core semantic association clues of the macro monetary policy semantic coding vector and the core clue coding vector of the public economic data item to obtain the core clue coding vector of the macro monetary policy and the core clue coding vector of the public economic data item; wherein, the core semantic association clue extraction formula is as follows.

[0038] v c1 = Leaky ReLU{Conv 1×1 (v1)+b1}

[0039]

[0040] Wherein, v1 is the macro monetary policy semantic coding vector, Conv 1×1 (·) is dot convolution processing, b1 is the bias vector, Leaky ReLU(·) is the Leaky ReLU activation function, v c1 represents the core clue coding vector of the macro monetary policy, M represents the set of public economic data item coding feature matrices, m1, m2,..., m i , m n respectively represent the 1st, 2nd, ith and nth public economic data item coding feature matrices, Conv 3×3 (·) is a convolutional layer with a 3×3 convolutional kernel, sigmoid is the sigmoid activation function, m i ′ represents the deep semantic feature matrix of the ith public economic data item, reshape(·) represents feature shape reshaping, v 2i represents the public economic data item coding feature vector corresponding to m i ′, max(·) and min(·) are respectively the maximum and minimum values in the vector, α, β and γ are all weight coefficients, μ(·) and σ(·) 2 are respectively the mean and variance of the vector, D i is the significant region attention focus coefficient, exp(·) is the exponential function value with the natural constant e as the base, k represents the number of feature matrices in the set of public economic data item coding feature matrices, v c2 represents the core clue coding vector of the public economic data item.

[0041] Specifically, the core feature weaving unit 352 is used to construct an inter-modal core clue weaving template between the core clue encoding vector of the macro monetary policy and the core clue encoding vector of the public economic data item to obtain a monetary policy-economic data core clue weaving template matrix. Since the macro monetary policy text and the public economic data each have unique information forms and structures, in order to deeply explore the potential connection between the two to enhance the system's ability to understand the deep meaning of background data, an inter-modal core clue weaving template between the core clue encoding vector of the macro monetary policy and the core clue encoding vector of the public economic data item is constructed to obtain a monetary policy-economic data core clue weaving template matrix. In an example, it can be calculated through dot product similarity measurement, attention mapping mechanism or other methods. The calculated monetary policy-economic data core clue weaving template matrix essentially reflects the global semantic constraint between monetary policy and economic data, and can also capture the fine-grained relationship between local significant regions. In this way, the system can discover and utilize the internal connection between the two at a deep level, thereby providing support for more accurate economic analysis. Specifically, the specific steps for constructing the inter-modal core clue weaving template between the core clue encoding vector of the macro monetary policy and the core clue encoding vector of the public economic data item are as follows: perform a linear mapping on the core clue encoding vector of the macro monetary policy and the core clue encoding vector of the public economic data item to obtain a monetary policy-economic data core correlation matrix; perform semantic alignment compensation on the monetary policy-economic data core correlation matrix to obtain the monetary policy-economic data core clue weaving template matrix. More specifically, in a specific example of the present application, the following template construction formula is used to construct an inter-modal core clue weaving template between the core clue encoding vector of the macro monetary policy and the core clue encoding vector of the public economic data item to obtain a monetary policy-economic data core clue weaving template matrix; where, the template construction formula is:

[0042]

[0043] Wherein, and φ(·) are respectively feature mapping functions, such as linear mapping or non-linear kernel function, M tem represents the monetary policy-economic data core clue weaving template matrix.

[0044] Specifically, the feature interaction encoding unit 353 is used to perform template-driven interaction encoding on the macro monetary policy semantic encoding vector and the core clue encoding vector of the public economic data item based on the monetary policy-economic data core clue weaving template matrix to obtain the background data deep semantic collaborative implicit encoding vector. Specifically, first, taking the macro monetary policy semantic encoding vector as the query vector, each public economic data item encoding feature matrix in the set of public economic data item encoding feature matrices as the key matrix, and the monetary policy-economic data core clue weaving template matrix as the prior information constraint matrix, and inputting them into the cross-modal template constraint encoder based on the heterogeneous transformer to obtain the set of cross-modal fine-grained interaction encoding vectors of the monetary policy-economic data; that is, by using the heterogeneous transformer and introducing the monetary policy-economic data core clue weaving template matrix as the prior information constraint matrix to discover and utilize the internal relationship between the monetary policy-economic data at a deep level, so as to provide support for more accurate economic analysis. Specifically, when evaluating the impact of a certain specific monetary policy on the development of the manufacturing industry, the system uses the macro monetary policy vector after semantic encoding as the query vector, obtains the response from the set of public economic data item encoding feature matrices, and each cross-modal fine-grained interaction encoding vector in the obtained set of cross-modal fine-grained interaction encoding vectors of the monetary policy-economic data reflects the fine interaction mode between the monetary policy and the specific economic data, enabling the system to provide more comprehensive and accurate economic prediction and analysis on the premise of combining the two. Furthermore, calculate the position-wise mean vector of the set of cross-modal fine-grained interaction encoding vectors of the monetary policy-economic data to obtain the background data deep semantic collaborative implicit encoding vector. It should be understood that each cross-modal fine-grained interaction encoding vector in the set of cross-modal fine-grained interaction encoding vectors of the monetary policy-economic data reflects the fine interaction mode between the monetary policy and the specific economic data. However, the number of these vectors is large and each represents different local features or interactions between regions. In order to extract the most representative and general information from these rich information, in the technical solution of this application, calculate the position-wise mean vector of the set of cross-modal fine-grained interaction encoding vectors of the monetary policy-economic data to obtain the background data deep semantic collaborative implicit encoding vector. By calculating the position-wise mean vector, a background data deep semantic collaborative implicit encoding vector that comprehensively reflects the deep relationship between the two can be obtained. This vector not only condenses the key information in the original data but also reveals the dynamic relationship between different economic variables and its impact on the target problem (such as the development trend of the information technology industry).This comprehensive analysis method not only improves the quality of data analysis but also provides a solid foundation for the subsequent intelligent analysis engine of economic data based on large language models, enabling the system to output more comprehensive, accurate, and understandable economic data analysis results, assisting users in making more informed choices in a complex economic environment. More specifically, in a specific example of this application, based on the template matrix for weaving the core clues of monetary policy - economic data, the following interactive encoding formula is used to perform template-driven interactive encoding on the macro monetary policy semantic encoding vector and the core clue encoding vector of the public economic data items to obtain the deep semantic collaborative implicit encoding vector of the background data; where the interactive encoding formula is:.

[0045]

[0046] Among them, L is the scale of m i , that is, the width of the matrix multiplied by the height, is matrix multiplication, softmax(·) represents the softmax function, and v ti represents the i-th monetary policy - economic data cross-modal fine-grained interactive encoding vector in the set of the monetary policy - economic data cross-modal fine-grained interactive encoding vectors, d is the number of vectors in the set of the monetary policy - economic data cross-modal fine-grained interactive encoding vectors, and v f represents the deep semantic collaborative implicit encoding vector of the background data.

[0047] Specifically, the analysis result generation module 360 is used to obtain economic data analysis results based on the deep semantic collaborative implicit encoding vector of the background data. In the technical solution of this application, after adding the natural language description of the economic data analysis problem to the tail of the deep semantic collaborative implicit encoding vector of the background data, it is input into the intelligent analysis engine of economic data based on large language models to obtain the economic data analysis results. It should be understood that although the deep semantic collaborative implicit encoding vector of the background data already contains rich information, it is essentially an abstract mathematical representation form, and it is still challenging to directly extract useful information for specific economic problems from it. Therefore, adding the natural language description of the specific economic data analysis problem to the tail of this vector can provide clear directions and context information for the subsequent large language model. In this way, the system can be more focused on solving the specific problems that users care about, better combining the deep meaning of the background data with the actual needs of users, so as to provide more personalized and targeted analysis results. In this way, the system not only improves the depth of understanding of the complex economic environment but also provides more accurate and personalized service support for users.

[0048] As described above, the intelligent economic data analysis system 300 based on big data according to the embodiments of the present application can be implemented in various wireless terminals, such as a server with an intelligent economic data analysis algorithm based on big data. In a possible implementation manner, the intelligent economic data analysis system 300 based on big data according to the embodiments of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the intelligent economic data analysis system 300 based on big data can be a software module in the operating system of the wireless terminal, or can be an application program developed for the wireless terminal; of course, the intelligent economic data analysis system 300 based on big data can also be one of the many hardware modules of the wireless terminal.

[0049] Alternatively, in another example, the intelligent economic data analysis system 300 based on big data and the wireless terminal can also be separate devices, and the intelligent economic data analysis system 300 based on big data can be connected to the wireless terminal through a wired and / or wireless network, and transmit and interact information according to a predefined data format.

[0050] Furthermore, an intelligent economic data analysis method based on big data is also provided.

[0051] Figure 4 It is a flowchart of the intelligent economic data analysis method based on big data according to the embodiments of the present application. As Figure 4 shown, the intelligent economic data analysis method based on big data according to the embodiments of the present application includes: S1, obtaining background data, where the background data includes public economic data and text descriptions of macro monetary policies; S2, obtaining a natural language description of an economic data analysis problem input by a user; S3, performing structured encoding on the public economic data to obtain a set of public economic data item encoding feature matrices; S4, performing semantic encoding on the text description of the macro monetary policy to obtain a macro monetary policy semantic encoding vector; S5, performing heterogeneous data collaborative parsing on the macro monetary policy semantic encoding vector and the set of public economic data item encoding feature matrices to obtain a background data deep semantic collaborative implicit encoding vector; S6, obtaining an economic data analysis result based on the background data deep semantic collaborative implicit encoding vector.

[0052] In summary, the wind-solar power generation energy storage management method according to the embodiments of the present application is clarified. It collects the energy storage parameters of the energy storage battery during the energy storage process, uses a deep neural network model as a feature extractor to capture the high-dimensional implicit features of each energy storage parameter and between each energy storage parameter, and performs decoding regression through a decoder to obtain a more accurate SOC measurement value.

[0053] The embodiments of the present disclosure have been described above. The above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to technologies in the market, or to enable other ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. Economic data intelligent analysis system based on big data, characterized by: include: A background data acquisition module, used to acquire background data, wherein the background data includes a text description of public economic data and macro-monetary policy; An economic data analysis problem acquisition module is used to acquire a natural language description of an economic data analysis problem input by a user; A structured coding module, used for performing structured coding on the public economic data to obtain a set of public economic data item coding feature matrices; A semantic coding module, used for semantically coding the text description of the macro-monetary policy to obtain a macro-monetary policy semantic coding vector; A semantic deep collaborative analysis module, used for performing heterogeneous data collaborative analysis on the set of the macro-monetary policy semantic coding vector and the public economic data item coding feature matrix to obtain a background data deep semantic collaborative implicit coding vector; The analysis result generation module is used to obtain the economic data analysis results based on the deep semantic collaborative implicit coding vector of the background data.

2. The economic data intelligent analysis system based on big data according to claim 1 is characterized in that: The structured coding module is used to: Performing data splitting on the public economic data based on the data items to obtain a set of public economic data items; Performing semantic embedding coding based on the BERT model on each public economic data item in the set of public economic data items to obtain a set of semantic embedding coding vectors of the public economic data items; The set of semantic embedding coding vectors of the public economic data items is input into a text convolutional neural network model to obtain a set of coding feature matrices of the public economic data items.

3. The economic data intelligent analysis system based on big data according to claim 2 is characterized in that: The semantic encoding module is used to: One-hot encoding the text description of the macro-monetary policy to obtain a plurality of macro-monetary policy one-hot encoding vectors; Inputting the plurality of one-hot encoded vectors into the transformer-based macro-monetary policy context encoder to obtain a plurality of macro-monetary policy semantic feature vectors; The multiple macro-monetary policy semantic feature vectors are cascaded to obtain the macro-monetary policy semantic coding vector.

4. The economic data intelligent analysis system based on big data according to claim 3 is characterized in that: The semantic depth collaborative analysis module includes: A key clue extraction unit, used for extracting the core semantic association clues of the macro-monetary policy semantic coding vector and the core clue coding vector of the public economic data item to obtain the macro-monetary policy core clue coding vector and the core clue coding vector of the public economic data item; A core feature weaving unit, used for constructing an inter-modal core clue weaving template between the macro-monetary policy core clue coding vector and the public economic data item core clue coding vector to obtain a monetary policy-economic data core clue weaving template matrix; A feature interaction coding unit is used to weave a template matrix based on the monetary policy-economic data core clues, and to perform template-driven interaction coding on the macro-monetary policy semantic coding vector and the public economic data item core clue coding vector to obtain the background data deep semantic collaborative implicit coding vector.

5. The economic data intelligent analysis system based on big data according to claim 4 is characterized in that: The key clue unit is used to: Performing point convolution coding-based core clue extraction on the macro-monetary policy semantic coding vector to obtain the macro-monetary policy core clue coding vector; Performing attention-based public economic data significant information attention analysis on each public economic data item coding feature matrix in the set of public economic data item coding feature matrices to obtain a sequence of significant area attention coefficients; Based on the sequence of the salient region attention coefficients, a core clue search is performed on the set of encoding feature matrices of the public economic data items to obtain the core clue encoding vector of the public economic data items.

6. The economic data intelligent analysis system based on big data according to claim 5 is characterized in that: The core feature weaving unit is used to: Linearly mapping the macro-monetary policy core clue coding vector and the public economic data item core clue coding vector to obtain a monetary policy-economic data core association matrix; The monetary policy-economic data core association matrix is ​​semantically aligned and compensated to obtain the monetary policy-economic data core clue weaving template matrix.

7. The economic data intelligent analysis system based on big data according to claim 6 is characterized in that: The core feature weaving unit is used to: Taking the macro-monetary policy semantic coding vector as the query vector, taking each public economic data item coding feature matrix in the set of public economic data item coding feature matrices as the key matrix, and taking the monetary policy-economic data core clue weaving template matrix as the prior information constraint matrix, inputting them into a cross-modal template constraint encoder based on a heterogeneous converter to obtain a set of monetary policy-economic data cross-modal fine-grained interactive coding vectors; The positional mean vector of the set of the monetary policy-economic data cross-modal fine-grained interaction coding vectors is calculated to obtain the background data deep semantic collaborative implicit coding vector.

8. The economic data intelligent analysis system based on big data according to claim 7 is characterized in that: The analysis result generating module is used for: After adding the natural language description of the economic data analysis problem to the tail of the background data deep semantic collaborative implicit coding vector, it is input into an economic data intelligent analysis engine based on a large language model to obtain the economic data analysis result.

9. A method for intelligent analysis of economic data based on big data, characterized in that: include: Obtaining background data, wherein the background data includes a text description of public economic data and macro-monetary policy; Obtaining a natural language description of an economic data analysis problem input by a user; Performing structured coding on the public economic data to obtain a set of public economic data item coding feature matrices; Performing semantic coding on the text description of the macro-monetary policy to obtain a macro-monetary policy semantic coding vector; Performing heterogeneous data collaborative analysis on the set of the macro-monetary policy semantic coding vector and the public economic data item coding feature matrix to obtain the background data deep semantic collaborative implicit coding vector; Based on the deep semantic collaborative implicit coding vector of the background data, the economic data analysis result is obtained.

Images