Organization decision-making method and equipment based on time sequence prediction and knowledge graph, and medium

Abstract

The invention relates to the technical field of organization decision management, in particular to an organization decision method and device based on time sequence prediction and a knowledge graph and a medium, and the method comprises the steps: obtaining a prediction target and a decision demand corresponding to the prediction target; based on the prediction target, obtaining an organization sub-behavior transfer matrix, organization knowledge graph data, index time sequence prediction data and target weight data; based on the decision demand, determining an organization sub-behavior probability matrix according to the organization sub-behavior transfer matrix, the organization knowledge graph data, the index time sequence prediction data and the target weight data; and determining a target decision result corresponding to the decision demand according to the organization sub-behavior probability matrix, thereby realizing accurate locking of the most probable organization sub-behavior mode in a comprehensive and complex information network, considering decision-related factors from a more stereoscopic and more comprehensive perspective, and improving decision-making efficiency. And the accuracy of the target decision result is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology in organizational decision-making management, and in particular to an organizational decision-making method, device and medium based on time series prediction and knowledge graph. Background Art

[0002] The current decision-making environment of various organizations is complex and changeable, which is caused by the complexity of data sources, changing market conditions and numerous internal management affairs. Traditional decision-making relies on the experience and intuition of managers. It was feasible in the past when the amount of data was small and the environment was simple. However, with the rapid increase in data volume and complex environment, traditional decision-making technology has made it difficult for organizations to make accurate decisions based on current needs.

[0003] With the development of artificial intelligence technology, computers can simulate human intelligent behavior, and artificial intelligence has also been applied to the field of decision-making. However, when the current models trained based on artificial intelligence are used to classify and predict business data, there are problems such as single considerations and models, which leads to low accuracy of decision-making results. Summary of the invention

[0004] Based on this, it is necessary to address the technical problem that the existing artificial intelligence training model has a single consideration factor and model when classifying and predicting business data, resulting in low accuracy of decision results. This application proposes an organizational decision-making method, device and medium based on time series prediction and knowledge graph.

[0005] In a first aspect of the present application, a method for organizational decision-making based on time series prediction and knowledge graph is provided, the method comprising: Obtaining a prediction target and a decision requirement corresponding to the prediction target; Based on the prediction target, obtain the organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data and target weight data; Based on the decision-making requirements, determining the organizational sub-behavior probability matrix according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data; According to the organizational sub-behavior probability matrix, a target decision result corresponding to the decision requirement is determined.

[0006] Furthermore, the step of determining the organizational sub-behavior probability matrix based on the decision-making requirements, according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data, includes: Based on the decision-making requirements, target data is generated according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data; The target data is input into a hidden Markov model to predict sub-behaviors, and the organizational sub-behavior probability matrix is ​​obtained.

[0007] Furthermore, the step of generating target data based on the decision-making requirements, according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data, includes: Obtaining sub-item prediction indicators corresponding to the decision-making requirements as a target indicator set; According to the organizational knowledge graph data, the indicator time series prediction data and the target weight data, predict the indicator value for each of the sub-item prediction indicators in the target indicator set to obtain a prediction result; Data fusion is performed according to the decision-making requirements, the prediction results corresponding to the target indicator set, and the organizational sub-behavior transfer matrix to obtain the target data.

[0008] Furthermore, the step of determining the target decision result corresponding to the decision requirement according to the organizational sub-behavior probability matrix includes: Inputting the decision requirement and the organizational sub-behavior probability matrix into a pre-trained result prediction model for classification prediction to obtain a prediction vector; Generate a decision result for each decision behavior in the decision requirement according to the prediction vector; According to each of the decision results corresponding to the decision requirement, the target decision result corresponding to the decision requirement is generated.

[0009] Furthermore, the step of obtaining the organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data and target weight data based on the prediction target includes: Obtaining a sub-behavior transfer matrix corresponding to the tissue type of the target tissue corresponding to the prediction target as the tissue sub-behavior transfer matrix; Acquire the knowledge graph data of the target organization corresponding to the prediction target as the organizational knowledge graph data, wherein the organizational knowledge graph data is used to describe key entities, key basic indicators and entity relationships; Obtaining target historical data of the target organization corresponding to the prediction target, predicting future indicator values ​​based on the target historical data and key basic indicators and entity relationships in the organizational knowledge graph data, and obtaining the indicator time series prediction data; Obtain indicator weight data of the target organization corresponding to the prediction target as target weight data.

[0010] Furthermore, the step of predicting future indicator values ​​based on the target historical data and the key basic indicators and entity relationships in the organizational knowledge graph data to obtain the indicator time series prediction data includes: The target historical data and the key basic indicators and entity relationships in the organizational knowledge graph data are input into a pre-trained sequence prediction model to predict the future indicator values ​​of the key basic indicators to obtain the indicator time series prediction data, wherein the sequence prediction model is a model trained based on the Transformer architecture.

[0011] Furthermore, before the step of obtaining the indicator weight data of the target organization corresponding to the prediction target as the target weight data, the step further includes: Obtaining non-quantifiable data of the target tissue; The non-quantifiable data is input into a pre-trained weight prediction model to predict the weight values ​​of the key basic indicators in the organizational knowledge graph data to obtain the indicator weight data of the target organization, wherein the weight prediction model is a model trained based on a generative language model.

[0012] Furthermore, the step of obtaining a sub-behavior transfer matrix corresponding to the tissue type of the target tissue corresponding to the prediction target as the tissue sub-behavior transfer matrix includes: Acquire a sub-behavior transfer matrix as an initial matrix according to the tissue type of the target tissue corresponding to the prediction target; generating adjustment data according to the current organizational life cycle of the target organization; The initial matrix is ​​corrected according to the adjustment data to obtain the organizational sub-behavior transfer matrix corresponding to the prediction target.

[0013] In a second aspect of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned organizational decision-making method based on time series prediction and knowledge graph when executing the computer program.

[0014] In a third aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned organizational decision-making method based on time series prediction and knowledge graph are implemented.

[0015] The organizational decision-making method, device and medium based on time series prediction and knowledge graph of this application have the following specific technical effects: The organizational knowledge graph data is derived from advanced knowledge representation and reasoning technology. It can represent various entities inside and outside the organization and their complex relationships in a structured and semantic way, thereby mining the associated information hidden behind the data; the indicator time series prediction data uses powerful time series analysis artificial intelligence technology to accurately capture the trends and laws of indicator values ​​over time, providing a dynamic basis for decision-making; the organizational sub-behavior transfer matrix is ​​based on behavioral modeling and analysis technology to effectively characterize the conversion pattern of organizational behavior. During the entire process, this application processes rich information sources (i.e., organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data, and target weight data) based on decision-making needs to determine the organizational sub-behavior probability matrix, and determines the target decision results corresponding to the decision-making needs based on the organizational sub-behavior probability matrix. It integrates various types of information sources, which are based on different artificial intelligence technologies. By integrating these information sources based on different artificial intelligence technologies for processing, the organizational sub-behavior probability matrix is ​​determined, thereby achieving accurate locking of the most likely organizational sub-behavior pattern in a comprehensive and complex information network. Compared with traditional methods, this application can consider decision-related factors from a more three-dimensional and comprehensive perspective, greatly improving the accuracy of target decision results, and better meeting the requirements of modern organizations for decision accuracy in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] in: Figure 1 An application environment diagram of an organizational decision-making method based on time series prediction and knowledge graph in one embodiment; Figure 2 is a flowchart of an organizational decision-making method based on time series prediction and knowledge graph in one embodiment; Figure 3 is a structural block diagram of an electronic device in one embodiment; Figure 4 FIG. 4 is another structural block diagram of an electronic device in an embodiment. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] The organizational decision-making method based on time series prediction and knowledge graph provided by the embodiment of the present invention can be applied in Figure 1 In the application environment, the client 110 communicates with the server 120 through the network.

[0020] The server 120 can obtain the prediction target and the decision requirement corresponding to the prediction target through the client 110. The server 120 is configured to implement the following steps: based on the prediction target, obtain the organization sub-behavior transition matrix, the organization knowledge graph data, the indicator time series prediction data and the target weight data; based on the decision requirement, determine the organization sub-behavior probability matrix according to the organization sub-behavior transition matrix, the organization knowledge graph data, the indicator time series prediction data and the target weight data; determine the target decision result corresponding to the decision requirement according to the organization sub-behavior probability matrix.

[0021] Organizational knowledge graph data is derived from advanced knowledge representation and reasoning technology. It can represent various entities inside and outside the organization and their complex relationships in a structured and semantic way, thereby mining the associated information hidden behind the data.

[0022] The indicator time series forecasting data uses powerful time series analysis artificial intelligence technology to accurately capture the trends and patterns of indicator values ​​changing over time, providing a dynamic basis for decision-making.

[0023] The organizational sub-behavior transfer matrix is ​​based on behavioral modeling and analysis technology, which effectively describes the conversion pattern of sub-behaviors of the organization's decision-making behavior.

[0024] During the entire process, this embodiment processes rich information sources (i.e., organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data, and target weight data) based on decision-making needs to determine the organizational sub-behavior probability matrix, and determines the target decision results corresponding to the decision-making needs based on the organizational sub-behavior probability matrix. It integrates various types of information sources, which are based on different artificial intelligence technologies. By integrating these information sources based on different artificial intelligence technologies, the organizational sub-behavior probability matrix is ​​determined, thereby achieving accurate locking of the most likely organizational sub-behavior pattern in a comprehensive and complex information network. Compared with traditional methods, this application can consider decision-related factors from a more three-dimensional and comprehensive perspective, greatly improving the accuracy of target decision results, and better meeting the requirements of modern organizations for decision accuracy in complex environments.

[0025] The client 110 may be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, and portable wearable devices. The server 120 may be implemented by an independent server or a server cluster consisting of multiple servers.

[0026] The present invention is described in detail below through specific embodiments.

[0027] See also Figure 2 As shown, Figure 2 A flowchart of an organizational decision-making method based on time series prediction and knowledge graph provided in an embodiment of the present invention includes the following steps: S1: Obtain a prediction target and a decision requirement corresponding to the prediction target; A forecast target is a description of the expected results of one or more aspects of the future set by an organization based on its own development plan or strategic needs. For example, the forecast target is to increase market share.

[0028] An organization refers to a collective or group of people who collaborate with each other to achieve certain goals, such as a corporate organization or a non-profit organization.

[0029] Decision requirements are specific issues or areas that need to be addressed in order to achieve the forecasting goal. Decision requirements include one or more decision actions. Each decision action is achieved through multiple sub-actions. For example, if the forecasting goal is to increase market share, the decision action is the market share growth rate.

[0030] Decision requirements are decomposed from forecast targets. They are the key content that enterprises must figure out or do something about in the process of pursuing forecast targets. If the forecast target is the destination, decision requirements are the key path or key task to the destination.

[0031] Sub-behaviors are subordinate to decision-making behaviors, and sub-behaviors are specific operational steps to achieve decision-making behaviors. For example, when the market share growth rate is used as the decision-making behavior, the decision-making behavior mainly includes four main sub-behaviors, including: customer acquisition (mainly including potential customer identification sub-behaviors and new channel development sub-behaviors), customer retention (mainly including customer service improvement sub-behaviors), competition response (mainly including competitive product analysis and refinement sub-behaviors, differentiated competition strategy formulation sub-behaviors) and product-related (mainly including product function optimization sub-behaviors and product cost control sub-behaviors).

[0032] Specifically, the prediction target and the decision requirements corresponding to the prediction target input by the user through the client can be obtained, the prediction target and the decision requirements corresponding to the prediction target input by the user through a third-party application can be obtained, and the prediction target and the decision requirements corresponding to the prediction target stored in advance by the user in a preset storage space can also be obtained.

[0033] Optionally, the user can input prediction demand description data on the client. The prediction demand description data is structured data and natural language data. The natural language data is a text or a voice. The structured data (that is, the prediction demand description data) has described the prediction target and the decision requirements corresponding to the prediction target in a preset format. If the natural language data (that is, the prediction demand description data) is voice, the voice is converted into text using a model trained based on speech recognition technology (ASR).

[0034] Through the model trained based on natural language technology, the text corresponding to the prediction demand description data is subjected to lexical, syntactic and semantic analysis, the entities and related information in the text are identified, and finally the prediction target and the decision demand corresponding to the prediction target are output. Among them, lexical analysis will decompose the words in the text, mark the part of speech, and clarify the basic attributes of each word; syntactic analysis focuses on the sentence structure, clarifies the grammatical relationship between words, and determines the structure such as subject, predicate, object, attributive, adverbial, complement, etc. Semantic analysis delves into the meaning of words and sentences; through these analyses, entities are identified, such as key things or concepts in the text, as well as related information, such as entity characteristics, restrictions, etc.

[0035] S2: Based on the prediction target, obtain the organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data and target weight data; Specifically, the sub-behavior transfer matrix of the organization is obtained from a preset storage space or a third-party application as the organizational sub-behavior transfer matrix. The knowledge graph data of the organization corresponding to the prediction target is obtained from a preset storage space or a third-party application as the organizational knowledge graph data. The sequence of indicator values ​​of the key basic indicators in the knowledge graph data of the organization corresponding to the prediction target in the future is obtained from the preset storage space or a third-party application as the indicator time series prediction data. The weight data of each key basic indicator in the knowledge graph data of the organization corresponding to the prediction target is obtained from the preset storage space or a third-party application as the target weight data.

[0036] The organizational sub-behavior transition matrix is ​​a matrix that can reflect the transition probabilities of all sub-behaviors corresponding to each decision-making behavior.

[0037] Organizational knowledge graph data covers various key entities, key basic indicators and entity relationships inside and outside the organization. Key entities include but are not limited to suppliers, customers and products. Key basic indicators are an indispensable part of organizational knowledge graph data. They are specific metrics used to quantitatively describe the organization's operating conditions, market position and other aspects. For example, key basic indicators include sales and market share. Entity relationships describe the logical connections between key entities. Logical connections include but are not limited to: the production relationship or supply relationship of a supplier to a certain product, the usage relationship or agency sales relationship of a customer to a product.

[0038] The indicator time series forecast data provides the indicator values ​​of each key basic indicator in the future period of time.

[0039] The target weight data clarifies the importance of each key basic indicator.

[0040] S3: Based on the decision-making requirements, determine the organizational sub-behavior probability matrix according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data; Specifically, based on the decision-making needs, the input vector or observation sequence of the hidden Markov model is generated according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data, so that the hidden Markov model predicts the sub-behavior, and the result of the hidden Markov model's prediction of the sub-behavior is used as the organizational sub-behavior probability matrix.

[0041] Hidden Markov Model (HMM) is a statistical model used to describe Markov processes with hidden unknown states. It contains a hidden state sequence and a corresponding observation sequence. There is a transition relationship between the two hidden states that satisfies the Markov property, that is, the next hidden state depends only on the current hidden state. Each hidden state will produce an observation value, and the entire model is determined by parameters such as the initial probability distribution, the state transition probability matrix, and the observation probability matrix. In many fields, such as speech recognition or natural language processing, hidden Markov models are widely used to infer hidden state sequences or estimate model parameters by learning observation sequences to solve practical classification or prediction problems.

[0042] Optionally, in step S3, when generating the input vector or observation sequence of the hidden Markov model, special attention is paid to multi-source data fusion.

[0043] Organization sub-behavior transfer matrix: represents the transition probability between each sub-behavior, corresponding to the transition of the hidden state.

[0044] Organizational knowledge graph data: constructed by knowledge representation and reasoning technology, it contains semantic information such as key entities, key basic indicators, and entity relationships (for example, supply relationships, agency relationships), providing structured support in the prediction and reasoning stages.

[0045] Indicator time series forecast data: Use time series modeling methods (based on Transformer's PatchTST) to predict the medium- and short-term indicator values ​​of key basic indicators.

[0046] Target weight data: indicates the importance of each key basic indicator in the entire forecast target or decision-making behavior. If drastic changes in the external environment (such as market sentiment, policy adjustments, etc.) are detected later, the weights in the target weight data can also be dynamically corrected.

[0047] Optionally, data fusion when generating the input vector or observation sequence of the hidden Markov model can be implemented in at least one more operational manner: (1) uniformly vectorizing or matrixing the above four types of data (i.e., the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data, and the target weight data); (2) Concatenate the matrix or vector dimensions and weight them based on the prior probability to obtain the importance of each source data; (3) The fused comprehensive feature vector / matrix is ​​used as the input of the subsequent prediction model (e.g., hidden Markov model) to more accurately capture the mutual influence between various information.

[0048] Optionally, in this step, to ensure that the use of HMM is more operational and professional, the following points need to be explained: • The initial parameters of the hidden Markov model can be obtained by combining supervised and unsupervised training (Baum-Welch algorithm) based on the existing sub-behavior sequence (that is, the observation sequence) to obtain the initial state distribution, transition probability matrix and observation probability matrix; • After training, evaluation indicators such as log-likelihood, AUC or F1 value can be used to measure the quality of the model's prediction of sub-behaviors; if some sub-behaviors occur rarely or in extreme cases, the probability distribution is corrected through smoothing methods (Laplace smoothing); F1 value is an indicator that comprehensively measures the precision and recall of the classification model. AUC (Area Under the Curve) is the area under the ROC curve, and its value range is between 0.5 and 1; when AUC=0.5, it means that the prediction effect of the model is equivalent to random guessing; the larger the AUC, the better the classification effect of the model. When AUC=1, it means that the model is a perfect classifier.

[0049] • An incremental learning mechanism can be introduced for scenarios that require online updates: When organizational behaviors change frequently or there are new data streams, the model parameters of the hidden Markov model are updated and iterated regularly to improve the timeliness and accuracy of sub-behavior predictions.

[0050] The organizational sub-behavior probability matrix is ​​a matrix that describes the probability of occurrence of each sub-behavior corresponding to the decision requirement, and can help analysts quantitatively evaluate the possibility of each sub-behavior in different environments or states.

[0051] S4: Determine the target decision result corresponding to the decision requirement according to the organizational sub-behavior probability matrix.

[0052] Specifically, a preset method is adopted to determine the target decision result corresponding to the decision demand according to the organizational sub-behavior probability matrix.

[0053] Optionally, the preset method uses a table lookup method.

[0054] Optionally, the preset method uses a multi-classification model. In complex decision-making scenarios, the multi-classification model is combined, and the decision requirement and the sub-behavior probability matrix are spliced ​​and input into the pre-trained result prediction model (multi-classifier based on deep learning), and the classification category corresponding to the predicted vector element is used as the final decision output. In this way, a more refined division of decision results can be achieved in a multi-dimensional space. To further enhance interpretability, while outputting the target decision result, an analysis of the contribution or importance of each sub-behavior in this decision is provided. By visually displaying the sub-behavior probability or the key path on the knowledge graph, users can understand the basis and credibility behind the decision.

[0055] Optionally, the preset method adopts a probability threshold, wherein the sub-behavior corresponding to the element in the organization sub-behavior probability matrix whose probability value is greater than the probability threshold is set to 1, and the sub-behavior corresponding to the element in the organization sub-behavior probability matrix whose probability value is less than or equal to the probability threshold is set to 0. After completing the setting of all elements of the organization sub-behavior probability matrix, a table lookup method is used to determine the target decision result corresponding to the decision requirement based on the organization sub-behavior probability matrix.

[0056] The target decision result describes the associated data between the decision behavior and the decision result. For example, if the decision behavior is the market share growth rate, then the decision result is to increase the market share growth rate.

[0057] The organizational knowledge graph data is derived from advanced knowledge representation and reasoning technology. It can represent various entities inside and outside the organization and their complex relationships in a structured and semantic way, thereby mining the associated information hidden behind the data; the indicator time series prediction data uses powerful time series analysis artificial intelligence technology to accurately capture the trends and laws of indicator values ​​over time, providing a dynamic basis for decision-making; the organizational sub-behavior transfer matrix is ​​based on behavioral modeling and analysis technology, which effectively characterizes the conversion pattern of sub-behaviors of the organization's decision-making behavior. During the entire process, this embodiment determines the organizational sub-behavior probability matrix by processing rich information sources (i.e., organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data, and target weight data) based on decision-making needs, and determines the target decision results corresponding to the decision-making needs based on the organizational sub-behavior probability matrix. It integrates various types of information sources, which are based on different artificial intelligence technologies. By integrating these information sources based on different artificial intelligence technologies, the organizational sub-behavior probability matrix is ​​determined, thereby achieving accurate locking of the most likely organizational sub-behavior pattern in a comprehensive and complex information network. Compared with traditional methods, this application can consider decision-related factors from a more three-dimensional and comprehensive perspective, greatly improving the accuracy of target decision results, and better meeting the requirements of modern organizations for decision accuracy in complex environments.

[0058] In one embodiment, the step of determining the organizational sub-behavior probability matrix based on the decision-making requirements, according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data, includes: S31: Based on the decision-making requirements, target data is generated according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data; Specifically, based on the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data, behavior probability prediction is performed for each sub-item prediction indicator corresponding to the decision requirement to obtain behavior probability prediction data, and an input vector or observation sequence of a hidden Markov model is formed based on the behavior probability prediction data, and the input vector or observation sequence is used as target data.

[0059] Optionally, step S31 pays special attention to multi-source data fusion. Step S31 data fusion step: weighted splicing of the organizational sub-behavior transfer matrix and indicator time series prediction data is performed according to the prior weight combined with the Attention mechanism, and at the same time, the entity relationship and sub-item prediction indicators extracted from the organizational knowledge graph data are embedded encoded, unified into the same vector space and then spliced, and finally the input vector or observation sequence of the hidden Markov model is obtained.

[0060] S32: Input the target data into a hidden Markov model to perform sub-behavior prediction, and obtain the organizational sub-behavior probability matrix.

[0061] Specifically, the target data is input into a hidden Markov model, and the hidden Markov model performs sub-behavior prediction, and finally the data output by the hidden Markov model is used as the organizational sub-behavior probability matrix.

[0062] Optionally, each row in the organizational sub-behavior probability matrix may correspond to a sub-behavior (that is, the sub-behavior corresponding to the decision requirement), and each column may be associated with different states or influencing factors. For example, for the behavior prediction of an organization, the rows may represent different types such as production sub-behaviors, marketing sub-behaviors, and R&D sub-behaviors. The columns may be related to time stages, market environment characteristics, etc. The values ​​in the matrix, i.e., probabilities, represent the probability of a specific sub-behavior occurring under a specific state or factor under the conditions of the hidden state and transition probability in the hidden Markov model.

[0063] Optionally, in order to ensure the model's ability to discern complex organizational behavior sequences, this embodiment also uses multi-hidden state or multi-perspective joint prediction (such as using competitor behavior or macroeconomic environment as additional hidden variables). The final output organizational sub-behavior probability matrix can be distributed in time series or in accordance with environmental or state characteristics, helping managers or analysts to examine the possibility of sub-behaviors from multiple perspectives.

[0064] This embodiment generates target data based on decision-making requirements by combining the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data. This method integrates multiple key information to make the target data comprehensive and targeted. The target data is input into the hidden Markov model for sub-behavior prediction. The model can effectively process sequence data with hidden states, thereby improving the accuracy of the organizational sub-behavior probability matrix, and further improving the accuracy of the target decision result.

[0065] In one embodiment, the step of generating target data based on the decision requirement, according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data, includes: S311: Obtaining sub-item prediction indicators corresponding to the decision-making requirements as a target indicator set; Specifically, according to the decision-making behavior in the decision-making needs, a table lookup method is used to determine the sub-item prediction indicators, and all the determined sub-item prediction indicators are used as the target indicator set.

[0066] Sub-item forecast indicators are specific indicators for forecasting and evaluating organizations from different dimensions. The selection range of sub-item forecast indicators includes but is not limited to: sales forecast indicators, market share change trend indicators, market sentiment scoring indicators, policy change risk probability indicators and supply chain delay risk indicators. Sales forecast indicators measure the sales amount in the future. Market share change trend indicators determine the trend of market share increase or decrease. Market sentiment scoring indicators quantify the emotional state of market participants. Policy change risk probability indicators assess the possibility of policy changes. Supply chain delay risk indicators estimate the delay risk of supply chain links.

[0067] S312: According to the organizational knowledge graph data, the indicator time series prediction data and the target weight data, predict the indicator value for each of the sub-item prediction indicators in the target indicator set to obtain a prediction result; Specifically, the organizational knowledge graph data, the indicator time series prediction data and the target weight data are input into a pre-trained target prediction model to predict the indicator value of each of the sub-item prediction indicators (the indicator value can be a specific numerical value or a text description data, for example, the text description data is a level description data), and all the data predicted for each of the sub-item prediction indicators are taken as a prediction result.

[0068] The pre-trained target prediction model is a pre-trained multi-classification model. The model structure and model training method of the target prediction model can be selected from the existing technology and will not be described in detail here.

[0069] S313: Perform data fusion according to the decision requirement, each prediction result corresponding to the target indicator set and the organizational sub-behavior transfer matrix to obtain the target data.

[0070] Specifically, a preset splicing symbol is used to splice the decision requirement, each prediction result corresponding to the target indicator set, and the organizational sub-behavior transfer matrix, and the spliced ​​data is used as the target data.

[0071] This embodiment can accurately locate the sub-item prediction indicators related to the decision by obtaining the sub-item prediction indicators corresponding to the decision-making needs to form a target indicator set, ensuring the pertinence and effectiveness of subsequent predictions. The behavior probability prediction is performed on each sub-item prediction indicator in the target indicator set based on the organizational knowledge graph data, the indicator time series prediction data and the target weight data. This prediction method combining multi-source data makes full use of the characteristics of different data. The organizational knowledge graph data contains rich structural relationships and semantic information within the organization. The indicator time series prediction data reflects the changing trend of key basic indicators over time, and the target weight data reflects the importance of each key basic indicator, thereby making the prediction result more scientific and accurate, thereby improving the accuracy of the decision result.

[0072] In one embodiment, the step of determining the target decision result corresponding to the decision requirement according to the organizational sub-behavior probability matrix includes: S41: Inputting the decision requirement and the organizational sub-behavior probability matrix into a pre-trained result prediction model for classification prediction to obtain a prediction vector; Specifically, the decision requirement and the organizational sub-behavior probability matrix are spliced ​​and input into a pre-trained result prediction model for classification prediction, and the predicted vector is used as a prediction vector.

[0073] The pre-trained result prediction model is a pre-trained multi-classification model. The model structure and model training method of the result prediction model can be selected from the existing technology and will not be described in detail here.

[0074] S42: generating a decision result of each decision behavior in the decision requirement according to the prediction vector; Specifically, a vector element with the largest value is extracted from the prediction vector for each decision behavior, and the classification category corresponding to the extracted vector element is used as the decision result of the decision behavior.

[0075] S43: Generate the target decision result corresponding to the decision requirement according to each of the decision results corresponding to the decision requirement.

[0076] The target decision result is the final conclusion or solution reached after a series of decision-making processes. It is formed by comprehensively considering the decision results of each decision-making behavior in the decision-making needs.

[0077] Specifically, based on the preset result generation template, the target decision result corresponding to the decision demand is generated according to each decision result corresponding to the decision demand. The preset result generation template provides a standardized framework for the entire process, like a precise mold, so that the generation of the target decision result has rules to follow. The preset result generation template can effectively integrate the various decision results corresponding to the decision demand, ensure that these results are reasonably organized, and avoid disorganized information. This process fully considers various factors related to decision needs, thereby improving the comprehensiveness and systematicness of the target decision results. In practical applications, this method can greatly improve the efficiency of decision result generation, improve the readability of target decision results, reduce errors that may occur due to manual integration, and lay a solid foundation for the accurate implementation of decisions.

[0078] This embodiment inputs the decision demand and the organizational sub-behavior probability matrix into a pre-trained result prediction model for classification prediction, and generates the target decision result corresponding to the decision demand based on the predicted data. It can effectively use the pre-trained result prediction model to process and analyze decision-related content, accurately derive results from decision-making behaviors, and then integrate to obtain complete target decision results, thereby providing a reliable basis for decision-making, improving the accuracy and efficiency of target decision results, and helping to make scientific and reasonable target decision results in complex decision scenarios involving multiple decision behaviors.

[0079] In one embodiment, the step of acquiring the organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data and target weight data based on the prediction target includes: S21: obtaining a sub-behavior transfer matrix corresponding to the tissue type of the target tissue corresponding to the prediction target as the tissue sub-behavior transfer matrix; The third-party platform or third-party organization can generate a sub-behavior transfer matrix for each organization type (such as retail, manufacturing, technology companies, etc.) based on the collected data. The present application can obtain the sub-behavior transfer matrix corresponding to the organization type of the target organization corresponding to the predicted target from the third-party platform or third-party organization, and use the obtained sub-behavior transfer matrix as the organization sub-behavior transfer matrix.

[0080] S22: Acquire knowledge graph data of the target organization corresponding to the prediction target as the organizational knowledge graph data, wherein the organizational knowledge graph data is used to describe key entities, key basic indicators, and entity relationships; The knowledge graph construction technology is used regularly to construct the knowledge graph based on the relevant data of the target organization, and the construction results are used as the knowledge graph data of the target organization. The knowledge graph data of the target organization is stored in the preset storage space.

[0081] Knowledge graph construction technology is a set of comprehensive technical means for creating knowledge graphs. It first uses information extraction technology, including entity extraction (entities in text can be identified through various methods such as rule-based and machine learning) and relationship extraction (determining the relationship between entities through pattern matching, statistical machine learning, deep learning, etc.). Then, knowledge fusion technology is used, such as entity alignment (matching the same entity with different representations based on attribute similarity or graph structure) and knowledge merging (integrating knowledge from multiple data sources and removing redundant contradictions). Finally, it also involves knowledge representation and storage technology, such as using semantic web-based representation methods (such as RDF triples) and graph database storage (such as Neo4j) to represent and store the extracted and fused knowledge in a suitable structure, thereby constructing a knowledge graph that contains rich entity and relationship knowledge and is easy to query and reason.

[0082] Specifically, the knowledge graph data of the target organization corresponding to the prediction target is obtained from a preset storage space, and the obtained knowledge graph data is used as the organization knowledge graph data.

[0083] S23: Obtain the target history data of the target organization corresponding to the prediction target, and predict the future indicator value according to the target history data and the key basic indicators and entity relationships in the organizational knowledge graph data to obtain the indicator time series prediction data; Target historical data refers to data related to the target organization that has occurred and been recorded in the past period of time. Target historical data reflects the status of the target organization in various aspects such as operations, activities, and development in the past.

[0084] Optionally, the target historical data includes quantifiable data.

[0085] Optionally, the target historical data includes quantifiable data and non-quantifiable data.

[0086] Quantifiable data refers to data that can be accurately measured, counted and expressed in numbers. These data can reflect certain characteristics or indicators of the target organization through clear numerical calculations. For example, quantifiable data include: annual turnover, net profit and debt-to-asset ratio, which can be described by specific numerical values.

[0087] Non-quantifiable data is difficult to accurately express with specific numbers, and relies more on description, feeling, evaluation, etc. to reflect certain aspects of the target organization. For example, non-quantifiable data includes the image of a brand in the minds of consumers. The image of a brand in the minds of consumers is a qualitative description based on consumer cognition and market impression. It cannot be simply measured with numbers, but it plays an important role in the market competitiveness of the brand.

[0088] Specifically, the target historical data of the target organization corresponding to the predicted target input by the user can be obtained, the target historical data of the target organization corresponding to the predicted target can be obtained from a preset storage space, and the target historical data of the target organization corresponding to the predicted target can also be obtained from a third-party application.

[0089] A future trend prediction model is used to predict future indicator values ​​based on the target historical data and the key basic indicators and entity relationships in the organizational knowledge graph data, and the predicted data is used as the indicator time series prediction data.

[0090] Optionally, the future trend prediction model uses a time series analysis model, such as an autoregressive moving average model (ARMA).

[0091] S24: Obtain indicator weight data of the target organization corresponding to the prediction target as target weight data.

[0092] Specifically, the indicator weight data of the target organization corresponding to the prediction target input by the user can be obtained, the indicator weight data of the target organization corresponding to the prediction target can be obtained from a preset storage space, and the indicator weight data of the target organization corresponding to the prediction target can also be obtained from a third-party application.

[0093] It is understandable that different organizations have different indicator weight data. For the same indicator weight data, the indicator weight data is updated periodically, that is, each time the indicator weight data is determined, the indicator weight data can be used for an organizational decision-making method based on time series prediction and knowledge graph for a period of time.

[0094] The target historical data of this embodiment carries information on the target organization's past operations, activities, development, and other aspects. These data include quantifiable data and non-quantifiable data, providing basic materials for prediction based on actual situations. The key basic indicators in the organizational knowledge graph data define the most core and critical measurement elements in the organization, and the entity relationships depict the interconnections between the key entities in the organization. By combining the target historical data with the organizational knowledge graph data, the laws and patterns hidden behind the data can be mined. Using these laws and patterns to build a prediction model can more accurately predict the future indicator values ​​of key basic indicators, thereby further improving the accuracy and efficiency of the target decision-making results.

[0095] In one embodiment, the step of predicting future indicator values ​​based on the target historical data and the key basic indicators and entity relationships in the organizational knowledge graph data to obtain the indicator time series prediction data includes: S221: Input the target historical data and the key basic indicators and entity relationships in the organizational knowledge graph data into a pre-trained sequence prediction model to predict the future indicator values ​​of the key basic indicators, and obtain the indicator time series prediction data, wherein the sequence prediction model is a model trained based on the Transformer architecture.

[0096] The sequence prediction model is a model trained by using the PatchTST model.

[0097] ‌PatchTST‌ (Patch Time series Transformer) is a model based on the Transformer architecture, specifically for long time series prediction tasks.

[0098] The Transformer architecture is a neural network architecture based on the self-attention mechanism. The Transformer architecture is mainly composed of an encoder and a decoder, and adopts an encoder-decoder architecture. The encoder is responsible for encoding the input sequence into a series of continuous representations, while the decoder decodes these representations into the target sequence.

[0099] Specifically, the target historical data and the key basic indicators and entity relationships in the organizational knowledge graph data are spliced ​​and input into a pre-trained sequence prediction model to predict the future indicator values ​​of the key basic indicators, and the predicted data is used as the indicator time series prediction data.

[0100] The powerful self-attention mechanism of the Transformer architecture can capture the long-term dependencies of time series data and achieve high-precision predictions. Therefore, this embodiment predicts the future indicator values ​​of key basic indicators through a model trained based on the Transformer architecture, thereby improving the accuracy of indicator time series prediction data, thereby further improving the accuracy and efficiency of target decision-making results.

[0101] In one embodiment, before the step of obtaining the indicator weight data of the target organization corresponding to the prediction target as the target weight data, the step further includes: S231: Acquire non-quantifiable data of the target tissue; Specifically, the non-quantifiable data of the target organization may be obtained from a user input, from a preset storage space, or from a third-party application.

[0102] Optional, non-quantifiable data includes, but is not limited to: market sentiment data and policy change data.

[0103] Market sentiment data is data that reflects the subjective feelings, attitudes and expectations of market participants (such as investors, consumers, etc.) towards market conditions. It is an intangible factor that is difficult to directly quantify but has an important impact on market behavior. Market sentiment can be optimistic, pessimistic or wait-and-see. It integrates people's cognition and psychological reactions to economic conditions, industry prospects, corporate performance and other aspects.

[0104] Policy change data refers to the information related to the introduction, adjustment or abolition of policies by government departments in various fields such as economy, society, environment, etc. These policies can include fiscal policy, monetary policy, industrial policy, regulatory policy, etc. Policy change data reflects the government's intention and decision on the development direction and regulation intensity of different fields.

[0105] S232: Input the non-quantifiable data into a pre-trained weight prediction model to predict the weight values ​​of the key basic indicators in the organizational knowledge graph data to obtain the indicator weight data of the target organization, wherein the weight prediction model is a model trained based on a generative language model.

[0106] The weight prediction model is a model obtained through training based on the LLM model.

[0107] The LLM model (Large Language Model) is an artificial intelligence model based on deep learning, designed to process and generate natural language text. By training on large-scale text data, LLM can understand and generate text similar to human language, and perform a variety of natural language processing tasks including text generation, translation, and sentiment analysis.

[0108] Indicator weight data describes the associated data between key basic indicators and weight values.

[0109] Specifically, first, these non-quantifiable data are processed according to the format required by the pre-trained weight prediction model, such as data standardization, encoding and other operations, so that they can be accepted by the model. Next, the key basic indicators in the organizational knowledge graph data of the target organization are obtained. After that, the processed non-quantifiable data and the obtained key basic indicators are spliced ​​and input into the model trained based on the generative language model. The model predicts the weight values ​​of the key basic indicators in the organizational knowledge graph data according to its own algorithm structure and training results. In this process, the model will analyze the potential relationship between non-quantifiable data and key basic indicators, and use the knowledge and pattern recognition capabilities obtained from the generative language model training to finally obtain the indicator weight data of the target organization. The indicator weight data of the target organization can reflect the relative importance of each key basic indicator in the target organization. ‌

[0110] This embodiment uses the powerful semantic understanding and data processing capabilities of the generative language model to explore the potential connections between non-quantifiable data and key basic indicators, so that the organization can more accurately consider the relative importance of each factor in the decision-making process, improve the scientificity and rationality of the decision, and also help to more comprehensively understand the interactive relationship between the internal structure of the organization and the external environment, thereby further improving the accuracy and efficiency of the target decision-making results.

[0111] In one embodiment, the step of obtaining a sub-behavior transfer matrix corresponding to a tissue type of a target tissue corresponding to the prediction target as the tissue sub-behavior transfer matrix includes: S211: acquiring a sub-behavior transfer matrix as an initial matrix according to the tissue type of the target tissue corresponding to the prediction target; Specifically, according to the tissue type of the target tissue corresponding to the prediction target, a sub-behavior transfer matrix is ​​obtained from the matrix library, and the obtained sub-behavior transfer matrix is ​​used as the initial matrix. In other words, the initial matrix is ​​the sub-behavior transfer matrix corresponding to the tissue type of the target tissue.

[0112] S212: Generate adjustment data according to the current organizational life cycle of the target organization; The current organizational life cycle of the target organization is predefined data. The current organizational life cycle of the target organization is the current organizational life cycle of the target organization.

[0113] The organizational life cycle refers to the various development stages that an organization goes through according to certain rules during the entire process from birth to death. The organizational life cycle reflects the changes in the organization's structure, functions, management model, resource requirements, development goals and other characteristics at different development stages.

[0114] Optionally, the organizational life cycle is any one of the start-up period, growth period, maturity period and decline period. It is understandable that the organizational life cycle can also be divided in other ways, which are not limited here.

[0115] Specifically, according to the current organizational life cycle of the target organization, the adjustment data is determined by using a table lookup method.

[0116] S213: Correcting the initial matrix according to the adjustment data to obtain the organizational sub-behavior transfer matrix corresponding to the prediction target.

[0117] Specifically, based on a preset adjustment formula and according to the adjustment data, the initial matrix is ​​corrected to obtain the organizational sub-behavior transfer matrix corresponding to the prediction target.

[0118] The adjustment formula is obtained by fitting multiple sets of historical data.

[0119] This embodiment obtains the initial matrix according to the organizational type, which can reflect the inherent behavioral pattern characteristics of different types of organizations, and lays a foundation for subsequent predictions that are consistent with organizational characteristics; considering the current organizational life cycle of the target organization, adjustment data is generated to correct the initial matrix, so that the corrected initial matrix can accurately reflect the special behavioral tendencies of the organization at different development stages. This embodiment can accurately construct an organizational sub-behavior transfer matrix that conforms to a specific organization in a specific period, greatly improving the accuracy of organizational behavior prediction, and further improving the accuracy and efficiency of target decision-making results.

[0120] In one embodiment, an electronic device is provided. The electronic device may be a computer device. The computer device may be a server 120. The internal structure diagram thereof may be as shown in FIG. Figure 3As shown. The electronic device includes a processor, a memory, a network interface and a database connected via a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the electronic device is used to communicate with an external client 110 via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a server 120 side of an organizational decision-making method based on time series prediction and knowledge graph.

[0121] In one embodiment, an electronic device is provided. The electronic device may be a computer device. The computer device may be a client 110. The internal structure diagram thereof may be as follows: Figure 4 As shown. The electronic device includes a processor, a memory, a network interface, a display screen and an input device connected via a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the electronic device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps on the client 110 side of a method for organizational decision-making based on time series prediction and knowledge graph.

[0122] In one embodiment, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the following steps are implemented: Obtaining a prediction target and a decision requirement corresponding to the prediction target; Based on the prediction target, obtain the organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data and target weight data; Based on the decision-making requirements, determining the organizational sub-behavior probability matrix according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data; According to the organizational sub-behavior probability matrix, a target decision result corresponding to the decision requirement is determined.

[0123] The organizational knowledge graph data is derived from advanced knowledge representation and reasoning technology. It can represent various entities inside and outside the organization and their complex relationships in a structured and semantic way, thereby mining the associated information hidden behind the data; the indicator time series prediction data uses powerful time series analysis artificial intelligence technology to accurately capture the trends and laws of indicator values ​​over time, providing a dynamic basis for decision-making; the organizational sub-behavior transfer matrix is ​​based on behavioral modeling and analysis technology to effectively characterize the conversion pattern of organizational behavior. During the entire process, this embodiment determines the organizational sub-behavior probability matrix by processing rich information sources (i.e., organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data, and target weight data) based on decision-making needs, and determines the target decision results corresponding to the decision-making needs based on the organizational sub-behavior probability matrix. It integrates various types of information sources, which are based on different artificial intelligence technologies. By integrating these information sources based on different artificial intelligence technologies, the organizational sub-behavior probability matrix is ​​determined, thereby achieving accurate locking of the most likely organizational sub-behavior pattern in a comprehensive and complex information network. Compared with traditional methods, this application can consider decision-related factors from a more three-dimensional and comprehensive perspective, greatly improving the accuracy of target decision results, and better meeting the requirements of modern organizations for decision accuracy in complex environments.

[0124] In one embodiment, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented: Obtaining a prediction target and a decision requirement corresponding to the prediction target; Based on the prediction target, obtain the organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data and target weight data; Based on the decision-making requirements, determining the organizational sub-behavior probability matrix according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data; According to the organizational sub-behavior probability matrix, a target decision result corresponding to the decision requirement is determined.

[0125] The organizational knowledge graph data is derived from advanced knowledge representation and reasoning technology. It can represent various entities inside and outside the organization and their complex relationships in a structured and semantic way, thereby mining the associated information hidden behind the data; the indicator time series prediction data uses powerful time series analysis artificial intelligence technology to accurately capture the trends and laws of indicator values ​​over time, providing a dynamic basis for decision-making; the organizational sub-behavior transfer matrix is ​​based on behavioral modeling and analysis technology to effectively characterize the conversion pattern of organizational behavior. During the entire process, this embodiment determines the organizational sub-behavior probability matrix by processing rich information sources (i.e., organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data, and target weight data) based on decision-making needs, and determines the target decision results corresponding to the decision-making needs based on the organizational sub-behavior probability matrix. It integrates various types of information sources, which are based on different artificial intelligence technologies. By integrating these information sources based on different artificial intelligence technologies, the organizational sub-behavior probability matrix is ​​determined, thereby achieving accurate locking of the most likely organizational sub-behavior pattern in a comprehensive and complex information network. Compared with traditional methods, this application can consider decision-related factors from a more three-dimensional and comprehensive perspective, greatly improving the accuracy of target decision results, and better meeting the requirements of modern organizations for decision accuracy in complex environments.

[0126] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or electronic device can refer to the relevant descriptions on the server 120 side and the client 110 side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0127] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0128] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0129] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. An organizational decision-making method based on time series prediction and knowledge graph, characterized in that: The method comprises: Obtaining a prediction target and a decision requirement corresponding to the prediction target; Based on the prediction target, obtain the organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data and target weight data; Based on the decision-making requirements, determining the organizational sub-behavior probability matrix according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data; According to the organizational sub-behavior probability matrix, a target decision result corresponding to the decision requirement is determined.

2. The organizational decision-making method based on time series prediction and knowledge graph according to claim 1 is characterized in that: The step of determining the organizational sub-behavior probability matrix based on the decision-making requirements, according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data, comprises: Based on the decision-making requirements, target data is generated according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data; The target data is input into a hidden Markov model to predict sub-behaviors, and the organizational sub-behavior probability matrix is ​​obtained.

3. The organizational decision-making method based on time series prediction and knowledge graph according to claim 2 is characterized in that: The step of generating target data based on the decision-making requirements and according to the organizational sub-behavior transfer matrix, the organizational knowledge graph data, the indicator time series prediction data and the target weight data comprises: Obtaining sub-item prediction indicators corresponding to the decision-making requirements as a target indicator set; According to the organizational knowledge graph data, the indicator time series prediction data and the target weight data, predict the indicator value for each of the sub-item prediction indicators in the target indicator set to obtain a prediction result; Data fusion is performed according to the decision-making requirements, the prediction results corresponding to the target indicator set, and the organizational sub-behavior transfer matrix to obtain the target data.

4. The organizational decision-making method based on time series prediction and knowledge graph according to claim 1 is characterized in that: The step of determining the target decision result corresponding to the decision requirement according to the organizational sub-behavior probability matrix includes: Inputting the decision requirement and the organizational sub-behavior probability matrix into a pre-trained result prediction model for classification prediction to obtain a prediction vector; Generate a decision result for each decision behavior in the decision requirement according to the prediction vector; According to each of the decision results corresponding to the decision requirement, the target decision result corresponding to the decision requirement is generated.

5. The organizational decision-making method based on time series prediction and knowledge graph according to claim 1 is characterized in that: The step of obtaining the organizational sub-behavior transfer matrix, organizational knowledge graph data, indicator time series prediction data and target weight data based on the prediction target includes: Obtaining a sub-behavior transfer matrix corresponding to the tissue type of the target tissue corresponding to the prediction target as the tissue sub-behavior transfer matrix; Acquire the knowledge graph data of the target organization corresponding to the prediction target as the organizational knowledge graph data, wherein the organizational knowledge graph data is used to describe key entities, key basic indicators and entity relationships; Obtaining target historical data of the target organization corresponding to the prediction target, predicting future indicator values ​​based on the target historical data and key basic indicators and entity relationships in the organizational knowledge graph data, and obtaining the indicator time series prediction data; Obtain indicator weight data of the target organization corresponding to the prediction target as target weight data.

6. The organizational decision-making method based on time series prediction and knowledge graph according to claim 5 is characterized in that: The step of predicting future indicator values ​​based on the target historical data and the key basic indicators and entity relationships in the organizational knowledge graph data to obtain the indicator time series prediction data includes: The target historical data and the key basic indicators and entity relationships in the organizational knowledge graph data are input into a pre-trained sequence prediction model to predict the future indicator values ​​of the key basic indicators to obtain the indicator time series prediction data, wherein the sequence prediction model is a model trained based on the Transformer architecture.

7. The organizational decision-making method based on time series prediction and knowledge graph according to claim 5 is characterized in that: Before the step of obtaining the indicator weight data of the target organization corresponding to the prediction target as the target weight data, the step further includes: Obtaining non-quantifiable data of the target tissue; The non-quantifiable data is input into a pre-trained weight prediction model to predict the weight values ​​of the key basic indicators in the organizational knowledge graph data to obtain the indicator weight data of the target organization, wherein the weight prediction model is a model trained based on a generative language model.

8. The organizational decision-making method based on time series prediction and knowledge graph according to claim 5 is characterized in that: The step of obtaining a sub-behavior transfer matrix corresponding to the tissue type of the target tissue corresponding to the prediction target as the tissue sub-behavior transfer matrix comprises: Acquire a sub-behavior transfer matrix as an initial matrix according to the tissue type of the target tissue corresponding to the prediction target; generating adjustment data according to the current organizational life cycle of the target organization; The initial matrix is ​​corrected according to the adjustment data to obtain the organizational sub-behavior transfer matrix corresponding to the prediction target.

9. An electronic device, characterized in that: The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the organizational decision-making method based on time series prediction and knowledge graph as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the organizational decision-making method based on time series prediction and knowledge graph as described in any one of claims 1 to 8.

Images