Enterprise potential intelligent identification method and system

Abstract

The invention discloses an enterprise potential intelligent identification method and system, and belongs to the technical field of enterprise intelligent analysis. The method comprises the following steps: constructing an enterprise behavior feature vector set by collecting an enterprise financial statement, behavior data, a semantic text and a visual map in a preset time sequence window, and carrying out topic modeling on an enterprise strategy text to obtain a strategy topic vector; and modeling enterprise behavior log data, extracting behavior topic vectors, and calculating a strategic offset metric value in combination with strategic response time-delay data. And calculating a strategic consistency score through a consistency score formula, and inputting the data into the enterprise potential generation network in combination with the enterprise technical entropy and the market entropy to obtain an enterprise potential score sequence. According to the scheme, monitoring of the matching degree of strategic execution and actual behaviors is enhanced, and prospective and dynamic evaluation is provided for enterprises. The method helps to identify early signals of potential outbreak, capture changes in a transformation period and a decline period, optimize resource allocation and realize more scientific strategic adjustment.

Description

Technical Field

[0001] This application belongs to the field of enterprise intelligent analysis technology, and specifically relates to an enterprise potential intelligent recognition method and system. Background Art

[0002] In the context of global competition and digital transformation, enterprise potential assessment has become the core basis for investment decisions, strategic planning, and resource allocation. Traditional assessment methods mainly rely on financial indicators (such as revenue growth rate, profit margin) or expert experience, and it is difficult to comprehensively capture the dynamic competitiveness of enterprises, especially the hidden potential such as strategic execution ability, technological innovation efficiency, and market adaptability. With the development of big data and artificial intelligence technologies, there is an urgent need for an intelligent assessment method that can integrate multi-source heterogeneous data, quantify the consistency between strategy and behavior, and dynamically predict the growth potential of enterprises.

[0003] Current enterprise potential assessment technologies mainly rely on single-dimensional data or static analysis frameworks. For example, enterprise competitiveness is evaluated through financial indicators (such as revenue, profit margin) or market research data. Although some methods introduce unstructured data (such as public opinion texts, patent information), they only stay at the level of simple keyword statistics or manual experience judgment.

[0004] However, the time series analysis capabilities of existing technologies are generally weak. They can only compare historical cross-sectional data and lack the ability to perform time series modeling on dynamic behavior patterns (such as R & D investment fluctuations, supply chain adjustment frequencies). In addition, existing technologies have not established a quantitative association mechanism between strategic texts and actual enterprise behaviors, and cannot capture the dynamic coupling relationship between strategic intentions and execution results, resulting in the evaluation results lagging behind the evolution of the true capabilities of enterprises. Summary of the Invention

[0005] The embodiments of this application provide an enterprise potential intelligent recognition method and system, which solve the problems that the time series analysis capabilities of existing technologies are generally weak, they can only compare historical cross-sectional data, and lack the ability to perform time series modeling on dynamic behavior patterns (such as R & D investment fluctuations, supply chain adjustment frequencies). In addition, existing technologies have not established a quantitative association mechanism between strategic texts and actual enterprise behaviors, and cannot capture the dynamic coupling relationship between strategic intentions and execution results, resulting in the evaluation results lagging behind the evolution of the true capabilities of enterprises.

[0006] In a first aspect, the embodiments of this application provide an enterprise potential intelligent recognition method, and the method includes: Obtain the structured data of enterprise financial statements, the time series data of enterprise behaviors, the semantic text data of enterprises, and the visual graph data of enterprises within a preset sampling time series window, and construct an enterprise behavior time series feature vector set according to the structured data of enterprise financial statements, the time series data of enterprise behaviors, the semantic text data of enterprises, and the visual graph data of enterprises; Obtain enterprise strategic texts within a preset sampling time series window, perform topic modeling and vectorization processing on the enterprise strategic texts to obtain enterprise strategic topic vectors at each time point; Obtain enterprise behavior log data within a preset sampling time series window, perform structured modeling and behavior pattern extraction on the enterprise behavior log data to obtain enterprise behavior topic vectors at each time point; Extract strategic response time lag data at each time point from enterprise strategic texts and enterprise behavior log data according to the time series annotation mechanism; Analyze enterprise strategic texts according to the behavior topic deviation analysis method to obtain enterprise strategic topic vectors at each time point, and analyze enterprise behavior log data according to the behavior topic deviation analysis method to obtain enterprise behavior data topic vectors at each time point. Calculate the cosine similarity between the enterprise strategic topic vector and the enterprise behavior data topic vector, and determine the strategic deviation metric value according to the cosine similarity; Calculate the strategic consistency score according to the preset sampling time series window, enterprise strategic topic vector, enterprise behavior topic vector, strategic response time lag data, strategic deviation metric value, and preset consistency scoring calculation formula; Obtain enterprise technical entropy and enterprise market entropy, and input the enterprise behavior time series feature vector set, enterprise strategic topic vector, enterprise behavior topic vector, strategic response time lag data, strategic deviation metric value, strategic consistency score, enterprise technical entropy, and enterprise market entropy into a preset enterprise potential generation network to generate an enterprise potential score sequence.

[0007] In a second aspect, an embodiment of the present application provides an enterprise potential intelligent recognition system, and the system includes: A behavior time series feature vector set construction module, configured to obtain enterprise financial statement structured data, enterprise behavior time series data, enterprise semantic text data, and enterprise visual map data within a preset sampling time series window, and construct an enterprise behavior time series feature vector set according to the enterprise financial statement structured data, enterprise behavior time series data, enterprise semantic text data, and enterprise visual map data; An enterprise strategic topic vector determination module, configured to obtain enterprise strategic texts within a preset sampling time series window, perform topic modeling and vectorization processing on the enterprise strategic texts to obtain enterprise strategic topic vectors at each time point; An enterprise behavior topic vector determination module, configured to obtain enterprise behavior log data within a preset sampling time series window, perform structured modeling and behavior pattern extraction on the enterprise behavior log data to obtain enterprise behavior topic vectors at each time point; A strategic response time lag data determination module, configured to extract strategic response time lag data at each time point from enterprise strategic texts and enterprise behavior log data according to the time series annotation mechanism; A strategic deviation metric determination module, configured to analyze an enterprise's strategic text according to a behavioral theme deviation analysis method to obtain the enterprise's strategic theme vectors at each time point, and analyze the enterprise's behavioral log data according to the behavioral theme deviation analysis method to obtain the enterprise's behavioral data theme vectors at each time point, calculate the cosine similarity between the enterprise's strategic theme vectors and the enterprise's behavioral data theme vectors, and determine a strategic deviation metric based on the cosine similarity; A strategic consistency score calculation module, configured to calculate a strategic consistency score according to a preset sampling time series window, the enterprise's strategic theme vectors, the enterprise's behavioral theme vectors, strategic response lag data, the strategic deviation metric, and a preset consistency score calculation formula; An enterprise potential score sequence generation module, configured to obtain an enterprise's technology entropy and an enterprise's market entropy, and input an enterprise's behavioral time series feature vector set, the enterprise's strategic theme vectors, the enterprise's behavioral theme vectors, strategic response lag data, the strategic deviation metric, the strategic consistency score, the enterprise's technology entropy, and the enterprise's market entropy into a preset enterprise potential generation network to generate an enterprise potential score sequence.

[0008] In a third aspect, an embodiment of the present application provides an electronic device, which includes a processor, a memory, and a program or instruction stored on the memory and executable on the processor. When the program or instruction is executed by the processor, the steps of the method described in the first aspect are implemented.

[0009] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0010] In the embodiments of the present application, structured data of enterprise financial statements, enterprise behavior time-series data, enterprise semantic text data, and enterprise visual graph data within a preset sampling time-series window are obtained, and an enterprise behavior time-series feature vector set is constructed based on the structured data of enterprise financial statements, enterprise behavior time-series data, enterprise semantic text data, and enterprise visual graph data; enterprise strategic texts within a preset sampling time-series window are obtained, and topic modeling and vectorization processing are performed on the enterprise strategic texts to obtain enterprise strategic topic vectors at each time point; enterprise behavior log data within a preset sampling time-series window are obtained, and structured modeling and behavior pattern extraction are performed on the enterprise behavior log data to obtain enterprise behavior topic vectors at each time point; strategic response time-delay data at each time point are extracted from the enterprise strategic texts and enterprise behavior log data according to a time-series annotation mechanism; the enterprise strategic texts are analyzed according to a behavior topic deviation analysis method to obtain enterprise strategic topic vectors at each time point, and the enterprise behavior log data are analyzed according to the behavior topic deviation analysis method to obtain enterprise behavior data topic vectors at each time point, the cosine similarity between the enterprise strategic topic vectors and the enterprise behavior data topic vectors is calculated, and a strategic deviation metric value is determined according to the cosine similarity; a strategic consistency score is calculated according to a preset sampling time-series window, enterprise strategic topic vectors, enterprise behavior topic vectors, strategic response time-delay data, strategic deviation metric values, and a preset consistency score calculation formula; enterprise technical entropy and enterprise market entropy are obtained, and the enterprise behavior time-series feature vector set, enterprise strategic topic vectors, enterprise behavior topic vectors, strategic response time-delay data, strategic deviation metric values, strategic consistency scores, enterprise technical entropy, and enterprise market entropy are input into a preset enterprise potential generation network to generate an enterprise potential score sequence. Through the above enterprise potential intelligent recognition method, by integrating multi-dimensional information, the development potential of an enterprise can be comprehensively and accurately evaluated. It not only strengthens the monitoring of the matching degree between strategic implementation and actual behavior, but also identifies the potential change trend through time-series analysis, providing a more forward-looking and dynamic evaluation for the enterprise. In addition, this method provides accurate decision-making support for enterprise management, helps identify early signals of potential outbreaks, captures changes during the transformation period and recession period, thereby optimizing resource allocation and achieving more scientific strategic adjustments. Description of the Drawings

[0011] Figure 1 is a schematic flowchart of the enterprise potential intelligent recognition method provided in Embodiment 1 of the present application; Figure 2 is a schematic flowchart of the enterprise potential intelligent recognition method provided in Embodiment 2 of the present application; Figure 3 is a schematic structural diagram of the enterprise potential intelligent recognition system provided in Embodiment 3 of the present application; Figure 4It is a schematic structural diagram of the electronic device provided in Embodiment 4 of the present application. Detailed implementation manners

[0012] In order to make the objectives, technical solutions, and advantages of the present application clearer, the following further describes the specific embodiments of the present application in detail with reference to the accompanying drawings. It can be understood that the specific embodiments described herein are only used to explain the present application, rather than limiting the present application. Additionally, it should be noted that for the sake of convenience of description, only the parts related to the present application are shown in the drawings, rather than all the content. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. When the operations are completed, the process can be terminated, but there can also be additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0013] The following will clearly describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art belong to the scope of protection of the present application.

[0014] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, rather than to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are usually of the same type, and the number of objects is not limited. For example, the first object can be one or multiple. In addition, "and / or" in the specification and claims means at least one of the connected objects, and the character " / " generally indicates an "or" relationship between the associated objects before and after.

[0015] The following describes in detail a RSMC chip, a multi-stage chip startup method, and a Beidou communication navigation device provided in the embodiments of the present application with reference to the accompanying drawings through specific embodiments and their application scenarios.

[0016] Embodiment 1 Figure 1 It is a schematic flowchart of the enterprise potential intelligent recognition method provided in Embodiment 1 of the present application. As Figure 1 shown, it specifically includes the following steps: S101. Obtain the structured data of the enterprise financial statements, the time-series data of enterprise behaviors, the semantic text data of the enterprise, and the visual graph data of the enterprise within a preset sampling time-series window, and construct an enterprise behavior time-series feature vector set based on the structured data of the enterprise financial statements, the time-series data of enterprise behaviors, the semantic text data of the enterprise, and the visual graph data of the enterprise.

[0017] The preset sampling time-series window can refer to a continuous time period selected on the time axis for constructing and analyzing the enterprise potential model.

[0018] The structured data of the enterprise financial statements can refer to tabular data that is standardized and can be directly used for modeling from the enterprise financial statements. For example, it can include indicators such as revenue, profit, cash flow, liabilities, assets, etc.; various financial ratios (such as asset-liability ratio, gross profit margin, net profit margin, etc.).

[0019] The time-series data of enterprise behaviors can be data that reflects various behavioral activities of the enterprise during operation and is recorded in chronological order. For example, it can include investment decision records (such as mergers and acquisitions, new product releases); major management actions (such as senior management changes, organizational adjustments); cooperation / termination events, market activity records; user operation data (such as the number of active users, transaction frequency).

[0020] The semantic text data of the enterprise can be text materials released by the enterprise that contain content such as strategies, intentions, value propositions, etc.

[0021] The visual graph data of the enterprise can be the enterprise operation, product, or industrial chain information presented in the form of images and transformed into available vectors through visual processing. It can include product pictures (obtained from the official website, mall); entity layout diagrams (such as factory layouts, retail terminals); industry maps (showing upstream and downstream associations).

[0022] The enterprise behavior time-series feature vector set can be a vector sequence constructed in chronological order after integrating the above-mentioned multi-modal data (structured data, text, images, time-series behaviors).

[0023] Within a preset sampling time series window (e.g., a continuous time period with a granularity of quarters or months, such as the last 8 quarters), first obtain the structured financial statement data of the enterprise within the corresponding time period from the enterprise public data source and the third-party commercial data interface. Specifically, it includes the key financial indicators of the balance sheet, income statement, and cash flow statement (such as operating income, net profit, debt ratio, net cash flow, etc.), and perform missing value processing, normalization, and time alignment operations on the data to generate a structured financial indicator vector in time series format. Secondly, obtain the behavioral time series data of the enterprise within this time period. The behavioral time series data comes from enterprise announcements, government-disclosed data, industry news, enterprise official website dynamics, etc. Use an event extraction algorithm (such as a named entity recognition and event classification model based on BERT) to label the event type and extract the timestamp for each piece of behavioral data, encode the behavioral events into structured event vectors, and construct a behavioral event sequence sorted by event time. At the same time, for the semantic text data generated by the enterprise within this time series window (such as enterprise strategy texts, strategic parts of annual reports, media interview transcripts, etc.), use a pre-trained language model (such as BERT, RoBERTa) to clean, split sentences, and vectorize the text, extract the semantic embedding representation of each text segment, and generate a semantic time vector sequence aligned by time. For the visual graph data related to the enterprise (such as enterprise official website product pictures, industrial layout pictures, store pictures, factory pictures, etc.), construct an image data set through an image acquisition module, and use a pre-trained convolutional neural network model (such as ResNet, EfficientNet) to extract image feature vectors, perform time annotation on the images or classify them into the corresponding time periods, and construct an image feature time series according to time. Finally, based on the above four types of data sources, perform data alignment and fusion processing according to each sampling moment t, splice or perform multi-modal feature fusion (such as using Transformer or AutoEncoder fusion strategies) on the financial structured indicator vector, behavioral event vector, semantic text vector, and image feature vector at the same time point, and construct a multi-modal behavioral feature vector of the enterprise at time point t Arrange the vectors at all time points in sequence to construct an enterprise behavioral time series feature vector set X = { , ,..., }, which serves as the input basis for the subsequent enterprise strategic consistency scoring and potential prediction models.

[0024] S102. Obtain the enterprise strategic text within the preset sampling time series window, perform topic modeling and vectorization processing on the enterprise strategic text to obtain the enterprise strategic theme vectors at each time point.

[0025] Corporate strategic text may refer to natural language text data that reflects the company's medium- and long-term development direction, goals, key investment areas, market expansion strategies, technology development paths, governance concepts, etc.

[0026] A time point may refer to an identifier of a specific moment or time period within a sampling time window, and is used to indicate the time attribution of an enterprise strategy text.

[0027] The enterprise strategy theme vector may refer to a numerical vector formed by modeling and vectorizing the enterprise strategy text through natural language processing and topic modeling algorithms, and is used to represent the core theme of the enterprise strategy at a certain point in time.

[0028] Within the preset sampling time window, the text data containing strategic content released by the enterprise can be automatically captured and downloaded through channels such as enterprise information disclosure platforms, annual report announcement systems, enterprise official websites, and third-party financial databases, covering semantic paragraphs such as strategic outlook, development plans, business plans, and speeches by senior executives. Then, the text is cleaned and preprocessed using natural language processing technology, including sentence segmentation, word segmentation, stop word removal, word form restoration, named entity recognition, and other steps, and the release date or timestamp of the text is extracted to map to the corresponding time point. Subsequently, topic modeling algorithms (such as LDA, BERTopic, Top2Vec, etc.) are used to identify the topics of the processed text and to mine the potential topic structure that reflects the strategic focus of the enterprise. On this basis, the text is vectorized using bag-of-words models, TF-IDF, Doc2Vec, or pre-trained language models (such as BERT, RoBERTa), and each strategic text is represented as a vector representation in the topic space. Finally, the topic vectors of multiple strategic texts belonging to the same time point are weighted averaged or aggregated to generate the corporate strategic topic vector corresponding to that time point.

[0029] S103, obtaining the enterprise behavior log data within a preset sampling time window, performing structured modeling and behavior pattern extraction on the enterprise behavior log data, and obtaining the enterprise behavior theme vector at each time point.

[0030] Enterprise behavior log data can be a collection of data that can record the behavior of an enterprise during its actual operation within a specific time period. These behaviors can be obtained from system logs, business operation records, product release records, organizational adjustment announcements, recruitment information, supply chain activities, patent applications, investment and mergers and acquisitions, market activities, etc. This type of data is time-series and action-oriented, and is usually recorded in the form of "behavior occurrence time + behavior type + behavior object + related indicators", reflecting the specific actions of the enterprise in operation, management, technology, market, etc.

[0031] The enterprise behavior theme vector can be a multi-dimensional feature representation formed by modeling and vectorizing the above-mentioned behavior log data at a certain point in time, which represents the main activity patterns and behavior characteristics of the enterprise behavior within that time point.

[0032] Within the preset sampling time series window, first obtain enterprise behavior log data from multiple data sources, including but not limited to enterprise official websites, news platforms, industrial and commercial registration systems, social media, patent databases, recruitment platforms, and supply chain systems, and extract business actions that occurred within this period for the enterprise, such as product release, patent application, senior management changes, investment acquisitions, organizational restructuring, etc. After obtaining the enterprise behavior log data within the preset sampling time series window, when performing structured modeling on the obtained enterprise behavior log data, first sort the original log text in chronological order and unify the time format, enterprise name, and representation of behavior terms. Then, use a named entity recognition (NER) model to identify the key elements in the log, including the behavior subject (enterprise or department), behavior actions (such as "investment", "cooperation", "merger and acquisition", "launch new products"), behavior objects (such as project names, cooperation partners, products, etc.), and the occurrence time, to construct a four-tuple of behavior events (subject, action, object, time). Then, based on the event four-tuple, perform standardization processing according to the semantic similarity criterion. For example, classify "invest in a project" and "take a stake in a company" into the "capital investment category" behavior theme. In the behavior pattern extraction stage, use unsupervised clustering (such as K-Means, DBSCAN) or topic modeling techniques (such as LDA, BERTopic) to perform semantic clustering and classification on a large number of structured behavior events, and extract behavior pattern themes representing the main strategic directions of the enterprise, such as "technology R & D-driven", "capital expansion type", "international operation type", etc. Finally, the log data at each time point is converted into a multi-dimensional behavior theme distribution vector, reflecting the activity of the enterprise in various behavior patterns at that time point.

[0033] S104, Extract the strategic response time lag data for each time point from the enterprise strategic text and enterprise behavior log data according to the time series annotation mechanism.

[0034] The time series annotation mechanism can be a method mechanism for automatically identifying and annotating the time points of key events in time series data and establishing the temporal relationships between events. It can include strategic time anchoring: extracting the time points of strategic releases or updates from strategic texts through natural language processing (such as time expression recognition) to form a strategic event sequence. Behavioral response window setting: For each strategic event, setting a behavioral response observation window (such as 1 month, 1 quarter, or a dynamically adjustable interval) after it. Response behavior recognition: In this window, analyzing the enterprise behavior logs to identify whether there are behaviors semantically related to the strategic event. Using semantic matching or vector similarity matching methods (such as BERT similarity, cosine similarity) to determine whether the "strategy-behavior" pair is associated. Lag time calculation: Once the response behavior is identified, calculating the time interval between its actual occurrence time and the release time of the strategic event, which is the strategic response lag data.

[0035] Strategic response lag data can refer to the time interval it takes for each strategic decision event to be reflected in the actual behavior of the enterprise. Its unit is usually "days", "weeks", or "months". For example, a certain enterprise proposed the "Expand Southeast Asian market" strategy in January 2024, and the actual first contract with Southeast Asian customers occurred in mid-March 2024, then the corresponding response lag is "2.5 months". This data reflects the agility and execution efficiency of the enterprise's strategic implementation.

[0036] The time series annotation mechanism refers to a set of rule-based or model-driven mechanisms for mapping text and behavior data to a timeline, used to unify different data sources in the temporal dimension and establish time tags for events. This mechanism includes the following steps: First, perform event time anchoring, that is, identify the time expressions in the enterprise strategic text. Using a time expression recognition model (such as HeidelTime, SUTime, or a BERT-based time entity recognition model), extract explicit or implicit time information (such as "the third quarter of 2023" or "the next three months") from the strategic text, and map these time expressions to standardized timestamps to form a set of time anchors for strategic events. Secondly, standardize the behavior log time. For each record in the enterprise behavior log, extract its original timestamp and construct a time series of behavior events. This process ensures that the occurrence time of each behavior event is accurately recorded and aligns the behavior data with the timeline. Then, perform time series window matching and annotation. For each time anchor of a strategic event, construct a behavior detection window according to the set response window (such as rolling back 30 days or 60 days, or by natural quarter). Then, associate the time tags of all behavior events within this time window with the corresponding strategic event. Next, perform topic alignment and response recognition. Compare the strategic topic vectors generated from each strategic text with the behavior topic vectors within the window, and use semantic similarity calculation (such as calculating the cosine similarity through Transformer encoding) to identify the behavior topic vectors whose similarity exceeds the set threshold and their corresponding time points. Finally, calculate the response time lag, that is, obtain the response time lag data for each strategic event through the time difference between the strategic event time and the behavior event time. This time lag data annotates the degree of delay of the strategic response and is analyzed as part of the time series.

[0037] S105. Analyze the enterprise strategic text according to the behavior topic deviation analysis method to obtain the enterprise strategic topic vectors at each time point, and, analyze the enterprise behavior log data according to the behavior topic deviation analysis method to obtain the enterprise behavior data topic vectors at each time point, calculate the cosine similarity between the enterprise strategic topic vectors and the enterprise behavior data topic vectors, and determine the strategic deviation metric value according to the cosine similarity.

[0038] The behavior topic deviation analysis method can be a technical method for analyzing the deviation degree between enterprise behavior data and strategic goals or strategic content. This method identifies the gap between the actual behavior of the enterprise in implementing the strategy and the expected strategic direction by performing topic modeling and deviation degree analysis on the behavior log data. Specifically, it will compare the actual behavior in the behavior log data with the corresponding strategic goals, calculate the behavior deviation amount, and evaluate the consistency between behavior execution and strategic expectations.

[0039] The enterprise strategic theme vector can be a vector representation obtained after analyzing the enterprise strategic text. It converts the strategic text into a numerical vector form through natural language processing techniques (such as topic modeling, TF-IDF, Word2Vec, etc.), and these vectors can effectively capture the semantic information and topic structure of the strategic text. The strategic theme vector at each time point reflects the core content and strategic direction of the strategic text at that time.

[0040] Enterprise behavior log data can be the original data that records all the behaviors or operations taken by an enterprise within a certain time range. It usually contains information such as the time when the behavior occurred, the type of behavior, the objects involved, and the results of the behavior. For example, during the implementation of the strategy, various specific actions (such as market promotion, product development, personnel transfer, etc.) will be recorded as behavior log data. These logs provide the original materials for analyzing the enterprise behavior patterns and evaluating the consistency between behaviors and strategies.

[0041] The enterprise behavior data theme vector can be a representation obtained by performing topic modeling and vectorization processing on the enterprise behavior log data. It converts the key features in the behavior logs (such as behavior categories, target objects, execution effects, etc.) into numerical vectors. These vectors represent the core features of the enterprise behaviors in different periods and can help analyze the topic distribution and execution trends of the behaviors.

[0042] Cosine similarity can be a mathematical measurement method used to calculate the similarity between two vectors. It determines their similarity by calculating the angle between the two vectors, and its value ranges from -1 (completely opposite) to 1 (completely identical).

[0043] The strategic deviation metric value can be a numerical indicator used to quantify the degree of deviation between the actual behaviors of an enterprise and its strategic goals. By calculating the cosine similarity between the enterprise strategic theme vector and the enterprise behavior data theme vector, a similarity value is obtained, and then this value is used to evaluate the deviation between the strategic goal and the actual execution. Usually, the closer the strategic deviation metric value is to 1, the more consistent the strategic execution is with the strategic goal; if the similarity is low, it means that the strategic execution deviates far from the goal.

[0044] First, a unified sampling time window can be set (e.g., divided by month, quarter, or half-year), and the enterprise strategic text and enterprise behavior log data are sorted into their respective corresponding time periods. For enterprise strategic text, unstructured text data related to enterprise strategy (such as development strategy reports, management discussions in financial reports, investor relations texts, etc.) are collected, cleaned and normalized through text preprocessing techniques. Subsequently, using topic modeling techniques, such as the Latent Dirichlet Allocation method or semantic embedding methods based on language models, the topic structure of the text in each time period is extracted and vectorized to obtain a strategic topic vector reflecting the enterprise's strategic intention. Next, structured modeling is performed on the enterprise behavior log data, including unified coding and classification of information such as operation types, execution objects, and occurrence times in the behavior logs, aggregation according to the time window, and then extraction of the behavior feature topics in this time period based on event frequency statistics, behavior sequence modeling, or deep embedding methods to form an enterprise behavior data topic vector, reflecting the main behavior topics or strategic tendencies executed by the enterprise in actual operations. In the step of strategic deviation analysis, first, the cosine similarity between the enterprise strategic topic vector and the corresponding enterprise behavior topic vector at each time point is calculated. This technology is used to measure the similarity degree of the directions of the two vectors in the semantic space and reflects the consistency in semantic expression. The closer the cosine similarity value is to 1, the more consistent the enterprise's actual behavior is with its strategic intention; the lower the value, the greater the deviation. On this basis, a set of deviation degree measurement methods is defined: for example, (1 minus the cosine similarity value) is used as the strategic deviation measurement value.

[0045] S106. Calculate the strategic consistency score according to the preset sampling time sequence window, enterprise strategic topic vector, enterprise behavior topic vector, strategic response time lag data, strategic deviation measurement value, and the preset consistency score calculation formula.

[0046] The preset sampling time sequence window, enterprise strategic topic vector, enterprise behavior topic vector, strategic response time lag data, and strategic deviation measurement value can be substituted into the preset consistency score calculation formula to obtain the strategic consistency score.

[0047] Based on the above technical solutions, optionally, the preset consistency score calculation formula is:

[0048] Among them, is the strategic consistency score; is the preset sampling time sequence window; t is each time point of the preset sampling time sequence window; is the preset similarity contribution weight; is the enterprise strategic topic vector; is the enterprise behavior topic vector; is the strategic behavior similarity; is the preset response time lag compensation weight; is the strategic response time lag data; is the preset offset penalty weight; is the strategic offset metric value.

[0049] In this solution, , , the relative importance of the three elements can be subjectively weighted according to corporate strategy research or management theory. Or through AHP: construct a pairwise comparison matrix, score according to relative importance, and extract the weight after passing the consistency test; entropy weight method: calculate the information entropy using the degree of variation of each index in the sample, so as to automatically determine the weight, which is suitable for big data scenarios that do not require manual scoring.

[0050] S107, obtain the enterprise technology entropy and the enterprise market entropy, and input the enterprise behavior time series feature vector set, the enterprise strategic theme vector, the enterprise behavior theme vector, the strategic response time lag data, the strategic offset metric value, the strategic consistency score, the enterprise technology entropy, and the enterprise market entropy into the preset enterprise potential generation network to generate an enterprise potential score sequence.

[0051] The enterprise technology entropy can be a metric index that measures the complexity, uncertainty, and innovation diffusion degree of the enterprise technology ability structure, and it is derived from the concept of information entropy. It reflects the diversity, dynamics, and synergy of the enterprise in aspects such as technology R & D, patent layout, and technology portfolio.

[0052] The enterprise market entropy can be an index that measures the diversity, uncertainty of the enterprise market behavior, and the complexity of its market response strategy, and it also comes from the concept of entropy. It reflects the "strategic diversity" and "market response uncertainty" of the enterprise in dimensions such as product portfolio, customer group, channel layout, and price strategy.

[0053] The preset enterprise potential generation network can be a deep learning model architecture for multi-source heterogeneous time series data fusion and potential evaluation. It predicts the medium- and long-term development potential of the enterprise by integrating multi-dimensional features such as enterprise strategic intention, behavior performance, response timeliness, offset degree, and system entropy.

[0054] The enterprise potential score sequence can be a multi-time point prediction value sequence output based on the above potential generation network, which is used to quantify the growth potential or development trend of the enterprise in different future time periods.

[0055] The enterprise technical entropy and enterprise market entropy can be read from the database. The enterprise behavior time series feature vector set, enterprise strategic theme vector, enterprise behavior theme vector, strategic response time lag data, strategic deviation metric value, strategic consistency score, enterprise technical entropy, and enterprise market entropy are input into a preset enterprise potential generation network, which is usually a deep neural network structure with multi-channel time series modeling capabilities, such as a multi-input Transformer architecture or a hybrid model integrating attention mechanism and gated recurrent structure, capable of processing multiple heterogeneous time series and performing interactive modeling on them. The network captures the deep laws between the internal development potential and external response situation of the enterprise by modeling the correlation relationships of each input feature and performing feature fusion, and finally outputs a sequence of enterprise potential scores, that is, prediction scores for the enterprise's strategic execution ability, growth potential, and innovation performance at several future time points, providing data support for subsequent enterprise governance, investment and financing evaluation, or industrial monitoring. For example, taking 12 months of data as a group of sequence samples and inputting them into the network model, the input tensor: Month 1: [behavior characteristics, strategic vector, behavior vector, response time lag, deviation value, consistency, technical entropy, market entropy], Month 2: [...], ... Month 12: [...] The model outputs a "sequence of enterprise potential scores" with a length of 12. For example: [0.71, 0.74, 0.78, 0.80, 0.83, 0.85, 0.84, 0.86, 0.87, 0.89, 0.91, 0.93] It represents the potential scores of the enterprise for each month in 2023. From this, it can be seen that its potential shows an obvious continuous growth trend, and strategic consistency and technology layout play a positive role. The trend of the score sequence on the time axis can be observed. If the score continues to rise, it indicates that the enterprise's potential is growing positively and the development trend is good; if the score fluctuates greatly or continues to decline, it may indicate a weakening of strategic execution ability or a decline in technology and market adaptability. Compare the potential scores at different stages (such as before and after the strategic release, between quarters) to evaluate the impact of specific strategic or behavioral changes. For example, whether the score shows a significant increase after a certain strategy is implemented can be used to reverse-verify the effectiveness of the strategy. Set a lower threshold for the potential score. If the score at a certain time point is lower than the threshold, the early warning mechanism is triggered, indicating possible risks such as strategic behavior disconnection, technology stagnation, or market degradation, to assist the enterprise in making timely adjustments.

[0056] The construction process of the preset enterprise potential generation network includes: ​​In order for the network to effectively process information from different data sources, various types of data must be preprocessed and standardized. The goal of preprocessing is to ensure that each data format is unified and noise is removed. The structured data of the enterprise's financial statements, behavioral time-series data, semantic text data, and visual graph data will be standardized according to their respective characteristics, such as the normalization of numerical data, the vectorization of text data (e.g., word embedding using TF-IDF or BERT), and the preprocessing of image data (such as cropping, scaling, normalization, etc.). After processing the data, this information will be transformed into a set of feature vectors suitable for input to the deep learning model. In particular, the enterprise's behavioral time-series feature vector set, which consists of data in multiple dimensions, such as financial indicators, behavioral sequences, semantic information, etc., after feature extraction and fusion, finally obtains a multi-dimensional and high-dimensional feature vector set. Then, the enterprise strategic theme vector and the enterprise behavioral theme vector, these two vectors are obtained through analysis and modeling based on the enterprise strategic text and behavioral log data. When constructing the enterprise potential generation network, these two vectors will serve as key information for model input. The strategic theme vector obtained by the strategic text through topic modeling and vectorization contains the macro direction and key points of the enterprise strategy, while the behavioral theme vector obtained by the behavioral log through structured modeling and behavioral pattern extraction reflects the actual operations and decision-making behaviors of the enterprise. In order for these vector information to be effectively learned by the network, appropriate encoding and embedding are required (for example, using neural network architectures such as RNN, LSTM, Transformer, etc. to process time-series data). The network also needs to understand the strategic response time lag data of the enterprise strategy and the strategic deviation metric value. The strategic response time lag refers to the time difference between strategic decision-making and execution, which affects the enterprise's adaptability and execution efficiency. The strategic deviation metric value measures the difference between the strategic goal and the actual behavior, and this indicator can help determine whether the enterprise is moving in the original strategic direction. For these two data, the network will capture their time characteristics and patterns through a time-series model (such as the combination of time series convolutional neural network CNN-LSTM) to provide a more accurate enterprise potential analysis. The enterprise's technological entropy and market entropy are important indicators to measure the enterprise's technological innovation and market environment changes. Technological entropy represents the diversity and complexity of the enterprise's technological development and innovation, while market entropy reflects the uncertainty and dynamic changes of the enterprise in the market. By calculating these two entropies, the network can understand the enterprise's capabilities in technological innovation and market adaptation, and thus assist in potential assessment. Technological entropy and market entropy will be used as one of the inputs to the network and jointly affect the generation of the potential score with other data. In terms of architecture, the preset enterprise potential generation network may adopt integrated deep learning methods, including deep learning models such as multi-layer perceptron (MLP), convolutional neural network (CNN), recurrent neural network (RNN), long short-term memory network (LSTM), self-attention mechanism (Self-Attention), etc.Specifically, MLP and CNN are suitable for processing structured data and image data, while RNN and LSTM can handle time series data well (such as behavioral time series data and strategic response delay data). To enhance the expressiveness of the model, a self-attention mechanism can also be added to the network, enabling the network to dynamically adjust weights according to the context information of the data and strengthening the performance of the model. The training process of the enterprise potential generation network is optimized according to the input data, with the goal of enabling the model to generate accurate enterprise potential scores. Commonly used training methods include supervised learning (such as regression problems) and reinforcement learning (such as adjusting scores based on environmental feedback). During the training process, the model continuously adjusts the weight parameters to minimize the prediction error or maximize the accuracy of the potential score. At the same time, cross-validation is used during the training process to prevent overfitting, and advanced optimization algorithms such as the Adam optimizer or SGD optimizer are used to accelerate the training process. After being trained, the enterprise potential generation network inputs all relevant feature data (including strategic data, behavioral data, technology, and market entropy, etc.), and finally outputs a sequence of enterprise potential scores. This score sequence reflects the potential change trend of the enterprise at different time periods. The higher the enterprise potential score, the stronger the possible innovation ability, higher market adaptability, and greater development space of the enterprise in the future.

[0057] In the embodiments of the present application, structured enterprise financial statement data, enterprise behavior time-series data, enterprise semantic text data, and enterprise visual graph data within a preset sampling time-series window are obtained, and an enterprise behavior time-series feature vector set is constructed based on the structured enterprise financial statement data, enterprise behavior time-series data, enterprise semantic text data, and enterprise visual graph data; enterprise strategic texts within a preset sampling time-series window are obtained, and topic modeling and vectorization processing are performed on the enterprise strategic texts to obtain enterprise strategic topic vectors at each time point; enterprise behavior log data within a preset sampling time-series window is obtained, and structured modeling and behavior pattern extraction are performed on the enterprise behavior log data to obtain enterprise behavior topic vectors at each time point; strategic response time-lag data at each time point is extracted from the enterprise strategic texts and enterprise behavior log data according to a time-series annotation mechanism; the enterprise strategic texts are analyzed according to a behavior topic deviation analysis method to obtain enterprise strategic topic vectors at each time point, and the enterprise behavior log data is analyzed according to the behavior topic deviation analysis method to obtain enterprise behavior data topic vectors at each time point. The cosine similarity between the enterprise strategic topic vectors and the enterprise behavior data topic vectors is calculated, and a strategic deviation metric value is determined according to the cosine similarity; a strategic consistency score is calculated according to a preset sampling time-series window, enterprise strategic topic vectors, enterprise behavior topic vectors, strategic response time-lag data, strategic deviation metric value, and a preset consistency score calculation formula; enterprise technology entropy and enterprise market entropy are obtained, and the enterprise behavior time-series feature vector set, enterprise strategic topic vectors, enterprise behavior topic vectors, strategic response time-lag data, strategic deviation metric value, strategic consistency score, enterprise technology entropy, and enterprise market entropy are input into a preset enterprise potential generation network to generate an enterprise potential score sequence. Through the above enterprise potential intelligent recognition method, by integrating multi-dimensional information, the development potential of an enterprise can be comprehensively and accurately evaluated. It not only strengthens the monitoring of the matching degree between strategic execution and actual behavior, but also identifies the potential change trend through time-series analysis, providing a more forward-looking and dynamic evaluation for the enterprise. In addition, this method provides accurate decision-making support for enterprise management, helps identify early signals of potential outbreaks, captures changes during the transformation period and recession period, thereby optimizing resource allocation and achieving more scientific strategic adjustments.

[0058] Based on the above technical solution, optionally, the determination process of enterprise technology entropy includes: Obtain the R & D investment data, number of valid invention patents, patent high-quality rate, achievement conversion rate, and innovation redundancy of the enterprise for each year, and calculate the enterprise technology entropy according to the R & D investment data, number of valid invention patents, patent high-quality rate, achievement conversion rate, innovation redundancy, and a preset technology entropy calculation formula; wherein, the preset technology entropy calculation formula is:

[0059] Wherein, is the enterprise technology entropy; n is the consecutive calculation years; i is the year serial number; is the preset R & D efficiency weight; is the R & D investment data; is the number of valid invention patents; the preset high-quality output weight; is the patent high-quality rate; is the preset achievement transformation ability weight; is the achievement transformation rate; is the preset innovation redundancy weight; is the innovation redundancy.

[0060] In this solution, the R & D investment data can refer to the capital investment of an enterprise in research and development during a specific period (such as one year). It usually includes the expenditures of the enterprise in technological R & D, product innovation, basic research, etc., reflecting the investment scale of the enterprise in technological progress and innovation.

[0061] The number of valid invention patents can refer to the number of invention patents applied for and authorized by an enterprise within a specific time. Valid patents usually refer to those patents that have passed formal examination and are recognized within the prescribed legal framework. It reflects the enterprise's ability and achievements in technological innovation.

[0062] The patent high-quality rate can refer to the proportion of patents with high innovation and market value among the patents applied for by an enterprise. High-quality patents usually mean having strong technological competitiveness, being able to effectively support the long-term development of the enterprise, and improving the enterprise's market position.

[0063] The achievement transformation rate can refer to the proportion of an enterprise's R & D achievements transformed from the laboratory or technology development stage into actual commercial products or technologies. It measures the enterprise's ability to transform R & D innovation into actual market value and economic benefits.

[0064] The innovation redundancy can refer to the R & D activities that may exist in duplicate or over-investment during the enterprise's innovation process. This redundancy usually reflects the efficiency and optimization space in the enterprise's innovation process. High redundancy may mean that resources are not effectively utilized, resulting in the innovation output being disproportionately lower than the input.

[0065] R&D investment data usually comes from a company's annual financial statements and internal budget reports, which clearly list the company's annual R&D investment. A company's R&D expenditures are usually recorded under the item of "R&D expenses" or "R&D investment" in the financial statements, or can be obtained by referring to the company's annual reports and audit reports. In addition, a company can also use its internal financial system to track R&D investment and compare it with other companies in the industry. Secondly, the number of valid invention patents is obtained through patent databases. These patent databases record all the patents applied for and obtained by a company within a certain period of time and can show whether the patents are valid. When querying, it is necessary to screen out those patents that are within the validity period and meet the invention patent standards, and usually confirm their validity through the patent's authorization date and legal status. To accurately count the number of valid invention patents, professional patent analysis tools (such as PatBase, Derwent Innovation, etc.) can be used for batch query. Next, the patent high-quality rate is evaluated based on the technical value and market influence of patents, usually measured by the number of citations of patents. A patent with a high number of citations indicates that it has a greater influence in the technical field and is therefore more valuable. The patent high-quality rate can also be analyzed through the patent conversion situation of a company, whether the company realizes the value of patents through patent licensing, technology cooperation or commercialized products, etc. Relevant data can be obtained from patent databases or evaluated by combining with industry reports and market feedback. The achievement conversion rate reflects the conversion of a company's R&D activities into actual marketable achievements, especially the implementation of new products, new technologies or new services. The data of the achievement conversion rate can be obtained by tracking the commercialization of new products and technologies launched by a company every year and the scope of technology application. A company usually discloses relevant data in its annual report or special achievement report, or obtains specific data on marketable achievements through cooperation with industry research institutions. Finally, the innovation redundancy is an indicator used to measure whether there is duplicate innovation or resource waste in a company's R&D process. By comparing the technical content, objectives and implementation paths of the company's R&D projects over the years and finding the overlapping parts, it is possible to evaluate whether there is duplicate investment. In addition, the R&D project management tools and resource management systems within a company (such as project management software) can provide detailed project resource allocation data, thus revealing whether there is redundant investment. These data can be collected and summarized through the company's internal management system, external patent databases, industry reports and the analysis of market research institutions.

[0066] 、 、 、 It can be set based on the experience of industry experts and their understanding of aspects such as technological innovation and market competition. For example, experts may believe that (related to Pi and Ri) should be larger because technological potential and R & D investment may have the most significant impact on the enterprise's technological level, while 、 、 are smaller because they may involve more detailed or indirectly influential factors (such as quality, stability, efficiency, etc.).

[0067] Based on the above technical solutions, optionally, the process of determining the enterprise market entropy includes: Obtain the coefficient of variation of the enterprise's sales in various markets, the market exit probability of the enterprise in the target market, the growth rate of the enterprise's market share in the target market, and the volatility of the enterprise's customer feedback. Calculate the enterprise market entropy according to the coefficient of variation of sales, the market exit probability, the volatility of customer feedback, and a preset market entropy calculation formula; where the preset market entropy calculation formula is:

[0068] Among them, is the enterprise market entropy; is the preset sales volatility sensitivity coefficient; is the coefficient of variation of sales; is the preset exit risk weight coefficient; is the market exit probability; is the preset share stagnation penalty coefficient; is the growth rate of the market share; is the preset customer feedback volatility weight coefficient; is the volatility of customer feedback.

[0069] The coefficient of variation of sales can be the ratio of the standard deviation to the mean of the enterprise's sales in different markets or multiple time periods, used to measure the degree of sales volatility.

[0070] The target market can refer to the market area, industry segment, or customer group that the enterprise strategically chooses to focus on investing resources in, usually the market that the enterprise believes has the most growth potential or core competitive advantage.

[0071] The market exit probability can be the possibility or frequency of the enterprise exiting a certain market segment or target market, which can be inferred through historical exit events, market production stoppage rates, or customer churn indicators.

[0072] The growth rate of the market share can be the year-on-year or month-on-month growth rate of the enterprise's market share in a certain market.

[0073] The volatility of customer feedback can be used to measure the degree of fluctuation of the evaluations (such as satisfaction, NPS, comment sentiment scores) obtained by the enterprise from customers within a certain time window.

[0074] Yes, you can obtain the sales data of different markets (divided by region, product line, and customer type) through the enterprise ERP system, financial statements, sales management platform, etc., organize the time series data according to the market dimension, and use statistical analysis tools (such as pandas and numpy libraries in Python) to calculate the standard deviation and mean of sales in each market, so as to obtain the coefficient of variation of sales in each market and evaluate its sales stability. Secondly, through data mining of the company's business development and market operation logs, identify the time nodes of the company's entry and exit of a certain market (which can be derived from public news, annual report disclosures, product on and off shelves records, etc.), and combine the total number of participating markets and the frequency of exiting the market to calculate the market exit probability in the target market to reflect its strategic stability. Thirdly, by retrieving industry statistics, market research reports (such as iResearch, IDC, Wind, etc.) and the company's own market sales data, analyze the company's market share at different time points in the target market, calculate the growth rate of market share, and evaluate its competitive expansion capabilities. Finally, we use customer feedback data (such as after-sales systems, satisfaction surveys, online reviews, social media comments, etc.) to conduct text sentiment analysis and score standardization, build a time series of customer feedback scores, and calculate its standard deviation and mean to obtain customer feedback volatility and determine the stability of customer satisfaction. Throughout the process, we need to rely on key steps such as data cleaning, label matching, market mapping, and time alignment to ensure data consistency and comparability, and ultimately form a key input indicator system to support enterprise market entropy modeling.

[0075] , , , Initial weights are set based on the judgment of industry experts, strategic analysts or academic researchers. Alternatively, a pairwise judgment matrix can be constructed to allow multiple experts to score the relative importance of each two indicators. After consistency testing and normalization, the weight of each indicator is obtained. This method can systematically introduce subjective judgments and convert them into quantitative weights.

[0076] Embodiment 2 Figure 2 Schematic diagram of the process of intelligent identification method of enterprise potential provided in Example 2 of this application. Figure 2 As shown, the specific steps include: S201, obtaining a historical enterprise potential score sequence, constructing a historical potential evolution time curve according to the historical enterprise potential score sequence, analyzing the historical potential evolution time curve using a trend modeling algorithm, and obtaining a historical trend change state of enterprise potential.

[0077] The historical enterprise potential score sequence can refer to the time series data formed by the "enterprise potential scores" calculated for an enterprise at each time point within a continuous time window (such as monthly, quarterly, annually).

[0078] The historical potential evolution time curve can be a continuous curve formed by visualizing the "historical enterprise potential score sequence" on the time axis, intuitively showing the dynamic evolution trend of the enterprise potential over time. Its essence is a two-dimensional function image, with the horizontal axis being time and the vertical axis being the potential score.

[0079] The trend modeling algorithm can be an algorithm used to analyze time series data and identify its long-term trend, periodicity, or mutation points. Commonly used algorithms include the moving average method: smoothing short-term fluctuations and extracting the long-term trend; exponential smoothing: assigning higher weights to recent data, suitable for dynamic prediction; linear / nonlinear regression trend analysis: fitting a straight line or polynomial curve to extract the trend; piecewise linear fitting: identifying multiple trend stages and their changes; time series decomposition (STL, EMD): decomposing the time series into three parts: trend, cycle, and residual; deep learning methods (such as LSTM): identifying complex nonlinear change patterns, suitable for long-term trend prediction; Bayesian change point detection: identifying the time points when the trend undergoes a mutation.

[0080] The historical trend change state can be an abstract description of the enterprise potential change state after trend modeling, which can include stable increase: the potential continuously improves, and the growth rate is stable; high-speed growth: the potential rises sharply within a short period; fluctuating adjustment: the potential score fluctuates up and down, with uncertainty; To obtain the historical enterprise potential score sequence, it is first necessary to continuously observe and evaluate the enterprise within a preset sampling time series window (such as monthly, quarterly, or annually). Specifically, based on multi-source information such as the enterprise's financial structured data, behavioral time series data, strategic text data, behavioral log data, and technological entropy and market entropy at each time point, it is input into the trained enterprise potential generation network model to output the enterprise potential score corresponding to the time point, and then a complete historical enterprise potential score sequence is formed. Next, arrange this score sequence in chronological order and visualize it to construct the historical potential evolution time curve, thereby intuitively showing the change trend of the enterprise potential. Subsequently, use the trend modeling algorithm to fit and analyze this time curve. The trend modeling algorithm can be selected from moving average, exponential smoothing, linear regression, polynomial regression, piecewise linear fitting, STL time series decomposition, Bayesian change point detection, or deep learning methods such as LSTM sequence modeling, etc., to identify the characteristic states such as stable growth, fluctuating callback, trend reversal, or platform oscillation in the enterprise potential change. Finally, through the analysis of the curve shape and trend change pattern, the historical trend change state of the enterprise potential can be extracted to help judge whether the enterprise is in the ascending period, mature period, adjustment period, or decline period, providing a basis for strategic planning and investment and financing decisions.

[0081] S202. Determine the historical fluctuation range and historical fluctuation frequency of the historical enterprise potential score sequence, and calculate the historical enterprise potential stability index according to the historical fluctuation range and historical fluctuation frequency.

[0082] The historical fluctuation range may refer to the numerical range within which the enterprise potential score fluctuates up and down in the historical time series, measuring the "fluctuation degree" of the score.

[0083] The historical fluctuation frequency may refer to the number of times the enterprise potential score changes significantly in the historical time series, used to measure the "frequency of change" of the score.

[0084] The historical enterprise potential stability index may be a comprehensive index measuring the historical change stability degree of the enterprise potential score.

[0085] Determining the historical fluctuation range and historical fluctuation frequency of the historical enterprise potential score sequence and calculating the historical enterprise potential stability index accordingly generally includes: First, obtain the potential score sequence generated by the enterprise within consecutive time windows (such as quarterly or annually) to form a set of time series data { , ,..., }, where each represents the enterprise potential score at the t-th time point. Next, to analyze the change of enterprise potential in the time dimension, calculate the historical fluctuation range. The fluctuation range reflects the range or dispersion degree of the score values between different time points. Common calculation methods include the difference between the maximum value and the minimum value (i.e., max( ) - min( ), and the standard deviation formula , where is the mean of the score sequence. Then, detect the fluctuation frequency of the time series scores, define a threshold for score change. When the difference ∣ - ∣ between any two adjacent score points exceeds this threshold, it is considered that a significant fluctuation has occurred. Count the cumulative number of such changes in the entire time series as the fluctuation frequency index. If normalization is required, the average number of fluctuations per unit time (such as the average annual fluctuation number) can be further calculated. After obtaining the fluctuation range and fluctuation frequency, use a preset stability evaluation formula to comprehensively evaluate the historical stability of the enterprise potential score to form the historical enterprise potential stability index. This index can be calculated using the following weighted reciprocal model:

[0086] Among them, It is the historical enterprise potential stability index; and are empirically set weight parameters, reflecting different degrees of emphasis on amplitude and stability. The higher the value of the finally generated stability index, the more stable the potential performance of the enterprise in the historical time period, and the more predictable the strategic execution and market response; a lower value indicates severe potential fluctuations and higher uncertainty and business risks for the enterprise.

[0087] S203, obtain historical external disturbance events, and determine the historical response speed of the enterprise to the historical external disturbance events according to the historical external disturbance events and the historical enterprise potential score sequence.

[0088] Historical external disturbance events may refer to external factors or unexpected events that may have a significant impact on aspects such as enterprise operation, strategy, innovation, or market performance during a specific historical period.

[0089] The historical response speed may refer to the degree of time delay in the reaction of the enterprise's potential score when facing the above external disturbance events. That is, the time interval between the occurrence of the external disturbance event and the obvious change in the enterprise's potential score (or key behavior indicators), reflecting the efficiency of the enterprise in perceiving external changes and making strategic or behavioral adjustments. The faster the response speed, the stronger the enterprise's ability to adapt to environmental changes, and the more agile the strategic adjustment or resource allocation.

[0090] A set of external event identification and classification mechanisms can be established. Specifically, by collecting data from sources such as authoritative news agencies, government announcements, industry association reports, and economic monitoring platforms, key time points and event contents that have had an obvious impact on the macro economy, industry pattern, policies and regulations, technological changes, market competition, or sudden public events in the past several years can be extracted. These events need to be classified according to their scope and nature of influence, such as macroeconomic category, policy change category, technological shock category, or major unexpected event category, and be structured and sorted, marking their occurrence time, affected objects, and expected impact directions. Next, these external events need to be associated with the historical potential score sequence of the enterprise. By aligning the time, observe whether the enterprise's potential score shows a significant change (such as a rapid increase, decrease, or trend inflection point) within a certain time after the occurrence of an external disturbance event. If the score change is obvious, the time interval between the event occurrence and the score change of the enterprise can be recorded, which is the historical response speed of the enterprise to this event. This process needs to use change recognition algorithms or trend analysis means to ensure that the change is driven by the event rather than normal fluctuations.

[0091] Based on the above technical solution, optionally, determining the historical response speed of the enterprise to the historical external disturbance events according to the historical external disturbance events and the historical enterprise potential score sequence includes: Determine the first historical preset time window and the second historical preset time window for the historical external disturbance event, and calculate the first potential slope of the historical enterprise potential score sequence within the first historical preset time window; Calculate the second potential slope of the historical enterprise potential score sequence within the second historical preset time window; Determine the time interval between the first historical preset time window and the second historical preset time window, and calculate the historical response speed of the enterprise to the historical external disturbance event according to the first potential slope, the second potential slope, and the time interval.

[0092] In this solution, the first historical preset time window can be a period of time before the occurrence of the external disturbance event, used to observe the change of the enterprise potential score during this period.

[0093] The second historical preset time window can be a period of time after the occurrence of the external disturbance event, used to observe the impact of the external disturbance event on the enterprise potential score.

[0094] The first potential slope can be the slope (i.e., the rate of change) of the enterprise potential score changing with time within the first historical preset time window. It represents the trend and rate of change of the enterprise potential score before the external disturbance event. If the slope is positive, it means the potential score is rising within this time window; if it is negative, it indicates that the potential score is decreasing.

[0095] The second potential slope can be the slope of the enterprise potential score changing with time within the second historical preset time window, used to measure the rate of change of the enterprise potential score after the occurrence of the external disturbance event. Similar to the first potential slope, it reflects the adaptive change situation of the enterprise after the external disturbance event.

[0096] The time interval can refer to the time difference from the end time of the first historical preset time window to the start time of the second historical preset time window. This time interval helps measure the potential change cycle of the enterprise before and after the occurrence of the external disturbance event. The longer the time interval, the greater the possible impact of the external disturbance event may be, and the longer it may take for the enterprise to adapt to the change.

[0097] The first historical preset time window can be a time period before the occurrence of a perturbation event. Specifically, the first historical preset time window usually starts from a period of time (such as several months or quarters) before the occurrence of an external perturbation event and lasts until the moment when the perturbation event occurs. The change in the enterprise potential score within this time window reflects the potential trend of the enterprise before the perturbation and is mainly used to calculate the potential change trend (potential slope) before the perturbation. The second historical preset time window usually starts from a period of time after the occurrence of an external perturbation event and lasts for a period of time (possibly several months or quarters) after the event. The second historical preset time window is used to calculate the potential change trend after the perturbation to help understand how the enterprise responds to external perturbations and whether it recovers or experiences a further decline in potential. Within the first historical preset time window, analyze the speed at which the enterprise potential score changes over time. Specifically, calculate the rate of change of the potential score within this time period, that is, the potential slope. The potential slope is usually determined by measuring the "growth rate" of the change in the potential score and reflects the growth or decline rate of the potential score during this period of time. Usually, the slope can be obtained by fitting the potential score data within this time window (such as methods like linear regression). The first potential slope indicates the speed at which the enterprise potential changes before the occurrence of the external perturbation event. If the first potential slope is positive, it means the enterprise potential is increasing; if it is negative, it means the potential is decreasing. Within the second historical preset time window, also calculate the speed of change of the potential score, that is, the potential slope. This slope reflects the potential change situation of the enterprise after the external perturbation event. If the change in the enterprise potential speeds up or slows down after the perturbation, the second potential slope will provide quantitative information on this trend. The second potential slope reflects the recovery or decline speed of the enterprise potential after the external perturbation. By fitting the potential score data within the second time window, the potential slope within this window can be obtained. This slope helps to reveal how the enterprise responds to the perturbation. The time interval refers to the time difference between the end moment of the first historical preset time window and the start moment of the second historical preset time window. Usually, this time interval is the moment when the perturbation event occurs. For example, if the perturbation event occurs in a specific month, then the end of the first time window and the start of the second time window both correspond to this event time point. This time interval is usually measured in quarters or months. Finally, by comparing the first potential slope (the potential change speed before the perturbation) and the second potential slope (the potential change speed after the perturbation), and combining the time interval when the perturbation occurs, the response speed of the enterprise to historical external perturbation events can be quantified. The calculation formula for the response speed is as follows:

[0098] where is the historical response speed; is the first potential slope; is the second potential slope; is the time interval.

[0099] In this solution, based on the historical enterprise potential score sequence, the potential slopes before and after the perturbation are calculated, and the response speed of the enterprise to historical external perturbation events is quantified. This provides an important reference basis for the enterprise's coping strategies when facing similar perturbations in the future, and can also reveal the enterprise's adaptability to external changes and the law of potential changes in different periods.

[0100] S204, obtain the historical potential evolution characteristics and historical potential change data of the enterprise, and construct an enterprise potential evolution analysis model according to the historical trend change state, historical enterprise potential stability index, historical response speed, historical potential evolution characteristics, and historical potential change data.

[0101] The historical potential evolution characteristics can be the change trend, fluctuation pattern, and frequency and amplitude of fluctuations of the enterprise potential score within a certain time range. It usually describes the performance of the enterprise's potential in different stages through data such as the change amplitude of the potential score, periodic fluctuations, trend turning points, and extreme values. For example, the sudden increase or decrease of the potential score, long-term stable growth or decline, periodic fluctuations, etc., are all manifestations of the historical potential evolution characteristics.

[0102] The historical potential change data can refer to the specific change situation of the enterprise potential score between each time point in the past period. It reflects the specific increase and decrease of the enterprise potential score in different historical periods and is usually presented in the form of a time series. By calculating the change amount, change speed, change direction, etc. of the enterprise potential score, it can provide an intuitive basis for evaluating the enterprise's development status.

[0103] The enterprise potential evolution analysis model can be a model that uses historical potential evolution characteristics, historical potential change data, and related analysis indicators (such as historical trend change state, historical potential stability, historical response speed, etc.) to predict and analyze the future potential of the enterprise. This model models the evolution process of the enterprise's historical potential through various analysis methods (such as time series analysis, regression analysis, machine learning algorithms, etc.) to predict the future potential change trend of the enterprise in different scenarios.

[0104] The historical potential score data of an enterprise are the basis for constructing a potential evolution analysis model. These data usually come from indicators that regularly evaluate the potential of an enterprise and may involve multiple dimensions such as financial performance, market performance, technological innovation, and strategic execution. Based on these data, the potential score of the enterprise can be calculated for each time point. By analyzing the change trend of the enterprise potential score, the long-term change pattern of the potential score over time can be identified. For example, the enterprise potential score may show a gradual increase, decrease, or fluctuation. The trend can help predict the future change path of the potential. Analyze the fluctuation of the enterprise potential score over a period of time to identify periods of high or low volatility of the potential. The greater the volatility, the more unstable the performance of the enterprise's potential in certain periods, indicating that it may be more affected by external or internal factors.

[0105] To obtain the historical potential evolution characteristics and historical potential change data of an enterprise, it is first necessary to collect the potential performance data of the enterprise at various historical time points. These data usually come from multi-dimensional indicators such as the enterprise's financial statements, market behavior, technological innovation, strategic execution results, industry trends, etc. Based on these indicators, a potential scoring system can be constructed to quantify the potential at each time point to form a potential scoring sequence. On this basis, multiple key features of potential evolution are extracted through methods such as time series analysis, statistical modeling, and external event mapping: trend modeling (such as moving average, exponential smoothing, and regression analysis) is used to identify the long-term trend of potential scores in the time dimension to form trend characteristics; volatility analysis (such as standard deviation and coefficient of variation) is used to characterize the stability of potential scores to form stability characteristics; extreme values, inflection points, or trend mutations are identified through mutation detection to form mutation characteristics; seasonal or cyclical changes in potential scores are identified in combination with periodic analysis methods to form periodic characteristics; and the reaction speed of potential scores is analyzed in combination with external disturbance events (such as policy changes and changes in competitive situations) to form response speed characteristics. The above features are then combined into historical potential evolution characteristics. Based on the above-mentioned historical potential evolution characteristics, historical trend change status, historical potential stability index and historical response speed data, a variety of modeling techniques can be used to construct an enterprise potential evolution analysis model. In the process of constructing an enterprise potential evolution analysis model, the historical potential evolution characteristics need to be taken as the core input first. This feature is formed by combining multi-dimensional features such as trend extraction, fluctuation amplitude calculation, extreme mutation identification and periodic fluctuation analysis of the potential change data of the enterprise at multiple time points. It can comprehensively characterize the evolution path of the enterprise potential over time. On this basis, the historical trend change status is extracted, and sliding window analysis, segmented regression, time series trend modeling and other means can be used to identify whether there is a significant growth trend, downward trend, fluctuation trend or stage change trend in the potential change, thereby reflecting the long-term direction and periodic characteristics of the enterprise potential development. At the same time, through a comprehensive analysis of statistical characteristics such as fluctuation frequency, fluctuation range, and abnormal point density in the evolution process, the historical potential stability index is further extracted to quantify the stability of the enterprise potential evolution process and identify whether the potential performance has violent fluctuations or highly controllable characteristics. In addition, by constructing a time alignment model between external disturbance events and potential changes, measuring the time lag and change speed of potential changes, the company's historical response speed index can be obtained, which reflects the company's sensitivity and resilience to changes in the external environment. After completing the above feature engineering, all constructed features are standardized or normalized to form a model input matrix. Subsequently, a variety of modeling techniques can be used, such as time window-based regression models, long short-term memory networks (LSTM), gradient boosting trees (GBDT), random forests, etc., to model and predict the evolution trend of corporate potential.The model output can include the classification of the evolution path of future potential (such as rising, falling, oscillating), the range of changes in the degree of evolution, potential abnormal evolution risks, etc., so as to provide data support and forward-looking insights for corporate strategic planning, investment judgment, and policy formulation. The key to constructing this model lies in the temporal expression of features, the pattern recognition of evolutionary behaviors, and the dynamic coordination among multi-dimensional indicators, so as to achieve accurate modeling and intelligent prediction of the evolution of corporate potential.

[0106] In this embodiment, by obtaining the historical corporate potential score sequence and constructing the historical potential evolution time curve, combining the trend modeling algorithm to extract the historical trend change state, further analyzing the fluctuation amplitude and frequency to calculate the stability index, and evaluating the response speed of the enterprise based on historical external perturbation events, and finally jointly constructing the corporate potential evolution analysis model with the extracted historical potential evolution characteristics, this process can achieve multi-dimensional dynamic characterization of the evolution trajectory of corporate potential, accurately identify the potential change patterns and influencing factors of the enterprise at different stages, improve the prediction accuracy and strategic judgment ability of the future potential trend of the enterprise, so as to provide systematic and scientific data support for investment and financing decisions, industrial policy formulation, and the strategic adjustment of the enterprise itself.

[0107] On the basis of the above technical solution, optionally, after constructing the corporate potential evolution analysis model according to the historical trend change state, the historical corporate potential stability index, the historical response speed, and the historical potential evolution characteristics, the method further includes: Constructing a potential evolution time curve based on the corporate potential score sequence, and using the trend modeling algorithm to analyze the potential evolution time curve to obtain the trend change state of the corporate potential; Determining the fluctuation amplitude and fluctuation frequency of the corporate potential score sequence, and calculating the corporate potential stability index according to the fluctuation amplitude and fluctuation frequency; Obtaining external perturbation events within a preset sampling time window, and determining the response speed of the enterprise to the external perturbation events according to the external perturbation events and the corporate potential score sequence; Inputting the trend change state, the corporate potential stability index, and the response speed into the corporate potential evolution analysis model to obtain the potential evolution characteristics of the enterprise.

[0108] In this solution, the potential evolution time curve can refer to a change curve drawn in chronological order based on the corporate potential score sequence (such as the potential score of each month / quarter). This curve reflects the evolution trajectory of the corporate potential at different time points, and is used to intuitively observe the growth, decline, or fluctuation trend of the corporate potential.

[0109] The trend change state can be the direction and morphological characteristics of enterprise potential change extracted by applying trend modeling algorithms (such as linear regression, time series decomposition, smoothing, etc.) to the "potential evolution time curve".

[0110] The fluctuation range can refer to the maximum difference in the potential score within a unit of time (such as a quarter or a year), which indicates the severity of the potential change.

[0111] The fluctuation frequency can be the number of times the potential score change direction (increase or decrease) switches within a certain time window, reflecting the frequency of the change.

[0112] The enterprise potential stability index can be a comprehensive stability measurement index calculated based on the fluctuation amplitude and frequency. Common methods include standard deviation, coefficient of variation, amplitude average, etc. The lower the value, the more stable the enterprise potential performance and the smaller the fluctuation; the higher the value, the more dramatic the potential change and the poor stability.

[0113] External disturbance events can refer to external environmental change events that may have a significant impact on the potential of the enterprise within the sampling time window.

[0114] Response speed can refer to the time interval (such as days, weeks, etc.) required for a company's potential score to change significantly after an external disturbance occurs, as well as the magnitude of the change. It reflects the company's "agility" and "adaptability" to changes in the external environment.

[0115] The potential evolution characteristics can be a set of high-order features that can represent the law of change of enterprise potential, which is extracted after the above indicators (trend change status, potential stability indicators, response speed) are input into the model as characteristic variables.

[0116] By constructing the enterprise potential score sequence, first, the potential evolution time curve can be obtained, and the trend modeling algorithm can be used to analyze this curve to identify the long-term change trend of the enterprise potential, that is, the trend change state. Then, by analyzing the fluctuation amplitude and frequency of the enterprise potential score sequence, the stability index of the enterprise potential can be calculated, so as to evaluate the degree and stability of potential fluctuations. At the same time, combined with the external disturbance events within the preset sampling time series window, analyze the response speed of the enterprise when facing these disturbance events, and reveal the adaptability of the enterprise. Input these data (trend change state, potential stability index, response speed) into the enterprise potential evolution analysis model, and the model will output the potential evolution characteristics of the enterprise based on these inputs, including potential evolution patterns, potential stability, adaptability, and potential evolution trends, so as to provide accurate decision-making support for enterprise management and help better understand the changes and future development directions of enterprise potential. For example, the model output may include the potential evolution trend: the model may output the overall change trend of the enterprise potential over a period of time, such as "The enterprise potential shows a steady upward trend and is expected to continue to grow in the next three quarters", or "The enterprise potential has experienced drastic fluctuations in the past year, and the future potential performance may be unstable". Potential stability: The model can evaluate the stability of the enterprise potential. For example, "The potential performance of this enterprise has high stability, with small fluctuations, indicating its strong market adaptability", or "The enterprise potential has large fluctuations and is easily affected by the external market and internal operations, and it is necessary to strengthen strategic adjustment and response capabilities". External disturbance response ability: The model can also reveal the response speed and effect of the enterprise to external disturbances. For example, "The response speed of the enterprise when facing changes in market demand is two months, and the reaction is relatively rapid", or "The enterprise responds slowly when facing external competition pressure, with a certain lag, which may affect its long-term development". Potential change pattern: The model can identify the fluctuation pattern of the potential score, such as "The enterprise potential score shows seasonal fluctuations, mainly affected by the release of annual financial reports and market off-peak and peak seasons", or "The enterprise potential shows periodic rises and falls, usually related to changes in industry policies". Adaptability assessment: The model can output the adaptability assessment of the enterprise, such as "The potential evolution performance of the enterprise in technological innovation is strong, with strong market adaptability and innovation capabilities", or "The strategic execution ability of the enterprise is weak, lacking effective market response strategies, and the future potential may decline".

[0117] In this solution, by constructing a potential evolution time curve based on the enterprise potential score sequence and conducting trend modeling analysis, it can help enterprises identify long-term trends and cyclical changes in potential changes, thereby providing a predictive basis for future potential development. At the same time, by analyzing the fluctuation range and frequency of potential scores, the stability of enterprise potential can be quantified, thereby identifying the enterprise's adaptability and market performance reliability in different time periods. Combined with the response speed of external disturbance events, the model can reveal the enterprise's responsiveness and response speed when facing changes in the external environment, thereby providing more accurate decision-making support for the enterprise. These analysis results provide enterprise management with an in-depth understanding of potential evolution, which helps to formulate more scientific strategic plans and measures to cope with market challenges, and enhance the long-term competitiveness and sustainable development capabilities of the enterprise.

[0118] Embodiment 3 Figure 3 Schematic diagram of the structure of the enterprise potential intelligent identification system provided in the third embodiment of the present application. Figure 3 As shown, specifically including: The behavior time series feature vector set construction module 301 is used to obtain the enterprise financial report structured data, enterprise behavior time series data, enterprise semantic text data and enterprise visual map data within a preset sampling time series window, and construct the enterprise behavior time series feature vector set according to the enterprise financial report structured data, enterprise behavior time series data, enterprise semantic text data and enterprise visual map data; The enterprise strategy theme vector determination module 302 is used to obtain the enterprise strategy text within a preset sampling time window, perform theme modeling and vectorization processing on the enterprise strategy text, and obtain the enterprise strategy theme vector at each time point; The enterprise behavior theme vector determination module 303 is used to obtain the enterprise behavior log data within a preset sampling time window, perform structured modeling and behavior pattern extraction on the enterprise behavior log data, and obtain the enterprise behavior theme vector at each time point; A strategic response time lag data determination module 304 is used to extract strategic response time lag data at each time point from the enterprise strategy text and enterprise behavior log data according to a time series annotation mechanism; A strategy deviation metric value determination module 305 is used to analyze the enterprise strategy text according to the behavior theme deviation analysis method to obtain the enterprise strategy theme vector at each time point, and to analyze the enterprise behavior log data according to the behavior theme deviation analysis method to obtain the enterprise behavior data theme vector at each time point, calculate the cosine similarity between the enterprise strategy theme vector and the enterprise behavior data theme vector, and determine the strategy deviation metric value according to the cosine similarity; The strategic consistency score calculation module 306 is configured to calculate a strategic consistency score according to a preset sampling time series window, an enterprise strategic theme vector, an enterprise behavior theme vector, strategic response delay data, a strategic deviation metric value, and a preset consistency score calculation formula; The enterprise potential score sequence generation module 307 is configured to obtain an enterprise technology entropy and an enterprise market entropy, and input an enterprise behavior time series feature vector set, an enterprise strategic theme vector, an enterprise behavior theme vector, strategic response delay data, a strategic deviation metric value, a strategic consistency score, an enterprise technology entropy, and an enterprise market entropy into a preset enterprise potential generation network to generate an enterprise potential score sequence.

[0119] The enterprise potential intelligent recognition system provided by the embodiments of the present application can implement Figure 1 each process implemented by the method embodiments. To avoid repetition, it will not be described in detail here.

[0120] Embodiment 4 As Figure 4 shown, the embodiments of the present application further provide an electronic device 400, including a processor 401, a memory 402, a program or instruction stored on the memory 402 and executable on the processor 401. When the program or instruction is executed by the processor 401, it implements each process of the method embodiment of the above enterprise potential intelligent recognition method, and can achieve the same technical effect. To avoid repetition, it will not be described in detail here.

[0121] It should be noted that the electronic device in the embodiments of the present application includes the above-mentioned mobile electronic device and non-mobile electronic device.

[0122] Embodiment 5 The embodiments of the present application further provide a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, it implements each process of the embodiment of the above cable installation process based on tension adaptive control system, and can achieve the same technical effect. To avoid repetition, it will not be described in detail here.

[0123] Wherein, the processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk, etc.

[0124] It should be noted that in this text, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, such that a process, method, article or system comprising a series of elements not only includes those elements but also other elements not expressly listed, or elements inherent to such process, method, article or system. Without further limitation, an element defined by the phrase "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or system comprising that element. In addition, it should be pointed out that the scope of the methods and systems in the embodiments of the present application is not limited to performing functions in the order shown or discussed, but may also include performing functions in a substantially simultaneous manner or in a reverse order according to the functions involved. For example, the described methods may be performed in an order different from that described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0125] Through the description of the above embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present application.

[0126] The embodiments of the present application have been described above in conjunction with the accompanying drawings. However, the present application is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present application, those of ordinary skill in the art can also make many forms without departing from the purpose of the present application and the scope protected by the claims, and all of them fall within the protection scope of the present application.

[0127] The above is only the preferred embodiment of the present application and the technical principles applied. The present application is not limited to the specific embodiments described herein. Various obvious changes, re-adjustments and substitutions that can be made by those skilled in the art will not depart from the protection scope of the present application. Therefore, although the present application has been described in more detail through the above embodiments, the present application is not limited to the above embodiments. Without departing from the concept of the present application, more other equivalent embodiments may be included, and the scope of the present application is determined by the scope of the claims.

Claims

1. An intelligent recognition method for enterprise potential, characterized in that The method includes: Obtain the structured data of the enterprise financial statements, the time-series data of enterprise behaviors, the semantic text data of the enterprise, and the visual graph data of the enterprise within a preset sampling time-series window, and construct an enterprise behavior time-series feature vector set according to the structured data of the enterprise financial statements, the time-series data of enterprise behaviors, the semantic text data of the enterprise, and the visual graph data of the enterprise; Obtain the enterprise strategy text within a preset sampling time-series window, perform topic modeling and vectorization processing on the enterprise strategy text to obtain the enterprise strategy theme vectors at each time point; Obtain the enterprise behavior log data within a preset sampling time-series window, perform structured modeling and behavior pattern extraction on the enterprise behavior log data to obtain the enterprise behavior theme vectors at each time point; Extract the strategic response time-lag data at each time point from the enterprise strategy text and the enterprise behavior log data according to the time-series annotation mechanism; Analyze the enterprise strategy text according to the behavior theme deviation analysis method to obtain the enterprise strategy theme vectors at each time point, and analyze the enterprise behavior log data according to the behavior theme deviation analysis method to obtain the enterprise behavior data theme vectors at each time point, calculate the cosine similarity between the enterprise strategy theme vectors and the enterprise behavior data theme vectors, and determine the strategic deviation metric value according to the cosine similarity; Calculate the strategic consistency score according to the preset sampling time-series window, the enterprise strategy theme vectors, the enterprise behavior theme vectors, the strategic response time-lag data, the strategic deviation metric value, and the preset consistency score calculation formula; Obtain the enterprise technology entropy and the enterprise market entropy, and input the enterprise behavior time-series feature vector set, the enterprise strategy theme vectors, the enterprise behavior theme vectors, the strategic response time-lag data, the strategic deviation metric value, the strategic consistency score, the enterprise technology entropy, and the enterprise market entropy into a preset enterprise potential generation network to generate an enterprise potential score sequence.

2. The method according to claim 1, wherein Wherein, The determination process of the enterprise technology entropy includes: Obtain the R & D investment data, the number of valid invention patents, the patent high-quality rate, the achievement conversion rate, and the innovation redundancy of the enterprise for each year, and calculate the enterprise technology entropy according to the R & D investment data, the number of valid invention patents, the patent high-quality rate, the achievement conversion rate, the innovation redundancy, and the preset technology entropy calculation formula; wherein, the preset technology entropy calculation formula is: Among them, is the enterprise technology entropy; n is the consecutive calculation years; i is the year serial number; is the preset R & D efficiency weight; is the R & D investment data; is the number of valid invention patents; preset high-quality output weight; is the patent high-quality rate; is the preset achievement transformation ability weight; is the achievement conversion rate; is the preset innovation redundancy weight; is the innovation redundancy.

3. The method according to claim 1, wherein Wherein, The determination process of the enterprise market entropy includes: Obtain the coefficient of variation of the sales volume of the enterprise in various markets, the market exit probability of the enterprise in the target market, the market share growth rate of the enterprise in the target market, and the customer feedback volatility of the enterprise, and calculate the enterprise market entropy according to the coefficient of variation of the sales volume, the market exit probability, the customer feedback volatility, and the preset market entropy calculation formula; wherein, the preset market entropy calculation formula is: Among them, is the enterprise market entropy; is the preset sales fluctuation sensitivity coefficient; is the coefficient of variation of sales; is the preset exit risk weight coefficient; is the market exit probability; is the preset share stagnation penalty coefficient; is the market share growth rate; is the preset customer feedback fluctuation weight coefficient; is the customer feedback volatility.

4. The method according to claim 1, wherein Wherein, The preset consistency score calculation formula is: Among them, is the strategic consistency score; is the preset sampling time series window; t is each time point of the preset sampling time series window; is the preset similarity contribution weight; is the enterprise strategic theme vector; is the enterprise behavior theme vector; is the strategic behavior similarity; is the preset response time lag compensation weight; is the strategic response time lag data; is the preset offset penalty weight; is the strategic offset measurement value.

5. The method according to claim 1, characterized in that Wherein, After generating the enterprise potential score sequence, the method further includes: Obtain the historical enterprise potential score sequence, construct a historical potential evolution time curve according to the historical enterprise potential score sequence, and analyze the historical potential evolution time curve by using a trend modeling algorithm to obtain the historical trend change state of the enterprise potential. Determine the historical fluctuation range and historical fluctuation frequency of the historical enterprise potential score sequence, and calculate the historical enterprise potential stability index according to the historical fluctuation range and historical fluctuation frequency; Obtain historical external disturbance events, and determine the historical response speed of the enterprise to the historical external disturbance events according to the historical external disturbance events and the historical enterprise potential score sequence; Obtain the historical potential evolution characteristics of the enterprise, and construct an enterprise potential evolution analysis model according to the historical trend change state, the historical enterprise potential stability index, the historical response speed, and the historical potential evolution characteristics.

6. The method according to claim 5, characterized in that Among them, After constructing the enterprise potential evolution analysis model according to the historical trend change state, the historical enterprise potential stability index, the historical response speed, and the historical potential evolution characteristics, the method further includes: Construct a potential evolution time curve based on the enterprise potential score sequence, and analyze the potential evolution time curve by using a trend modeling algorithm to obtain the trend change state of the enterprise potential; Determine the fluctuation range and fluctuation frequency of the enterprise potential score sequence, and calculate the enterprise potential stability index according to the fluctuation range and fluctuation frequency; Obtain external disturbance events within a preset sampling time sequence window, and determine the response speed of the enterprise to the external disturbance events according to the external disturbance events and the enterprise potential score sequence; Input the trend change state, the enterprise potential stability index, and the response speed into the enterprise potential evolution analysis model to obtain the potential evolution characteristics of the enterprise.

7. The method according to claim 5, wherein Among them, Determining the historical response speed of the enterprise to the historical external disturbance events according to the historical external disturbance events and the historical enterprise potential score sequence includes: Determine the first historical preset time window and the second historical preset time window of the historical external disturbance events, and calculate the first potential slope of the historical enterprise potential score sequence within the first historical preset time window; Calculate the second potential slope of the historical enterprise potential score sequence within the second historical preset time window; Determine the time interval between the first historical preset time window and the second historical preset time window, and calculate the historical response speed of the enterprise to the historical external disturbance events according to the first potential slope, the second potential slope, and the time interval.

8. An enterprise potential intelligent recognition system, characterized in that, The system includes: A behavioral time sequence feature vector set construction module, configured to obtain the structured data of the enterprise financial statements, the enterprise behavioral time sequence data, the enterprise semantic text data, and the enterprise visual graph data within a preset sampling time sequence window, and construct an enterprise behavioral time sequence feature vector set according to the structured data of the enterprise financial statements, the enterprise behavioral time sequence data, the enterprise semantic text data, and the enterprise visual graph data; An enterprise strategic theme vector determination module, configured to obtain the enterprise strategic text within a preset sampling time sequence window, perform topic modeling and vectorization processing on the enterprise strategic text to obtain the enterprise strategic theme vector at each time point; An enterprise behavior theme vector determination module, configured to obtain the enterprise behavior log data within a preset sampling time sequence window, perform structured modeling and behavior pattern extraction on the enterprise behavior log data to obtain the enterprise behavior theme vector at each time point. A strategic response latency data determination module, configured to extract strategic response latency data at each time point from enterprise strategic texts and enterprise behavior log data according to a time series annotation mechanism; A strategic deviation metric determination module, configured to analyze enterprise strategic texts according to a behavior theme deviation analysis method to obtain enterprise strategic theme vectors at each time point, and analyze enterprise behavior log data according to the behavior theme deviation analysis method to obtain enterprise behavior data theme vectors at each time point, calculate the cosine similarity between the enterprise strategic theme vectors and the enterprise behavior data theme vectors, and determine a strategic deviation metric according to the cosine similarity; A strategic consistency score calculation module, configured to calculate a strategic consistency score according to a preset sampling time series window, enterprise strategic theme vectors, enterprise behavior theme vectors, strategic response latency data, strategic deviation metrics, and a preset consistency score calculation formula; An enterprise potential score sequence generation module, configured to obtain enterprise technology entropy and enterprise market entropy, and input an enterprise behavior time series feature vector set, enterprise strategic theme vectors, enterprise behavior theme vectors, strategic response latency data, strategic deviation metrics, strategic consistency scores, enterprise technology entropy, and enterprise market entropy into a preset enterprise potential generation network to generate an enterprise potential score sequence.

9. An electronic device, characterized in that, It includes a processor, a memory, and a program or instruction stored on the memory and executable on the processor. When the program or instruction is executed by the processor, the steps of the enterprise potential intelligent recognition method according to any one of claims 1-7 are implemented.

10. A readable storage medium, characterized in that, A program or instruction is stored on the readable storage medium. When the program or instruction is executed by the processor, the steps of the enterprise potential intelligent recognition method according to any one of claims 1-7 are implemented.

Images