Enterprise Intelligent Decision Analysis System and Method Based on Big Data

Abstract

This application relates to the field of intelligent decision-making. Specifically, it discloses an enterprise intelligent decision-making analysis system and method based on big data. By obtaining the historical decision-making data and the to-be-decided data of the enterprise, where the historical decision-making data includes decision-making content and tags, and using data analysis and processing technologies based on deep learning to perform semantic analysis and mutual correlation and fusion on the historical decision-making data and the to-be-decided data, so as to intelligently obtain the type tags of the recommended decisions. In this way, the system can identify decision-making patterns and trends from historical data, provide data-based insights, help enterprises evaluate the potential impacts of different decision-making scenarios, improve the quality and accuracy of decision-making, and thus provide references for future decisions.

Description

Technical Field

[0001] This application relates to the field of intelligent analysis, and more specifically, to an enterprise intelligent decision-making analysis system and method based on big data. Background Art

[0002] With the development of information technology and the explosive growth of data volume, enterprises generate a large amount of data every day, including data from multiple channels such as internal systems, external sources, and social media. These data are valuable assets that can help enterprises better understand market trends, customer needs, competitor dynamics, etc., so as to support the decision-making and business development of enterprises.

[0003] However, in the face of such a large and complex amount of data, traditional decision-making methods and analysis tools often cannot effectively process and utilize this data. Secondly, traditional systems may lack in-depth analysis of historical data, resulting in decision-makers being unable to obtain the necessary insights to guide future decisions, thus affecting the comprehensiveness and rationality of decision-making results.

[0004] Therefore, there is an expectation for an enterprise intelligent decision-making analysis system based on big data. Summary of the Invention

[0005] To solve the above technical problems, this application is proposed. Embodiments of this application provide an enterprise intelligent decision-making analysis system and method based on big data, which obtain historical decision-making data and to-be-decision-making data of an enterprise, wherein the historical decision-making data includes decision content and labels, and use data analysis and processing technologies based on deep learning to perform semantic analysis and mutual correlation fusion on the historical decision-making data and the to-be-decision-making data, so as to intelligently obtain the type labels of recommended decisions. In this way, the system can identify decision-making patterns and trends from historical data, provide data-based insights, help enterprises evaluate the potential impacts of different decision-making scenarios, improve the quality and accuracy of decision-making, and thus provide references for future decisions.

[0006] According to one aspect of this application, an enterprise intelligent decision-making analysis system based on big data is provided, which includes:

[0007] A decision-making data acquisition module, configured to acquire historical decision-making data and to-be-decision-making data of an enterprise, wherein the historical decision-making data includes decision content and labels;

[0008] A historical decision-making data semantic understanding module, configured to perform historical decision-making data semantic understanding on each decision example in the historical decision-making data to obtain a plurality of historical decision-making data semantic understanding feature vectors;

[0009] A historical decision data semantic interaction enhancement module, which is used to perform historical decision data semantic interaction enhancement on the historical decision data semantic understanding feature vectors of the multiple historical decision data after arranging them into a historical decision data semantic understanding associated interaction feature matrix to obtain a historical decision data semantic understanding associated interaction enhancement matrix;

[0010] A to-be-decided data semantic understanding module, which is used to perform to-be-decided data semantic understanding on the to-be-decided data to obtain a to-be-decided data semantic understanding vector;

[0011] A feature fusion module, which is used to fuse the to-be-decided data semantic understanding vector and the historical decision data semantic understanding associated interaction enhancement matrix to obtain a to-be-decided data semantic query representation vector;

[0012] A recommendation result generation module, which is used to obtain a recommendation result based on the to-be-decided data semantic query representation vector.

[0013] According to another aspect of the present application, there is provided a big data-based enterprise intelligent decision analysis method, which includes:

[0014] Obtain the historical decision data and to-be-decided data of an enterprise, wherein the historical decision data includes decision content and labels;

[0015] Perform historical decision data semantic understanding on each decision example in the historical decision data to obtain multiple historical decision data semantic understanding feature vectors;

[0016] Perform historical decision data semantic interaction enhancement on the historical decision data semantic understanding feature vectors of the multiple historical decision data after arranging them into a historical decision data semantic understanding associated interaction feature matrix to obtain a historical decision data semantic understanding associated interaction enhancement matrix;

[0017] Perform to-be-decided data semantic understanding on the to-be-decided data to obtain a to-be-decided data semantic understanding vector;

[0018] Fuse the to-be-decided data semantic understanding vector and the historical decision data semantic understanding associated interaction enhancement matrix to obtain a to-be-decided data semantic query representation vector;

[0019] Obtain a recommendation result based on the to-be-decided data semantic query representation vector.

[0020] Compared with the prior art, an enterprise intelligent decision-making analysis system and method based on big data provided by the present application obtains historical decision-making data and to-be-decision data of an enterprise, wherein the historical decision-making data includes decision content and tags, and uses data analysis and processing technologies based on deep learning to perform semantic analysis and mutual correlation and fusion on the historical decision-making data and the to-be-decision data, so as to intelligently obtain the type tags of recommended decisions. In this way, the system can identify decision-making patterns and trends from historical data, provide data-based insights, help enterprises evaluate the potential impacts of different decision-making schemes, improve the quality and accuracy of decision-making, and thus provide references for future decisions. BRIEF DESCRIPTION OF THE DRAWINGS

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

[0022] Figure 1 It is a block diagram of an enterprise intelligent decision-making analysis system based on big data according to an embodiment of the present application.

[0023] Figure 2 It is a schematic diagram of the architecture of an enterprise intelligent decision-making analysis system based on big data according to an embodiment of the present application.

[0024] Figure 3 It is a block diagram of a historical decision-making data semantic interaction enhancement module in an enterprise intelligent decision-making analysis system based on big data according to an embodiment of the present application.

[0025] Figure 4 It is a flowchart of an enterprise intelligent decision-making analysis method based on big data according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although some embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand the present disclosure. It should be understood that the accompanying drawings and embodiments of the present disclosure are only for exemplary purposes and are not used to limit the protection scope of the present disclosure.

[0027] It should be understood that the various steps recited in the method embodiments of the present disclosure may be executed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this regard.

[0028] It should be noted that the terms "first", "second", and "third" involved in the embodiments of the present application are only used to distinguish similar objects and do not represent a specific order for the objects. It can be understood that "first", "second", and "third" can be interchanged in a specific order or sequence when permitted. It should be understood that the objects distinguished by "first", "second", and "third" can be interchanged appropriately so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.

[0029] As shown in the present disclosure and the claims, unless the context clearly indicates otherwise, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0030] The various exemplary embodiments, features, and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the drawings denote elements having the same or similar functions. Although various aspects of the embodiments are shown in the drawings, the drawings do not have to be drawn to scale unless otherwise specified.

[0031] With the development of information technology and the explosive growth of data volume, enterprises generate a large amount of data every day, including data from multiple channels such as internal systems, external sources, and social media. These data are valuable assets that can help enterprises better understand market trends, customer needs, competitor dynamics, etc., so as to support the decision-making and business development of enterprises.

[0032] However, in the face of such a large and complex amount of data, traditional decision-making methods and analysis tools often cannot effectively process and utilize this data. Secondly, traditional systems may lack in-depth analysis of historical data, resulting in decision-makers being unable to obtain the necessary insights to guide future decisions, thus affecting the comprehensiveness and rationality of decision-making results.

[0033] Therefore, there is a need for an enterprise intelligent decision-making analysis system based on big data. By obtaining the historical decision-making data and the to-be-decided data of an enterprise, where the historical decision-making data includes decision content and tags, and by using data analysis and processing techniques based on deep learning to perform semantic analysis and mutual correlation fusion on the historical decision-making data and the to-be-decided data, the type tags of recommended decisions can be obtained intelligently. In this way, the system can identify decision-making patterns and trends from historical data, provide data-based insights, help enterprises evaluate the potential impacts of different decision-making scenarios, improve the quality and accuracy of decision-making, and thus provide references for future decisions.

[0034] Figure 1 It is a block diagram of an enterprise intelligent decision-making analysis system based on big data according to an embodiment of the present application. Figure 2 It is a schematic diagram of the architecture of an enterprise intelligent decision-making analysis system based on big data according to an embodiment of the present application. As Figure 1 and Figure 2 shown, the enterprise intelligent decision-making analysis system 100 based on big data according to an embodiment of the present application includes: a decision-making data acquisition module 110, configured to acquire the historical decision-making data and the to-be-decided data of an enterprise, where the historical decision-making data includes decision content and tags; a historical decision-making data semantic understanding module 120, configured to perform historical decision-making data semantic understanding on each decision example in the historical decision-making data to obtain a plurality of historical decision-making data semantic understanding feature vectors; a historical decision-making data semantic interaction enhancement module 130, configured to perform historical decision-making data semantic interaction enhancement on the historical decision-making data semantic understanding correlation interaction feature matrix formed by arranging the plurality of historical decision-making data semantic understanding feature vectors to obtain a historical decision-making data semantic understanding correlation interaction enhancement matrix; a to-be-decided data semantic understanding module 140, configured to perform to-be-decided data semantic understanding on the to-be-decided data to obtain a to-be-decided data semantic understanding vector; a feature fusion module 150, configured to fuse the to-be-decided data semantic understanding vector and the historical decision-making data semantic understanding correlation interaction enhancement matrix to obtain a to-be-decided data semantic query representation vector; and a recommendation result generation module 160, configured to obtain a recommendation result based on the to-be-decided data semantic query representation vector.

[0035] In the embodiment of the present application, the decision data acquisition module 110 is configured to acquire the historical decision data and the to-be-decided data of an enterprise, where the historical decision data includes decision content and tags. It should be understood that the decision content refers to the specific information involved in the decision itself, including the background, objectives, alternative plans, expected results, relevant data, and analysis of the decision. The tags are the classifications or markings of the historical decision results, such as success, failure, high risk, high return, etc. These tags help to classify and evaluate the decision results. Based on this, in the technical solution of the present application, the historical decision data and the to-be-decided data of the enterprise are acquired, where the historical decision data includes decision content and tags. That is, by understanding the background and content of previous decisions, the system can provide data-based recommendations and suggestions, and establish a data-based decision-making culture, so as to improve the scientificity and objectivity of decision-making, thereby assisting decision-makers to make more informed choices and learn and improve future decisions.

[0036] In the embodiment of the present application, the historical decision data semantic understanding module 120 is configured to perform historical decision data semantic understanding on each decision example in the historical decision data to obtain a plurality of historical decision data semantic understanding feature vectors. Specifically, in the embodiment of the present application, the historical decision data semantic understanding module is configured to: input each decision example in the historical decision data into a historical decision data semantic understanding device based on a gated recurrent unit to obtain the plurality of historical decision data semantic understanding feature vectors. Correspondingly, considering that there are mutual semantic associations between the contexts of each decision example in the historical decision data, and the semantic associations and influences between different word distances are different. Therefore, in order to better understand the context and influencing factors behind the decision, in the technical solution of the present application, each decision example in the historical decision data is input into a historical decision data semantic understanding device based on a gated recurrent unit to obtain the plurality of historical decision data semantic understanding feature vectors. It is worth mentioning that the gated recurrent unit can process sequential data, understand the semantic information in each decision text, and capture the internal connections between words and phrases. And the gated recurrent unit is designed to remember long-term dependencies, which helps to understand the long-term factors in the decision-making process, reflects the complexity of the semantic associations in the decision content, and provides more accurate and in-depth decision analysis and intelligent decision support for the enterprise.

[0037] In the embodiment of the present application, the historical decision data semantic interaction enhancement module 130 is configured to perform historical decision data semantic interaction enhancement on the historical decision data semantic understanding associated interaction feature matrix after arranging the plurality of historical decision data semantic understanding feature vectors to obtain a historical decision data semantic understanding associated interaction enhancement matrix. Figure 3It is a block diagram of a historical decision data semantic interaction enhancement module in an enterprise intelligent decision-making analysis system based on big data according to an embodiment of the present application. Specifically, in the embodiment of the present application, as Figure 3 shown, the historical decision data semantic interaction enhancement module 130 includes: a semantic feature arrangement unit 131 for arranging the multiple historical decision data semantic understanding feature vectors into the historical decision data semantic understanding associated interaction feature matrix; and a semantic interaction unit 132 for inputting the historical decision data semantic understanding associated interaction feature matrix into the historical decision data semantic interaction enhancement module based on a bidirectional interaction attention mechanism to obtain the historical decision data semantic understanding associated interaction enhancement matrix.

[0038] Specifically, the semantic feature arrangement unit 131 is used to arrange the multiple historical decision data semantic understanding feature vectors into the historical decision data semantic understanding associated interaction feature matrix. It should be understood that considering that each historical decision data semantic understanding feature vector in the multiple historical decision data semantic understanding feature vectors respectively expresses the semantic information of a historical decision case, and there are mutual associations and influences between different historical decision examples. Based on this, in order to better understand and analyze the implicit association relationships contained in each historical decision data semantic understanding feature vector, in the technical solution of the present application, the multiple historical decision data semantic understanding feature vectors are arranged into the historical decision data semantic understanding associated interaction feature matrix. Specifically, by arranging multiple vectors in the form of a matrix, it is easier to identify patterns and trends in the data, such as common decision-making paths or results, which helps to analyze the semantic relevance between different decision-making features, discover the potential semantic association relationships between them, and thus provide more comprehensive and in-depth decision-making support for enterprises.

[0039] Specifically, the semantic interaction unit 132 is configured to input the historical decision data semantic understanding associated interaction feature matrix into the historical decision data semantic interaction enhancement module based on the bidirectional interaction attention mechanism to obtain the historical decision data semantic understanding associated interaction enhancement matrix. In particular, the historical decision data semantic interaction enhancement module based on the bidirectional interaction attention mechanism is the historical decision data semantic interaction enhancement module based on the bidirectional attention mechanism. Correspondingly, considering that different historical decision data semantic understanding associated interaction features in the historical decision data semantic understanding associated interaction feature matrix have different importance and influence. Therefore, in order to capture and mine key and important historical decision data semantic understanding associated interaction feature information from the historical decision data semantic understanding associated interaction feature matrix, reduce the interference of redundant information and irrelevant features, in the technical solution of this application, the historical decision data semantic understanding associated interaction feature matrix is input into the historical decision data semantic interaction enhancement module based on the bidirectional interaction attention mechanism to obtain the historical decision data semantic understanding associated interaction enhancement matrix. It is worth mentioning that the bidirectional interaction attention mechanism can simultaneously consider the context information in the horizontal and vertical directions of each element in the input data, providing a more comprehensive semantic understanding. Specifically, the historical decision data semantic interaction enhancement module based on the bidirectional interaction attention mechanism introduces the attention mechanism, and the model can identify and strengthen those historical decision data semantic understanding associated interaction features that are crucial for understanding the decision data, while ignoring irrelevant information, and assigns dynamic weights to each element in the historical decision data semantic understanding associated interaction feature matrix. These weights indicate the importance and key nature of each element for the current task, so as to generate an enhanced historical decision data semantic understanding associated interaction enhancement matrix, which is used to provide a richer and more accurate data representation.

[0040] In the embodiment of the present application, the to-be-decided data semantic understanding module 140 is configured to perform to-be-decided data semantic understanding on the to-be-decided data to obtain a to-be-decided data semantic understanding vector. Specifically, in the embodiment of the present application, the to-be-decided data semantic understanding module is configured to: input the to-be-decided data into a to-be-decided data semantic understander based on a gated recurrent unit to obtain the to-be-decided data semantic understanding vector. Correspondingly, considering that a large amount of text semantic information is included in the to-be-decided data, and the gated recurrent unit is a variant of the recurrent neural network suitable for processing sequence data, it can effectively extract key features in the to-be-decided data. Based on this, in the technical solution of the present application, the to-be-decided data is input into a to-be-decided data semantic understander based on a gated recurrent unit to obtain the to-be-decided data semantic understanding vector. It should be understood that the to-be-decided data often has a sequential nature, such as time series data or data with a certain order relationship. By inputting the to-be-decided data into the to-be-decided data semantic understander based on the gated recurrent unit, the excellent sequence modeling ability of the gated recurrent unit can be utilized to better learn and capture the dependency relationships, sequence semantic information, and patterns in the to-be-decided data, thereby generating a more representative to-be-decided data semantic understanding vector and providing richer information for subsequent decision-making.

[0041] In the embodiment of the present application, the feature fusion module 150 is used to fuse the semantic understanding vector of the data to be decision - made and the semantic understanding associated interaction enhancement matrix of the historical decision - making data to obtain the semantic query representation vector of the data to be decision - made. Specifically, in the embodiment of the present application, the feature fusion module is used to: use the semantic understanding vector of the data to be decision - made as a query vector, and calculate the vector product between the query vector and the semantic understanding associated interaction enhancement matrix of the historical decision - making data to obtain the semantic query representation vector of the data to be decision - made. Correspondingly, considering that the semantic understanding vector of the data to be decision - made contains the representation obtained after semantic understanding of the data to be decision - made, which includes important semantic information and features of the data, this vector can be regarded as an abstract representation of the data to be decision - made for subsequent decision - making processes. The semantic understanding associated interaction enhancement matrix of the historical decision - making data contains the semantic understanding information of the historical decision - making data, and also considers the association and interaction between historical data. Therefore, in the technical solution of the present application, the semantic understanding vector of the data to be decision - made is used as a query vector, and the vector product between the query vector and the semantic understanding associated interaction enhancement matrix of the historical decision - making data is calculated to obtain the semantic query representation vector of the data to be decision - made. In detail, by calculating the vector product between the query vector and the semantic understanding associated interaction enhancement matrix of the historical decision - making data, the semantic association and information interaction between the query vector and the historical decision - making data can be realized, helping the model to better understand the semantic similarity and correlation between data, so as to obtain a more comprehensive semantic query representation vector of the data to be decision - made, thereby improving the accuracy and comprehensiveness of decision - making.

[0042] In the embodiment of the present application, the recommendation result generation module 160 is used to obtain a recommendation result based on the semantic query representation vector of the data to be decision - made. Specifically, in the embodiment of the present application, the recommendation result generation module is used to: input the semantic query representation vector of the data to be decision - made into a decision - label recommender based on a classifier to obtain the type label of the recommended decision. That is, through the classification processing of the semantic query representation vector of the data to be decision - made, the type label of the recommended decision can be intelligently obtained. In this way, the system can identify decision - making patterns and trends from historical data, provide data - based insights, help enterprises evaluate the potential impacts of different decision - making scenarios, improve the quality and accuracy of decision - making, and thus provide a reference for future decision - making.

[0043] In a preferred embodiment, the recommendation result generation module includes: a feature normalization enhancement unit, configured to perform target-domain-based class label regression normalization enhancement on the semantic query representation vector of the data to be decision-making to obtain an optimization factor; a feature weighting unit, configured to weight the semantic query representation vector of the data to be decision-making with the optimization factor as a weight to obtain an optimized semantic query representation vector of the data to be decision-making; and a decision label recommendation unit, configured to input the optimized semantic query representation vector of the data to be decision-making into a decision label recommender based on a classifier to obtain a type label for the recommended decision.

[0044] Specifically, the feature normalization enhancement unit is configured to perform target-domain-based class label regression normalization enhancement on the semantic query representation vector of the data to be decision-making to obtain an optimization factor. In particular, in the technical solution of the present application, the historical decision-making data and the data to be decision-making may come from different time points and business environments, resulting in differences in the inherent features of the data. The semantic understanding unit based on the gated recurrent unit may have biases in understanding and representing different data sets, especially if the historical data and the data to be decision-making are different in language use and expression; the bidirectional interactive attention mechanism may cause the distribution of the enhanced matrix in the feature space to be inconsistent due to the correlation between different historical decision-making examples when enhancing the historical decision-making data. As a query vector, the matching degree between the semantic understanding vector of the data to be decision-making and the historical decision-making data may vary due to data differences, affecting the consistency of the query representation vector. When calculating the vector product between the query vector and the enhanced matrix, the feature distribution of the semantic query representation vector of the data to be decision-making may be biased towards a specific subspace due to the dominant role of some row vectors in the matrix. If the feature distribution of the semantic query representation vector of the data to be decision-making is inconsistent, it may be difficult for the decision label recommender based on the classifier to capture the stable mapping relationship between features and type labels during training, resulting in a decrease in classification accuracy. The inconsistency of the feature distribution may lead to a high degree of uncertainty when recommending decision type labels, affecting the reliability of decision support. Based on this, target-domain-based class label regression normalization enhancement is performed on the semantic query representation vector of the data to be decision-making.

[0045] More specifically, the feature normalization enhancement unit is configured to: multiply the semantic query representation vector of the data to be decision - made by the classification weight matrix of the decision - label recommender based on the classifier to obtain a semantic query representation weight feature vector of the data to be decision - made; concatenate the semantic query representation vector of the data to be decision - made and the semantic query representation weight feature vector of the data to be decision - made to obtain a semantic query representation - weight information joint vector of the data to be decision - made; pass the semantic query representation - weight information joint vector through a first fully - connected layer and then through a sigmoid function to obtain a first activation output value; add the semantic query representation vector of the data to be decision - made and the semantic query representation weight feature vector of the data to be decision - made element - by - element to obtain a semantic query representation - weight information summation vector of the data to be decision - made; pass the semantic query representation - weight information summation vector through a second fully - connected layer and then through a sigmoid function to obtain a second activation output value; calculate the mean of the first activation output value and the second activation output value, and subtract the mean from 1 to obtain a first weighting coefficient; calculate the mean of the first activation output value and the second activation output value to obtain a second weighting coefficient; calculate the natural exponential function value with the eigenvalue of the semantic query representation weight feature vector of the data to be decision - made as the power to obtain a first exponential semantic query representation weight feature vector of the data to be decision - made; calculate the natural exponential function value with the eigenvalue of the semantic query representation vector of the data to be decision - made as the power to obtain a second exponential semantic query representation vector of the data to be decision - made; and, based on the first weighting coefficient and the second weighting coefficient, calculate the weighted sum of the first exponential semantic query representation weight feature vector of the data to be decision - made and the second exponential semantic query representation vector of the data to be decision - made to obtain a weighted - sum feature vector, and calculate the two - norm of the weighted - sum feature vector to obtain the optimization factor.

[0046] In the above - mentioned preferred embodiment, specifically, the feature normalization enhancement unit is configured to: perform target - domain class - label regression normalization enhancement on the semantic query representation vector of the data to be decision - made according to the following enhancement formula to obtain the optimization factor; where the enhancement formula is:

[0047]

[0048] where, v c represents the semantic query representation vector of the data to be decision - made, M represents the classification weight matrix of the decision - label recommender based on the classifier, represents matrix multiplication, represents element - by - element addition, concat represents the concatenation function, W 1 represents the first weight matrix of the first fully - connected layer, b 1 represents the first bias vector of the first fully - connected layer, t 1 represents the first activation output value, W2 The second weight matrix representing the second fully connected layer, b 2 The second bias vector representing the second fully connected layer, t 2 Represents the second activation output value, sigmoid represents the sigmoid function, ‖·‖ 2 Represents the two-norm of the vector, and α represents the optimization factor.

[0049] That is, in the technical solution of this application, the manifold geometric consistency of the overall feature distribution of the semantic query representation vector of the data to be decision-making is poor, resulting in relatively poor class label regression normality with respect to the classifier-based decision label recommender when it passes through the classifier-based decision label recommender, affecting the accuracy of the classification result. Therefore, in the technical solution of this application, the class label regression normality of the semantic query representation vector of the data to be decision-making is enhanced based on the target domain. It performs auxiliary analysis of the feature attribute layout of the semantic query representation vector of the data to be decision-making with the classification weight matrix of the classifier-based decision label recommender, and performs backward correlation prediction with the statistical norm of the differential features in the class high-dimensional space between the auxiliary analysis result and the original semantic query representation vector of the data to be decision-making, so as to improve the description significance of the improved semantic query representation vector of the data to be decision-making for the class probability of the predetermined class of the classifier-based decision label recommender. In this way, maintaining the geometric consistency of the semantic query representation vector of the data to be decision-making during the migration transformation process to the class label regression domain of the classifier-based decision label recommender not only helps to reduce the regression bias in the classification process, but also can improve the generalization ability of the classifier-based decision label recommender for new data, and finally achieve a more accurate classification result.

[0050] In summary, the enterprise intelligent decision-making analysis system 100 based on big data according to the embodiments of this application is elucidated. It obtains the historical decision-making data and the data to be decision-making of the enterprise. Among them, the historical decision-making data includes decision content and labels, and uses deep learning-based data analysis and processing technologies to perform semantic analysis and mutual correlation fusion on the historical decision-making data and the data to be decision-making, so as to intelligently obtain the type labels of the recommended decisions. In this way, the system can identify decision-making patterns and trends from historical data, provide data-based insights, help enterprises evaluate the potential impacts of different decision-making scenarios, improve the quality and accuracy of decision-making, and thus provide references for future decision-making.

[0051] Figure 4 Is a flowchart of the enterprise intelligent decision-making analysis method based on big data according to the embodiments of this application. As Figure 4As shown, the big data-based enterprise intelligent decision-making analysis method according to an embodiment of the present application includes: S110, obtaining historical decision-making data and to-be-decision data of an enterprise, wherein the historical decision-making data includes decision contents and tags; S120, performing semantic understanding of historical decision-making data on each decision example in the historical decision-making data to obtain a plurality of historical decision-making data semantic understanding feature vectors; S130, arranging the plurality of historical decision-making data semantic understanding feature vectors into a historical decision-making data semantic understanding associated interaction feature matrix and then performing historical decision-making data semantic interaction enhancement to obtain a historical decision-making data semantic understanding associated interaction enhancement matrix; S140, performing semantic understanding of the to-be-decision data to obtain a to-be-decision data semantic understanding vector; S150, fusing the to-be-decision data semantic understanding vector and the historical decision-making data semantic understanding associated interaction enhancement matrix to obtain a to-be-decision data semantic query representation vector; and S160, obtaining a recommendation result based on the to-be-decision data semantic query representation vector.

[0052] Here, those skilled in the art can understand that the specific operations of each step in the above big data-based enterprise intelligent decision-making analysis method have been introduced in detail in the description of the big data-based enterprise intelligent decision-making analysis system above with reference to Figures 1 to 3 and thus, the repeated description thereof will be omitted.

[0053] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired manner (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that includes one or more integrated available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as digital video discs (DVDs)), or semiconductor media (such as solid state disks (SSDs)), etc.

[0054] Those of ordinary skill in the art will realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. A professional technician can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of this application.

[0055] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of devices or units can be in electrical, mechanical, or other forms.

[0056] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or may be distributed over multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0057] In addition, each functional unit in various embodiments of the present application may be integrated in a processing unit, may exist separately as individual physical units, or two or more units may be integrated in one unit.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit of the technical solutions of the present invention.

Claims

1. An enterprise intelligent decision-making analysis system based on big data, characterized in that: include: A decision data acquisition module is used to acquire the enterprise's historical decision data and pending decision data, wherein the historical decision data includes decision content and labels; A historical decision data semantic understanding module, used for performing historical decision data semantic understanding on each decision example in the historical decision data to obtain a plurality of historical decision data semantic understanding feature vectors; A historical decision data semantic interaction enhancement module, used to perform historical decision data semantic interaction enhancement after arranging the plurality of historical decision data semantic understanding feature vectors into a historical decision data semantic understanding associated interaction feature matrix to obtain a historical decision data semantic understanding associated interaction enhancement matrix; A semantic understanding module for data to be decided, used for performing semantic understanding of the data to be decided on the data to be decided to obtain a semantic understanding vector of the data to be decided; A feature fusion module, used for fusing the semantic understanding vector of the data to be decided and the semantic understanding association interaction enhancement matrix of the historical decision data to obtain a semantic query representation vector of the data to be decided; A recommendation result generation module, used to obtain a recommendation result based on the semantic query representation vector of the data to be decided; The historical decision data semantic interaction enhancement module includes: A semantic feature arrangement unit, used for arranging the plurality of historical decision data semantic understanding feature vectors into the historical decision data semantic understanding associated interaction feature matrix; The semantic interaction unit is used to input the historical decision data semantic understanding associated interaction feature matrix into the historical decision data semantic interaction enhancement module based on the bidirectional interactive attention mechanism to obtain the historical decision data semantic understanding associated interaction enhancement matrix.

2. The enterprise intelligent decision-making analysis system based on big data according to claim 1 is characterized in that: The historical decision data semantic understanding module is used to: Each decision example in the historical decision data is input into a historical decision data semantic understander based on a gated recurrent unit to obtain the plurality of historical decision data semantic understanding feature vectors.

3. The enterprise intelligent decision-making analysis system based on big data according to claim 2 is characterized in that: The semantic understanding module of the data to be decided is used to: The data to be decided is input into a semantic understander of the data to be decided based on a gated recurrent unit to obtain a semantic understanding vector of the data to be decided.

4. The enterprise intelligent decision-making analysis system based on big data according to claim 3 is characterized in that: The feature fusion module is used to: The semantic understanding vector of the data to be decided is used as a query vector, and the vector product between the query vector and the historical decision data semantic understanding associated interaction enhancement matrix is ​​calculated to obtain the semantic query representation vector of the data to be decided.

5. The enterprise intelligent decision-making analysis system based on big data according to claim 4 is characterized in that: The recommendation result generating module is used to: The semantic query representation vector of the data to be decided is input into a decision label recommender based on a classifier to obtain a type label of the recommended decision.

6. The enterprise intelligent decision-making analysis system based on big data according to claim 4 is characterized in that: The recommendation result generating module comprises: A feature normative enhancement unit, used for performing a class label regression normative enhancement based on a target domain on the semantic query representation vector of the decision-making data to obtain an optimization factor; A feature weighting unit, used to weight the semantic query representation vector of the data to be decided by taking the optimization factor as a weight to obtain an optimized semantic query representation vector of the data to be decided; The decision label recommendation unit is used to input the optimized semantic query representation vector of the data to be decided into a decision label recommender based on a classifier to obtain a type label of the recommended decision.

7. The enterprise intelligent decision-making analysis system based on big data according to claim 6 is characterized in that: The feature specification enhancement unit is used to: Multiplying the semantic query representation vector of the data to be decided by the classification weight matrix of the decision label recommender based on the classifier to obtain a semantic query representation weight feature vector of the data to be decided; Cascading the semantic query representation vector of the data to be decided and the semantic query representation weight feature vector of the data to be decided to obtain a semantic query representation-weight information joint vector of the data to be decided; The semantic query representation-weight information joint vector of the decision data passes through the first fully connected layer and then passes through the sigmoid function to obtain a first activation output value; Add the semantic query representation vector of the data to be decided and the weight feature vector of the semantic query representation of the data to be decided by position to obtain a semantic query representation-weight information sum vector of the data to be decided; The semantic query representation of the data to be decided-weight information sum vector passes through the second fully connected layer and then passes through the sigmoid function to obtain a second activation output value; Calculating a mean of the first activation output value and the second activation output value, and subtracting the mean from 1 to obtain a first weighting coefficient; Calculating a mean of the first activation output value and the second activation output value to obtain a second weighting coefficient; Calculate the value of a natural exponential function by raising the eigenvalue of the semantic query representation weight feature vector of the data to be decided to a power to obtain a first exponential semantic query representation weight feature vector of the data to be decided; Calculate the value of a natural exponential function by using the eigenvalue of the semantic query representation vector of the data to be decided as a power to obtain a second exponential semantic query representation vector of the data to be decided; Based on the first weighting coefficient and the second weighting coefficient, the weighted sum of the first index semantic query representation weight feature vector of the data to be decided and the second index semantic query representation vector of the data to be decided is calculated to obtain a weighted sum feature vector, and the bi-norm of the weighted sum feature vector is calculated to obtain the optimization factor.

8. A method for enterprise intelligent decision-making analysis based on big data, characterized in that: include: Acquire the enterprise's historical decision data and pending decision data, wherein the historical decision data includes decision content and labels; Performing historical decision data semantic understanding on each decision example in the historical decision data to obtain a plurality of historical decision data semantic understanding feature vectors; After arranging the plurality of historical decision data semantic understanding feature vectors into a historical decision data semantic understanding associated interaction feature matrix, the historical decision data semantic interaction enhancement is performed to obtain a historical decision data semantic understanding associated interaction enhancement matrix; Performing semantic understanding of the data to be decided on the data to be decided to obtain a semantic understanding vector of the data to be decided; The semantic understanding vector of the data to be decided and the semantic understanding association interaction enhancement matrix of the historical decision data are integrated to obtain a semantic query representation vector of the data to be decided; Obtaining a recommendation result based on the semantic query representation vector of the data to be decided; The method of arranging the plurality of historical decision data semantic understanding feature vectors into a historical decision data semantic understanding associated interaction feature matrix and then performing historical decision data semantic interaction enhancement to obtain a historical decision data semantic understanding associated interaction enhancement matrix includes: Arranging the plurality of historical decision data semantic understanding feature vectors into the historical decision data semantic understanding associated interaction feature matrix; The historical decision data semantic understanding associated interaction feature matrix is ​​input into the historical decision data semantic interaction enhancement module based on the bidirectional interactive attention mechanism to obtain the historical decision data semantic understanding associated interaction enhancement matrix.

Images