Precise investment attraction robot control system and method based on cross-platform AI

Abstract

The invention discloses a cross-platform AI-based precise investment attraction robot control system and method, and relates to the technical field of AI precise investment attraction robots, and the system comprises a multi-source data collection module, an industrial data management module, an enterprise feature extraction module, an enterprise uplink module, an industrial map module, a precise mining module, a robot management module, and an operation monitoring module. The system can give full play to the advantages of the big data and the artificial intelligence technology, comprehensively obtains the investment attraction industry chain data from different platforms, carries out the deep analysis and precise mining of the data, provides precise and efficient support for the investment attraction work, assists the improvement of the investment attraction efficiency and quality, and promotes the industrial development and economic growth. The risk assessment and value assessment model of the target enterprise is established to improve the investment attraction efficiency, reduce the investment risk, provide powerful support for precise investment attraction of the government, and solve the problems of incomplete data acquisition, non-deep analysis, inaccurate risk assessment and the like in the current investment attraction work.

Description

Technical Field

[0001] The present invention relates to the technical field of AI precise investment promotion robots, and particularly to a cross-platform artificial intelligence-based precise investment promotion robot control system and method thereof. Background Art

[0002] In the current field of investment promotion work, many thorny problems are faced. For example, data acquisition often lacks comprehensiveness, resulting in information loss; when analyzing the acquired data, it is often not in-depth enough to dig out valuable content; and the risk assessment is also inaccurate, which brings many uncertainties and potential risks to the investment promotion work.

[0003] In view of this, the present invention emerges as the times require. Its core goal is to carefully construct a precise investment promotion robot system based on the AI big data industrial chain. This system will comprehensively and fully utilize the two cutting-edge technologies of big data and artificial intelligence. On the one hand, it can widely and precisely obtain investment promotion industrial chain data from multiple sources, covering various dimensions and levels of data information, without missing any potentially valuable details; on the other hand, it deeply and carefully analyzes the massive data obtained to dig out the deep-seated laws and trends hidden behind the data.

[0004] Moreover, the system will also establish risk assessment and value assessment models for target enterprises. The risk assessment model can comprehensively consider various factors, including but not limited to the financial status of the enterprise, market environment, industry competition situation, etc., so as to accurately assess the risks that the target enterprise may face and provide a reliable risk warning for investment promotion decisions; the value assessment model will start from multiple angles such as the profitability, development potential, and innovation ability of the enterprise to accurately measure the value of the target enterprise and help the investment promotion party screen out the enterprises with the most investment value.

[0005] Through such a complete and advanced system, it is expected to significantly improve the efficiency of investment promotion work, make the investment promotion process more efficient and precise, and at the same time effectively reduce investment risks, provide strong technical support and decision-making basis for the government to carry out precise investment promotion work, help the government achieve more excellent results in investment promotion, and promote the vigorous development of the regional economy and the optimization and upgrading of the industrial structure. Summary of the Invention

[0006] The cross-platform AI precise investment promotion robot control system and method proposed by the present invention are used to solve the problems mentioned in the above-mentioned prior art.

[0007] To achieve the above object, the present invention adopts the following technical solution: A cross-platform AI-based precise investment promotion robot control system includes:

[0008] Multi-source data collection module: The multi-source data collection module supports two collection methods, namely parsing the page structure based on XPath and using a deep learning model to identify page image content. It sets multi-source data collection rules in a visual way and automatically completes data collection tasks in real time based on a scheduled task or event trigger mechanism.

[0009] Industrial data governance module: With the help of metadata management, data standard management, and data quality management data governance function components, an industrial data governance system is established to complete the aggregation, cleaning, processing, and modeling of multi-source industrial data.

[0010] Enterprise panoramic portrait module: Extract features from data that meet quality requirements, assign various data labels to enterprises, and quantitatively reflect the characteristics of enterprises in terms of industry, technology, capital, and talent.

[0011] Enterprise on-chain module: Use an artificial intelligence classification algorithm model to classify the leading industries of enterprises. Select linear kernel and Gaussian kernel functions to construct an SVM classification model. Use the enterprise data with known leading industries as the training set to train the model. Input the feature data of the enterprise to be classified into the SVM model to obtain the classification result.

[0012] Industrial map module: Use a graph database to store the results of enterprise spectrum classification and on-chain. Analyze the relationships between industries through graph algorithms to construct an industrial map, and use a visualization tool to generate an industrial map portrait.

[0013] Precision mining module: Train and optimize the model through machine learning algorithms, and use a comprehensive evaluation model to complete the establishment of a target enterprise value evaluation model.

[0014] Robot management module: Manage the function maintenance and upgrade of robots, send control instructions to robots, and realize the background management function of robots.

[0015] Operation monitoring module: Through built-in intelligent algorithms, automatically identify abnormal situations during the operation of robots, immediately trigger an early warning mechanism, and notify operation and maintenance personnel in a timely manner.

[0016] Furthermore, locate the elements in the page and extract the required data. For the situation where the page structure is complex and it is difficult to accurately locate data through XPath, intercept the target page image, use a deep learning model to identify the text and table information in the screenshot, and use AIGC to automatically perform data structure conversion to form structured data that meets the format requirements.

[0017] Further, the industrial data governance module establishes an industrial data governance system through metadata management, data standard management, and data quality management, completes the aggregation, cleaning, processing, and modeling of multi-source industrial data, and uses the cosine similarity enterprise name matching algorithm to automatically collect enterprise investment and financing, talent recruitment, administrative penalties, intellectual property, qualification certifications, and legal document data into the correct enterprise directory;

[0018] In the scenario of enterprise name matching, consider the enterprise name as a vector, with each character regarded as a dimension of the vector. According to the frequency or weight of the character's appearance, assign a value to each dimension to convert the enterprise name into a vector representation;

[0019] The formula for cosine similarity is:

[0020] Similarity(A,B)=(A·B) / (||A||×||B||)

[0021] Among them, A and B are two vectors, A·B is their dot product, ||A|| and ||B|| are their norms respectively. Convert the enterprise names into vectors A and B, then A·B is the sum of the products of the values on their corresponding dimensions, and ||A|| and ||B|| are the square roots of the sum of the squares of the values on their respective dimensions;

[0022] Use the cosine similarity formula to calculate the similarity degree between two enterprise names, set the threshold of similarity within a certain range, and determine that two enterprise names exceeding this threshold are the same enterprise, so as to automatically collect various behavioral data of the enterprise under the same enterprise directory.

[0023] Further, the enterprise panoramic portrait module uses data statistics, machine learning, and natural language processing technologies to deeply mine and analyze the collected enterprise information, directory information, qualification certifications, business information, negative information, and other information, identify patterns, trends, and associations in the data, and display the enterprise panoramic portrait through a graphical interface, including the visual presentation and interaction of various types of information.

[0024] Further, the enterprise on-chain module uses an artificial intelligence classification algorithm model to classify the leading industries of enterprises, constructs an SVM classification model with a Gaussian kernel function, and uses the enterprise data with known leading industries as the given training data set, {(x_1,y_1),(x_2,y_2),...,(x_m,y_m)}, where x_i is the input data and y_i is the corresponding leading industry classification. The optimization objective of the SVM model is:

[0025]

[0026] Among them, α i is the Lagrange multiplier, yi is a sample point, X i The corresponding label, K(x i , x j ), is the Gaussian kernel function, n is the size of the training set, and this optimization problem is restricted by the following constraints: Among them, C is the regularization parameter, which is used to control the balance between the generalization ability of the model and the training error. After solving the above optimization problem, the Lagrange multiplier α corresponding to each sample point is obtained i , for the new sample X to be classified, its class prediction is realized through the following decision function: Among them, b is the bias term, which is determined by the support vectors.

[0027] Furthermore, the industrial map module uses knowledge extraction technology to automatically identify company names, product names, and enterprise qualifications from industry reports, academic papers, news reports, social media, and enterprise databases, extracts various industrial entities and the relationships between entities, forms basic map data elements, organizes the extracted knowledge according to the data structure of the industrial map, forms a complete industrial map, and finally uses a graph database for the storage and display of the industrial map. The graph database used has the function of convenient retrieval based on entities and relationships.

[0028] Furthermore, the precise mining module uses a comprehensive evaluation model to establish a target enterprise value evaluation model according to the following steps:

[0029] S1: First, determine the alternative object set U = {U1, U2,..., Un}, that is, each enterprise; the index set X = {X1, X2,..., Xm}, including enterprise technological innovation, market share, operating conditions, and investment and financing status indicators; the comment set V = {v1, v2,..., vp}, including excellent, good, general, and poor;

[0030] S2: Determine the weight set of each index

[0031] S3: Establish an evaluation matrix R = (r ij ) m×p , where r ij represents the membership degree of the i-th index to the j-th comment;

[0032] S4: Calculate the fuzzy comprehensive evaluation set Here Adopt synthesis operators such as multiplication and bounded sum. When using the M(·, +) operator,

[0033] Furthermore, the robot management module performs function maintenance and upgrade management on the robot, sends control commands to the robot, and realizes the background management function of the robot.

[0034] Furthermore, the operation monitoring module automatically identifies abnormal situations during the operation of the robot through built-in intelligent algorithms, and immediately triggers an early warning mechanism to notify the operation and maintenance personnel in a timely manner via email, text message, and system notification. Description of the Drawings

[0035] Figure 1 It is a structural diagram of a cross-platform artificial intelligence precise investment promotion robot system proposed by the present invention;

[0036] Figure 2 It is a schematic flow chart of the application method of the cross-platform artificial intelligence precise investment promotion robot system proposed by the present invention. Detailed Embodiments

[0037] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0038] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention.

[0039] In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the described features. In the description of the present invention, "a plurality" means two or more, unless otherwise specifically defined. In addition, the terms "installation", "connection", and "connection" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the internal communication of two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations. The present invention will be further described in detail below with reference to the accompanying drawings.

[0040] Reference Figures 1 to 2 : A cross-platform AI-based precise investment promotion robot control system and method, the system includes the following modules:

[0041] The multi-source data collection module uses artificial intelligence, big data, knowledge graphs and other technologies to collect various types of industry data from multiple data channels such as government open data websites, corporate information disclosure platforms and professional databases to ensure the comprehensiveness and diversity of industry data. Multi-source data collection rules are set in a visual way, and data collection tasks are automatically completed regularly or in real time based on scheduled tasks or event trigger mechanisms.

[0042] The industrial data governance module uses data governance functional components such as metadata management, data standard management, and data quality management to establish a multi-level, multi-dimensional, and multi-topic industrial data governance system to complete the aggregation, cleaning, processing, and modeling of multi-source industrial data.

[0043] The enterprise panoramic portrait module extracts features from data that meets quality requirements and labels the enterprise with various data tags to quantitatively reflect the enterprise's characteristics in terms of industry, technology, capital, talent, etc.

[0044] The enterprise chain module uses an artificial intelligence classification algorithm model to classify the dominant industries of enterprises, selects appropriate linear kernels and Gaussian kernel functions to build an SVM classification model, uses enterprise data of known dominant industries as a training set to train the model, and inputs the characteristic data of the enterprises to be classified into the trained SVM model to obtain the classification results, complete the enterprise classification and chaining, and establish a complete industry map portrait.

[0045] The industry map module uses a graph database to store the results of enterprise classification and chain-linking, analyzes the relationship between industries through graph algorithms, constructs an industry map, and uses visualization tools to generate an industry map portrait to display the distribution and correlation of each industry.

[0046] The precise mining module trains and optimizes the model through machine learning algorithms, and uses a comprehensive evaluation model to complete the establishment of a target enterprise value assessment model, comprehensively considering multiple dimensions such as the enterprise's financial status, market competitiveness, technological innovation capabilities, and legal compliance to improve the accuracy of risk and value assessment.

[0047] The robot management module maintains and upgrades the robot functions, sends control instructions to the robot, and realizes the robot's background management functions.

[0048] The operation monitoring module, through built-in intelligent algorithms, automatically identifies abnormal situations during the operation of the robot, such as data loss, format mismatch, performance bottlenecks, etc., and immediately triggers an early warning mechanism, notifying the operation and maintenance personnel in a timely manner through various methods such as emails, text messages, and system notifications to ensure that problems are quickly responded to and processed.

[0049] In the present invention, the multi-source data acquisition module has the ability to visually configure acquisition rules, supporting two acquisition methods: parsing the page structure based on XPath and identifying page image content using a deep learning model. By writing XPath expressions, specific elements on the page can be accurately located, thus extracting the required data. For cases where the page structure is complex and it is difficult to accurately locate data through XPath, by capturing the target page image, using a deep learning model to identify information such as text and tables in the screenshot, and automatically performing data structure conversion using AIGC to form structured data that meets the format requirements.

[0050] In the present invention, the industrial data governance module, with the help of data governance functional components such as metadata management, data standard management, and data quality management, establishes a multi-level, multi-dimensional, and multi-topic industrial data governance system to complete the aggregation, cleaning, processing, and modeling of multi-source industrial data. Using the cosine similarity enterprise name matching algorithm, data such as enterprise investment and financing, talent recruitment, administrative penalties, intellectual property rights, qualification certifications, and legal documents are automatically aggregated under the correct enterprise directory, realizing the alignment and integration of information from different sources.

[0051] In the scenario of enterprise name matching, consider the enterprise name as a vector, and each character (or character combination) as a dimension of the vector. According to the frequency or weight of the character appearance, assign a value to each dimension to convert the enterprise name into a vector representation.

[0052] The formula for cosine similarity is:

[0053] Similarity(A,B)=(A·B) / (||A||×||B||)

[0054] Where A and B are two vectors, A·B is their dot product, and ||A|| and ||B|| are their norms (or lengths) respectively.

[0055] Convert the enterprise names into vectors A and B, then A·B is the sum of the products of the values on their corresponding dimensions, and ||A|| and ||B|| are the square roots of the sum of the squares of the values on their respective dimensions.

[0056] The cosine similarity formula is used to calculate the similarity degree between two enterprise names. The threshold of similarity is set within a reasonable range. Two enterprise names exceeding this threshold are determined to be the same enterprise, so as to automatically collect various behavioral data of the enterprise under the same enterprise directory.

[0057] In the present invention, the enterprise panoramic portrait module utilizes technologies such as data statistics, machine learning, and natural language processing (NLP) to deeply mine and analyze the collected enterprise information, directory information, qualification certifications, business information, negative information, and other information, identify patterns, trends, and associations in the data, and display the enterprise panoramic portrait using a graphical interface, including the visual presentation and interaction of various types of information.

[0058] In the present invention, the enterprise on-chain module uses an artificial intelligence classification algorithm model to classify the leading industries of enterprises, selects an appropriate Gaussian kernel function to construct an SVM classification model, and uses the enterprise data with known leading industries as the given training data set, {(x_1,y_1),(x_2,y_2),...,(x_m,y_m)}, where x_i is the input data and y_i is the corresponding leading industry classification. The optimization objective of the SVM model is:

[0059]

[0060] Among them, α i is the Lagrange multiplier, y i is the label corresponding to the sample point X i K(x i , x j ) is the Gaussian kernel function, and n is the size of the training set. This optimization problem is restricted by the following constraint conditions:

[0061] 0 ≤ α i ≤ C, i = 1,..., n

[0062]

[0063] Here, C is the regularization parameter, which is used to control the balance between the generalization ability of the model and the training error.

[0064] After solving the above optimization problem, the Lagrange multiplier α i corresponding to each sample point can be obtained. For the new sample X to be classified, its class prediction can be achieved through the following decision function:

[0065]

[0066] Among them, b is the bias term, which can be determined by the support vectors (i.e., the sample points where α i > 0).

[0067] In the present invention, the industrial map module uses knowledge extraction technology to automatically identify entities such as company names, product names, and enterprise qualifications from structured and unstructured data such as industry reports, academic papers, news reports, social media, and enterprise databases, extracts various industrial entities and the relationships between entities to form basic map data elements, then organizes the extracted knowledge according to the data structure of the industrial map to form a complete industrial map, and finally uses a graph database for the storage and display of the industrial map. The graph database used has the function of convenient retrieval based on entities and relationships.

[0068] In the present invention, the precise mining module uses a comprehensive evaluation model to establish a target enterprise value evaluation model according to the following steps:

[0069] S1: First, determine the set of alternative objects U = {U1, U2,..., Un}, that is, each enterprise; the index set X = {X1, X2,..., Xm}, including indicators such as enterprise technological innovation, market share, operating conditions, and investment and financing conditions; the evaluation set V = {v1, v2,..., vp}, including excellent, good, general, poor, etc.

[0070] S2: Then determine the weight set of each index

[0071] S3: Establish an evaluation matrix R = (r ij ) m×p , where r ij represents the membership degree of the i-th index to the j-th evaluation.

[0072] S4: Finally, calculate the fuzzy comprehensive evaluation set Here uses synthesis operators such as multiplication and bounded sum, such as the M(·, +) operator,

[0073] The precise mining module comprehensively considers multiple dimensions such as the financial status, market competitiveness, technological innovation ability, and legal compliance of the enterprise to accurately screen out high-quality enterprises.

[0074] In the present invention, the robot management module performs functional maintenance and upgrade management on the robot, sends control commands to the robot, and realizes the background management function of the robot.

[0075] In the present invention, the operation monitoring module automatically identifies abnormal situations during the operation of the robot, such as data loss, format mismatch, performance bottleneck, etc., through built-in intelligent algorithms, and immediately triggers an early warning mechanism to notify the operation and maintenance personnel in a timely manner through multiple methods such as emails, text messages, and system notifications to ensure that problems are quickly responded to and processed.

[0076] The technical solution of the present invention will be clearly and completely described below in conjunction with embodiments. The described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments, and are not used to limit the invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work belong to the scope of protection of the present invention.

[0077] The following gives an optimal embodiment:

[0078] S1. Complete the industrial data collection in two ways: parsing the page structure based on XPath and using a deep learning model to identify the page image content. By writing XPath expressions, specific elements in the page can be accurately located, so as to extract the required data. For the case where the page structure is complex and it is difficult to accurately locate data through XPath, by intercepting the target page image, using a deep learning model to identify information such as text and tables in the screenshot, and automatically performing data structure conversion by AIGC to form structured data that meets the format requirements.

[0079] S2. Build a multi-level, multi-dimensional, and multi-topic industrial data governance system with the help of data governance functional components such as metadata management, data standard management, and data quality management, and complete the aggregation, cleaning, processing, and modeling of multi-source industrial data. Using the cosine similarity enterprise name matching algorithm, automatically collect data such as enterprise investment and financing, talent recruitment, administrative penalties, intellectual property rights, qualification certifications, and legal documents under the correct enterprise directory, and achieve the alignment and integration of information from different sources.

[0080] In the scenario of enterprise name matching, regard the enterprise name as a vector, and regard each character (or character combination) as a dimension of the vector. According to the frequency or weight of the character appearance, assign a value to each dimension, and convert the enterprise name into a vector representation.

[0081] The formula for cosine similarity is:

[0082] Similarity(A,B)=(A·B) / (||A||×||B||)

[0083] Among them, A and B are two vectors, A·B is their dot product, and ||A|| and ||B|| are their norms (or lengths) respectively.

[0084] Convert the enterprise names into vectors A and B, then A·B is the sum of the products of the values on their corresponding dimensions, and ||A|| and ||B|| are the square roots of the sum of the squares of the values on their respective dimensions.

[0085] Use the cosine similarity formula to calculate the similarity between two enterprise names, set the similarity threshold within a reasonable range, and determine that two enterprise names exceeding this threshold belong to the same enterprise, so as to automatically collect various behavioral data of enterprises under the same enterprise list.

[0086] S3. Use the artificial intelligence classification algorithm model to classify the leading industries of enterprises, select an appropriate Gaussian kernel function to construct an SVM classification model, and use the enterprise data with known leading industries as the given training data set, \(\{(x_1,y_1),(x_2,y_2),...,(x_m,y_m)\}\), where \(x_i\) is the input data and \(y_i\) is the corresponding leading industry classification. The optimization objective of the SVM model is:

[0087]

[0088] where, \(\alpha\) i is the Lagrange multiplier, \(y\) i is the label corresponding to the sample point \(X\) i , \(K(x\) i , \(x\) j ) is the Gaussian kernel function, and \(n\) is the size of the training set. This optimization problem is restricted by the following constraints:

[0089] \(0\leq\alpha\) i \(\leq C, i = 1,...,n\)

[0090]

[0091] Here, \(C\) is the regularization parameter, which is used to control the balance between the generalization ability of the model and the training error.

[0092] S4. After solving the above optimization problem, the Lagrange multiplier \(\alpha\) i corresponding to each sample point can be obtained. For the new sample \(X\) to be classified, its class prediction can be achieved through the following decision function:

[0093]

[0094] where, \(b\) is the bias term, which can be determined by the support vectors (i.e., the sample points where \(\alpha\) i \(> 0\)).

[0095] S5. Use knowledge extraction technology to automatically identify entities such as company names, product names, and enterprise qualifications from structured and unstructured data such as industry reports, academic papers, news reports, social media, and enterprise databases, extract various industrial entities and the relationships between entities to form basic graph data elements, and then organize the extracted knowledge according to the data structure of the industrial graph to form a complete industrial graph. Finally, use a graph database to store and display the industrial graph, and the graph database used has the function of convenient retrieval based on entities and relationships.

[0096] S6. Train and optimize the model through machine learning algorithms, and use a comprehensive evaluation model to complete the establishment of the target enterprise value evaluation model, comprehensively considering multiple dimensions such as the financial status, market competitiveness, technological innovation ability, and legal compliance of the enterprise to improve the accuracy of risk and value evaluation.

[0097] S7: First, determine the alternative object set U = {U1, U2,..., Un}, that is, each enterprise; the index set X = {X1, X2,..., Xm}, including indicators such as enterprise technological innovation, market share, operating conditions, and investment and financing conditions; the comment set V = {v1, v2,..., vp}, including excellent, good, general, poor, etc.

[0098] S8: Then determine the weight set of each index

[0099] S9: Establish an evaluation matrix R = (r ij ) m×p , where r ij represents the membership degree of the i-th index to the j-th comment.

[0100] S10: Finally, calculate the fuzzy comprehensive evaluation set Here When using synthesis operators such as multiplication and bounded sum, such as the M(·, +) operator,

[0101] The above is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A cross-platform AI precise investment promotion robot control system, characterized in that include: Multi-source data collection module: supports two collection methods: parsing page structure based on XPath and identifying page image content using deep learning models. It sets multi-source data collection rules and automatically completes data collection tasks in real time based on scheduled tasks or event trigger mechanisms. Industrial data governance module: Establish an industrial data governance system through metadata management, data standard management and data quality management data governance functions to complete the aggregation, cleaning, processing and modeling of multi-source industrial data; Enterprise Panoramic Portrait Module: Extract features from data that meet quality requirements, label the enterprise with various data tags, and quantitatively reflect the characteristics of the enterprise in terms of industry, technology, capital, and talent; Enterprise on-chain module: Use artificial intelligence classification algorithm model to classify enterprises by their dominant industries, select linear kernel and Gaussian kernel function to build SVM classification model, use enterprise data of known dominant industries as training set to train the model, input feature data of enterprises to be classified into SVM model to obtain classification results; Industry map module: Use the graph database to store the results of enterprise classification and chaining, analyze the relationship between industries through graph algorithms, build industry maps, and use visualization tools to generate industry map portraits; Precision mining module: train and optimize the model through machine learning algorithms, and use comprehensive evaluation models to complete the establishment of the target enterprise value assessment model; Robot management module: maintains and upgrades the robot functions, sends control instructions to the robot, and implements the robot's background management functions; Operation monitoring module: Through the built-in intelligent algorithm, it automatically identifies abnormal situations during the operation of the robot, immediately triggers the early warning mechanism, and immediately notifies the operation and maintenance personnel.

2. The control system of the cross-platform artificial intelligence precise investment promotion robot according to claim 1, characterized in that, Locate the elements in the page and extract the required data. For situations where the page structure is complex and it is difficult to accurately locate the data through XPath, capture the target page image, use the deep learning model to recognize the text and table information in the screenshot, and use AIGC to automatically convert the data structure to form structured data that meets the format requirements.

3. The cross-platform artificial intelligence-based precise investment promotion robot control system according to claim 1, wherein The industrial data governance module establishes an industrial data governance system through metadata management, data standard management, and data quality management, completes the aggregation, cleaning, processing, and modeling of multi-source industrial data, and uses the cosine similarity enterprise name matching algorithm to automatically aggregate enterprise investment and financing, talent recruitment, administrative penalties, intellectual property rights, qualification certification, and legal document data into the correct enterprise directory; In the scenario of enterprise name matching, the enterprise name is regarded as a vector, and each character is regarded as a dimension of the vector. A value is assigned to each dimension according to the frequency or weight of the character, and the enterprise name is converted into a vector representation. The formula for cosine similarity is: Similarity(A,B)=(A·B) / (||A||×||B||) Among them, A and B are two vectors, A·B is their dot product, ||A|| and ||B|| are their moduli respectively. When converting enterprise names into vectors A and B, then A·B is the sum of the products of the values on their corresponding dimensions, and ||A|| and ||B|| are the square roots of the sum of the squares of the values on their respective dimensions; The cosine similarity formula is used to calculate the similarity degree between two enterprise names. The threshold of the similarity degree is set within a certain range. Two enterprise names exceeding this threshold are determined to be the same enterprise, so as to automatically collect various behavior data of the enterprise under the same enterprise directory.

4. The cross-platform artificial intelligence-based precise investment promotion robot control system according to claim 1, characterized in that The enterprise panoramic portrait module uses data statistics, machine learning, and natural language processing technologies to deeply mine and analyze the collected enterprise information, directory information, qualification certifications, business information, negative information, and other information, identify patterns, trends, and associations in the data, and uses a graphical interface to display the enterprise panoramic portrait, including the visual presentation and interaction of various types of information.

5. The cross-platform artificial intelligence-based precise investment promotion robot control system according to claim 1, characterized in that The enterprise on-chain module uses an artificial intelligence classification algorithm model to classify the leading industries of enterprises. A Gaussian kernel function is used to construct an SVM classification model. The enterprise data with known leading industries is used as the given training data set, {(x_1,y_1),(x_2,y_2),...,(x_m,y_m)}, where x_i is the input data and y_i is the corresponding leading industry classification. The optimization objective of the SVM model is: Among them, α i is the Lagrange multiplier, y i is the sample point, x i is the corresponding label, K(x i , x j ) is the Gaussian kernel function, n is the size of the training set, and this optimization problem is restricted by the following constraint conditions: Among them, C is the regularization parameter, which is used to control the balance between the generalization ability of the model and the training error. After solving the above optimization problem, the Lagrange multiplier α i corresponding to each sample point is obtained. For the new sample X to be classified, its class prediction is realized through the following decision function: Among them, b is the bias term, which is determined by the support vectors.

6. The cross-platform artificial intelligence-based precise investment promotion robot control system according to claim 1, characterized in that, The industrial map module uses knowledge extraction technology to automatically identify company names, product names, and enterprise qualifications from industry reports, academic papers, news reports, social media, and enterprise databases, extracts various industrial entities and the relationships between entities to form basic map data elements, organizes the extracted knowledge according to the map structure of the industrial map to form a complete industrial map, and finally uses a graph database for the storage and display of the industrial map. The graph database used has the function of convenient retrieval based on entities and relationships.

7. A method for applying the cross-platform artificial intelligence precise investment promotion robot control system according to any one of claims 1-6, characterized in that, The precise mining module uses a comprehensive evaluation model to establish a target enterprise value evaluation model according to the following steps: S1: First, determine the alternative object set U = {U1, U2,..., Un}, that is, each enterprise; the index set X = {X1, X2,..., Xm}, including enterprise technological innovation, market share, business conditions, and investment and financing status indicators; the comment set V = {v1, v2,..., vp}, including excellent, good, general, and poor; S2: Determine the weight set of each index S3: Establish the evaluation matrix R = (r ij ) m×p , where r ij represents the membership degree of the i-th index to the j-th comment; S4: Calculate the fuzzy comprehensive evaluation set Here When using the composition operator, the M(·, +) operator 8. The control system of the cross-platform artificial intelligence precise investment promotion robot according to claim 1, characterized in that, The robot management module conducts functional maintenance and upgrade management of the robot, sends control commands to the robot, and realizes the background management function of the robot.

9. The cross-platform artificial intelligence-based precision investment promotion robot control system according to claim 1, characterized in that, The operation monitoring module automatically identifies abnormal situations during the operation of the robot through built-in intelligent algorithms, and immediately triggers an early warning mechanism, and instantly notifies the operation and maintenance personnel through email, text message, and system notification methods.

Images