Enterprise management intelligent decision-making system integrated with big data analysis

Abstract

The invention discloses an enterprise management intelligent decision-making system integrated with big data analysis, and relates to the technical field of enterprise intelligent management, and the system comprises an installation package receiving and analyzing module which is used for connecting an enterprise management software system and receiving and analyzing an installation package when upgrading is triggered; the installation package portrait construction module is used for extracting a function label word vector and a key restoration module based on an NLP technology to form an upgrade package portrait; the data portrait construction module is used for extracting and analyzing multi-source historical operation data and forming an enterprise data portrait containing module use popularity and fault records; and the quantitative matching module is used for carrying out upgrading matching calculation according to the two types of portraits, outputting an upgrading quantitative index, and comparing the upgrading quantitative index with a preset index to generate an upgrading decision result. Therefore, the technical effects of optimizing the upgrading suitability and enhancing the autonomous analysis capability and the active decision-making capability are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise intelligent management, and particularly to an enterprise management intelligent decision-making system integrating big data analysis. Background Art

[0002] Currently, enterprise management mostly relies on digital software for operation, and these software are upgraded regularly to introduce new functions, fix bugs or improve performance. However, the upgrade of enterprise software is usually uniformly pushed by software vendors, and the upgrade content is mainly based on general requirements or industry trends, rather than the personalized needs of specific enterprises. This "one-size-fits-all" upgrade method may cause the following problems for enterprises: The new functions after upgrade may not match the actual business needs of the enterprise, resulting in waste of resources; the enterprise may introduce unnecessary complexity or performance degradation due to the upgrade, especially when the upgrade content is incompatible with the running state of the enterprise's existing system; the enterprise lacks effective tools to evaluate the necessity of upgrade and can only passively accept the upgrade without being able to make active decisions. Summary of the Invention

[0003] The present invention provides an enterprise management intelligent decision-making system integrating big data analysis to solve the technical problems of poor software upgrade adaptability and lack of independent analysis and active decision-making ability in the prior art, and to achieve the technical effects of online monitoring, complete monitoring and good early warning ability.

[0004] The enterprise management intelligent decision-making system integrating big data analysis provided by the present invention includes: An installation package receiving and parsing module, which is used to connect to the enterprise management software system and receive and parse the upgrade installation package when the enterprise management software system triggers an upgrade.

[0005] An installation package portrait construction module, which is used to perform semantic label recognition on the upgrade installation package based on NLP technology and construct an upgrade package portrait, and the upgrade package portrait includes a function label word vector set and a key repair module set of the upgrade package.

[0006] A data portrait construction module, which is used to extract multi-source historical operation data from multiple heterogeneous software systems of the target enterprise, analyze the multi-source historical operation data, and construct an enterprise data portrait, and the enterprise data portrait includes a usage heat set and a failure module set of function modules.

[0007] A quantization matching module, which is used to perform upgrade matching quantization calculation according to the upgrade package portrait and the enterprise data portrait, output an upgrade quantization index, compare the upgrade quantization index with a preset quantization index, and return an upgrade decision result.

[0008] In a feasible implementation manner, upgrade matching quantitative calculation is performed according to the upgrade package portrait and the enterprise data portrait. The quantitative matching module includes: An overlapping matching degree calculation unit, configured to perform cosine similarity matching on the function tag word vector set of the upgrade package portrait and the usage heat set of the enterprise data portrait, and output a function overlapping matching degree.

[0009] An association matching degree calculation unit, configured to perform fault association recognition on the key repair module set of the upgrade package portrait and the fault module set of the enterprise data portrait, and output a fault association matching degree.

[0010] A quantitative index generation unit, configured to calculate the function overlapping matching degree and the fault association matching degree, and output an upgrade quantitative index based on the upgrade installation package.

[0011] In a feasible implementation manner, upgrade matching quantitative calculation is performed according to the upgrade package portrait and the enterprise data portrait. The quantitative matching module further includes: A data analysis unit, configured to analyze the multi-source historical operation data to obtain a function module time factor set, where the function module time factor does not represent the interval time of each function module since the last upgrade.

[0012] A weight calculation and matching degree update unit, configured to calculate weights for the modules with overlapping functions according to the function module time factor set, update the function overlapping matching degree, calculate weights for the modules with associated faults according to the function module time factor set, and update the fault association matching degree.

[0013] An index update unit, configured to recalculate the updated function overlapping matching degree and the updated fault association matching degree, and update the upgrade quantitative index.

[0014] In a feasible implementation manner, the upgrade quantitative index is compared with a preset quantitative index to return an upgrade decision result. The execution steps of the quantitative matching module include: If the upgrade quantitative index is less than the preset quantitative index, cache the upgrade installation package.

[0015] If the upgrade quantitative index is greater than or equal to the preset quantitative index, perform system upgrade on the enterprise management software system according to the upgrade installation package.

[0016] In a feasible implementation manner, to return an upgrade decision result, the quantitative matching module further includes: An enterprise management cloud center establishment unit, configured to establish an enterprise management cloud center, where the enterprise management cloud center is connected to multi-terminal enterprise management software systems, and one enterprise user corresponds to each enterprise management software system.

[0017] A multi - terminal upgrade decision - making and execution unit, which is used to obtain multiple upgrade decision results returned by the multi - terminal enterprise management software system according to the enterprise management cloud center, and upgrade the multi - terminal enterprise management software system with the multiple upgrade decision results.

[0018] In a feasible implementation manner, to obtain a functional label word vector set, the execution steps of the installation package portrait construction module include: Perform text processing on the upgrade installation package based on NLP technology, and output upgrade installation text.

[0019] Define a functional keyword sample, and obtain the context word vectors of each functional keyword in the upgrade installation text.

[0020] Aggregate the context word vectors to obtain functional labels representing each keyword, and output the functional label word vector set.

[0021] In a feasible implementation manner, to obtain a key repair module set, the execution steps of the installation package portrait construction module include: Define a repair action keyword sample, identify the relevant modules of each repair action keyword in the upgrade installation text, and output the key repair module set.

[0022] In a feasible implementation manner, to obtain a usage heat set, the execution steps of the data portrait construction module include: Analyze the multi - source historical operation data, where the multi - source historical operation data includes call logs, operation frequencies, exception records, and response delays.

[0023] Obtain the call frequencies, continuous call cycles, and the number of operating users of each functional module.

[0024] Perform normalization processing on the call frequencies, continuous call cycles, and the number of operating users of each functional module to obtain the usage heat set.

[0025] In a feasible implementation manner, to obtain a faulty module set, the execution steps of the data portrait construction module include: Analyze the multi - source historical operation data to obtain the fault frequencies, fault levels, and fault influence scopes of each functional module.

[0026] Perform normalization processing on the fault frequencies, fault levels, and fault influence scopes of each functional module to obtain the faulty module set.

[0027] The present invention discloses an enterprise management intelligent decision-making system integrating big data analysis, including: an installation package receiving and parsing module for connecting to an enterprise management software system, receiving and parsing an upgrade installation package when the system triggers an upgrade; an installation package portrait construction module for performing semantic tag recognition on the upgrade installation package based on NLP technology to construct an upgrade package portrait including a functional tag word vector set and a key repair module set; a data portrait construction module for extracting multi-source historical operation data from multiple heterogeneous software systems of a target enterprise and constructing an enterprise data portrait through analysis, including a usage heat set and a failure module set of each functional module; a quantization matching module for performing quantization calculation of upgrade matching based on the upgrade package portrait and the enterprise data portrait, generating an upgrade quantization index, comparing the index with a preset quantization index, and finally outputting an upgrade decision result. The enterprise management intelligent decision-making system integrating big data analysis disclosed by the present invention solves the technical problems of poor software upgrade adaptability and lack of independent analysis and active decision-making capabilities of enterprises, and achieves the technical effects of optimizing upgrade adaptability, enhancing independent analysis capabilities and active decision-making capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 FIG. is a schematic structural diagram of the enterprise management intelligent decision-making system integrating big data analysis of the present invention.

[0029] Figure 2 FIG. is a schematic flow diagram of obtaining a functional tag word vector set in the enterprise management intelligent decision-making system integrating big data analysis of the present invention.

[0030] Description of reference numerals: installation package receiving and parsing module 11, installation package portrait construction module 12, data portrait construction module 13, quantization matching module 14. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] The above technical solutions will be described in detail below in combination with the accompanying drawings of the specification and specific embodiments to better understand the above technical solutions. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments for explaining the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention. In addition, it should be noted that for the sake of description, only parts related to the present invention are shown in the drawings rather than all.

[0032] Embodiment Figure 1 FIG. is a schematic structural diagram of the enterprise management intelligent decision-making system integrating big data analysis of the present invention, wherein the enterprise management intelligent decision-making system integrating big data analysis includes: The installation package receiving and parsing module 11 is used to connect to the enterprise management software system and receive and parse the upgrade installation package when the enterprise management software system triggers an upgrade.

[0033] Specifically, the installation package receiving and parsing module 11 is used to implement the docking with the enterprise management software system and complete the receiving and parsing process of the upgrade installation package before the system upgrade.

[0034] Specifically, the module establishes a communication channel with the enterprise management software system through a network interface (such as API or WebSocket) to ensure the stability and real-time nature of data transmission; when the enterprise management software system triggers an upgrade, the module obtains the binary file of the upgrade installation package from the server side through a preset communication protocol (such as HTTP or FTP); then, the obtained installation package is structurally processed to extract key information therein (such as function description, repair module, version number, etc.).

[0035] Exemplarily, the upgrade installation package may include content such as update files of software function modules, configuration parameters, database structure change instructions, script files, etc.; the operations performed by the installation package receiving and parsing module 11 for parsing include: performing integrity verification on the installation package (such as through algorithms such as MD5, SHA, etc.); decompressing the content of the installation package to identify the module types and version information contained therein; parsing the upgrade script or configuration file to extract the key parameters required for the upgrade; and passing the parsing results to subsequent modules for generating an upgrade package portrait. For example, assume that an enterprise management software system triggers an upgrade, and the upgrade installation package is a ZIP file containing a function description file (in JSON format) and a list of repair modules (in XML format). The ZIP file is received through the HTTP protocol, and the metadata in the JSON and XML files is extracted after decompression to achieve parsing. The parsing results show that the upgrade package contains a total of 3 new function modules and 2 repair modules. This information will be passed to subsequent modules for generating an upgrade package portrait and performing matching calculations.

[0036] The installation package receiving and parsing module 11 plays an entry role in the entire system, ensuring that subsequent modules can analyze and make decisions based on accurate installation package information.

[0037] The upgrade package portrait construction module 12 is used to perform semantic tag recognition on the upgrade installation package based on NLP technology and construct an upgrade package portrait, where the upgrade package portrait includes a functional tag word vector set and a key repair module set of the upgrade package.

[0038] Specifically, the upgrade package portrait construction module 12 is used to perform semantic analysis on the content of the installation package based on natural language processing (NLP) technology after completing the receiving and parsing of the upgrade installation package, so as to construct a functional portrait of the upgrade package to assist subsequent upgrade evaluation, risk prediction, and module-level dependency management.

[0039] Specifically, the functional label word vector set is a set of vectors formed by converting functional keywords into word vectors (e.g., through the Word2Vec model), which is used to quantify the semantic information of functional modules. The key repair module set is a set of modules related to repair actions identified in the obtained installation package, such as "permission control module", "data synchronization module", "report engine module", etc., which is used for subsequent matching calculations.

[0040] In some implementation manners, such as Figure 2 shown, the steps of the installation package portrait construction module 12 for obtaining the functional label word vector set include: S100: Perform text processing on the upgrade installation package based on NLP technology, and output the upgrade installation text; S200: Define functional keyword samples, and obtain the context word vectors of each functional keyword in the upgrade installation text; S300: Aggregate the context word vectors to obtain functional labels representing each keyword, and output the functional label word vector set.

[0041] Specifically, the functional keyword samples are a predefined set of keywords, which are used to identify the boundary points of functional modules in the installation package, that is, the keywords corresponding to the functional change points. These keywords can be defined by domain experts or extracted from historical data through machine learning models.

[0042] Specifically, the functional label is a semantic label assigned to each functional keyword based on the context semantic aggregation result, such as "performance enhancement type", "compatibility update type", "security repair type", etc.

[0043] Specifically, first, based on natural language processing (NLP) technology, including semantic recognition models such as BERT, Word2Vec, FastText, etc., perform preprocessing operations on the text information in the upgrade installation package to obtain the upgrade installation text. Among them, the upgrade installation text may include change logs, upgrade description documents, script comments, configuration descriptions, etc. The text processing process includes character cleaning, language detection, word segmentation (e.g., splitting "customer management module" into "customer", "management", "module"), part-of-speech tagging, stop word filtering (e.g., "of", "is", "and", etc.), stemming (e.g., unifying "management" and "manager" into "management"), etc.

[0044] Specifically, next, load predefined samples of functional keyword (such as "customer management", "order processing", "data analysis", etc.), and use them as matching benchmarks to locate the predefined functional keywords in the upgrade installation text, and based on the context window mechanism (such as 5 words before and after), extract the context content when these keywords appear in the text; then, through a pre-trained word vector model (such as Word2Vec, GloVe, BERT, etc.), convert each word and its context information in the text into a fixed-length vector representation (i.e., context word vector) to capture the semantic features of the word.

[0045] Furthermore, perform an aggregation operation (such as average pooling, weighted aggregation, attention mechanism, etc.) on the context word vectors corresponding to each functional keyword, so as to generate a vector representing the semantic features of the keyword as the functional label vector of the functional keyword. Then, summarize the functional label vectors of all functional keywords to form a set of functional label word vectors of the upgrade installation package, which is output as part of the upgrade package portrait for subsequent analysis and invocation.

[0046] In the above process, through the aggregation of context word vectors, the module can capture the semantic features of functional keywords, avoid semantic deviation caused by isolated keywords, and thus improve the semantic matching accuracy: at the same time, by dynamically updating the samples of functional keywords, it can adapt to the changes in functional descriptions in different fields and enterprises, reduce manual intervention and improve the intelligence level of the system.

[0047] In some implementation manners, obtain a set of key repair modules, and the execution steps of the installation package portrait construction module 12 include: Define a sample of repair action keywords, identify the relevant modules of each repair action keyword in the upgrade installation text, and output a set of key repair modules.

[0048] Specifically, first, define a set of repair action keyword samples to represent common repair behaviors or actions in the upgrade installation package, which can be constructed by means of manual collation, historical patch analysis, trouble ticket mining, etc. Exemplarily, the repair action keyword samples include but are not limited to: "repair", "correct", "adjust", "optimize", "patch", "resolve", "avoid", "trim", "rectify", "eliminate", etc., which are respectively used to locate semantic segments related to defect repair, security reinforcement, performance optimization, etc. in the upgrade installation text.

[0049] Specifically, scan the upgrade installation text to identify sentences or paragraphs containing keywords of repair actions, and use entity recognition and dependency syntactic analysis to further analyze technical entities such as system modules, components, class names, function names, and configuration items mentioned in their context. For example, if the text appears "repair the performance problem of the order processing module", then "repair" is associated with "order processing module".

[0050] Optionally, perform lexical normalization processing on the extraction results (such as name de-duplication, module name standardization mapping) to form a denoised module set (i.e., the key repair module set), which is achieved by the following methods: use named entity recognition (NER) technology to identify module names; perform matching based on existing module dictionaries or module mapping relationships; perform reverse parsing in combination with code comments or path information (such as "fixed the token refresh logic in the com.company.auth permission control module").

[0051] Optionally, structurally organize the above-identified repair actions and their associated modules to form the key repair module set involved in this upgrade installation package, and output it as part of the upgrade package portrait; among them, this module set can include information such as module name, repair type, influence range, and repair location, so as to provide high-precision data support for subsequent matching calculations.

[0052] The data portrait construction module 13 is used to extract multi-source historical operation data from multiple heterogeneous software systems of the target enterprise, analyze the multi-source historical operation data, and construct an enterprise data portrait, where the enterprise data portrait includes the usage heat set and the fault module set of functional modules.

[0053] Specifically, multi-source historical operation data refers to various types of data generated during the operation of enterprise software systems, including call logs, operation frequencies, exception records, response delays, etc. The multi-source historical operation data corresponds to multiple heterogeneous software systems (such as software systems applicable to different enterprises, scenarios, users, etc.). That is to say, it is necessary to obtain the multi-source historical operation data of all software systems (multiple heterogeneous software systems) involved in the software to be updated in order to judge the possible impacts caused by software updates. This is because different heterogeneous software systems have their own requirements and operating environments, which may lead to different impacts on software updates in different software systems (the upgrade of a single system may affect the normal operation of other systems).

[0054] Specifically, the enterprise data portrait is a knowledge abstraction result obtained by analyzing the above multi-source historical operation data; for example, through semantic modeling, clustering analysis, and label annotation of historical data, a comprehensive portrait information that can characterize the actual usage scenarios and module quality status of enterprise software is constructed.

[0055] Specifically, the usage heat set refers to the set of the usage frequencies and activity levels of each functional module quantified by analyzing multi-source historical operation data; the faulty module set refers to the set of functional modules that frequently experience faults or performance issues identified by analyzing multi-source historical operation data.

[0056] The above-mentioned data portrait construction module 13 can generate an accurate usage heat set and faulty module set through comprehensive analysis of the enterprise's multi-source historical operation data, providing reliable data support for subsequent matching calculations.

[0057] In some implementation manners, to obtain the applicable usage heat set, the execution steps of the data portrait construction module 13 include: Analyze the multi-source historical operation data, where the multi-source historical operation data includes call logs, operation frequencies, exception records, and response delays; obtain the call frequency, continuous call period, and number of user operations of each functional module; perform normalization processing on the call frequency, continuous call period, and number of user operations of each functional module to obtain the usage heat set.

[0058] Specifically, first, by docking with the software system deployed by the target enterprise, obtain the historical data during the actual operation of the system, including: call logs, recording the call time, call times, call sources, etc. of each functional module or interface; operation frequencies, counting the number of operations of the functional module by users in the front-end interface or system operations; exception records, including events such as errors, exceptions, and failed retries generated during the operation of the module; response delays, recording performance indicators such as the response time and delay fluctuation of the module when it is called.

[0059] Furthermore, based on the above multi-source historical operation data, extract the following core usage indicators for each functional module: the number of times the module is called per unit time (call frequency), the time span during which a certain module is continuously accessed for a period of time (i.e., continuous call period), and the number of independent users who access or operate the module within the statistical period (number of user operations). Then, for the unified quantification and comparison between different indicators, perform normalization processing on the above three indicators. Linear normalization, Z-score standardization, quantile normalization, etc. can be used to convert the indicator values into standardized values in the [0, 1] interval to eliminate the differences in different indicator dimensions and magnitudes, and fuse the normalized call frequency, continuous call period, number of user operations, etc. indicators to construct the usage heat vector of each functional module. This usage heat vector is used to quantitatively represent the usage heat of each functional module, and multiple usage heat vectors are aggregated to form the usage heat set of the system.

[0060] Exemplarily, assuming that the statistical period is 24 hours, extract the indicators and perform normalization processing to construct the usage heat vector and form the usage heat set: Table 1 Core usage metrics of functional modules Function module ID Call frequency (times / hour) Continuous call period (minutes) Number of operating users (persons) M001 3 180 2 M002 2 300 2 M003 1 60 1 Table 2 Normalization results of core usage metrics of functional modules Function module ID Call frequency (normalized) Continuous call period (normalized) Number of operating users (normalized) M001 1.0 (180-60) / (300-60)=0.5 (2-1) / (2-1)=1.0 M002 (2-1) / (3-1)=0.5 (300-60) / (300-60)=1.0 (2-1) / (2-1)=1.0 M003 0.0 0.0 0.0 Fuse the normalized metrics with equal weights to generate a module heat vector: Table 3 Usage heat set represented by the usage heat vector Function module ID Usage heat vector M001 [1.0,0.5,1.0] M002 [0.5,1.0,1.0] M003 [0.0,0.0,0.0] Optionally, in some other implementation manners, the construction process of the usage heat set can be further optimized, including: introducing a time decay weight to assign a higher weight to recent usage behaviors; using a clustering algorithm to classify the module heat patterns (such as high-frequency, low-frequency, fluctuating); combining user portraits to realize the analysis of module heat differences under different user groups; introducing negative metrics such as exception rate and response delay to correct the module heat score and improve the stability evaluation ability.

[0061] Through the above steps, the data portrait construction module 13 can achieve: Accurately depict the usage intensity and activity of each functional module; provide a quantitative module heat ranking to provide data support for subsequent test priority ranking, upgrade impact assessment, resource scheduling, etc.; facilitate the comparison and analysis of module heat across time periods, versions, and enterprises.

[0062] In some implementation manners, obtain a set of faulty modules. The execution steps of the data portrait construction module 13 include: Analyze the multi-source historical operation data to obtain the fault frequency, fault level, and fault impact range of each functional module; perform normalization processing on the fault frequency, fault level, and fault impact range of each functional module to obtain a set of faulty modules.

[0063] Specifically, to obtain the set of faulty modules of each functional module in the software system of the target enterprise, first, based on the multi-source historical operation data, analyze and extract the fault frequency, fault level, and fault impact range corresponding to different functional modules, that is, count the number of times each module fails within a set time window, obtain the classification based on the severity of the fault (for example, level 1 fault: system unavailable, service interruption; level 2 fault: function anomaly but system available; level 3 fault: performance degradation or occasional error), and evaluate the number of users, system components, or business process nodes affected by the fault. Among them, the above classification based on the severity of the fault is the classification result under the preset rules of the target scenario.

[0064] Furthermore, to achieve unified measurement among different metrics, the above three metrics are normalized, specifically including: linearly normalizing the failure frequency; numerically mapping the failure level (e.g., mapping level 1 to 1.0 and level 3 to 0.3); proportionally processing the impact scope according to the number of users or modules; the normalized metric values will be uniformly transformed into the interval [0, 1] through a data fusion method (such as weighted summation), and frequentness screening will be performed through a preset selection rule to form a set of faulty modules, which is convenient for subsequent fusion and sorting. Optionally, the frequentness screening includes screening based on relative percentages, such as selecting the modules corresponding to the first 50% in the serialized fusion results as faulty modules.

[0065] The quantization matching module 14 is configured to perform an upgrade matching quantization calculation according to the upgrade package profile and the enterprise data profile, output an upgrade quantization metric, compare the upgrade quantization metric with a preset quantization metric, and return an upgrade decision result.

[0066] Specifically, the upgrade matching quantization calculation refers to the process of comparing the upgrade package profile and the enterprise data profile through a mathematical model or algorithm and calculating the corresponding quantization metric. The obtained upgrade quantization metric is used to measure the matching degree between the upgrade package and the enterprise requirements. The preset quantization is a predefined threshold used to determine whether the upgrade quantization metric meets the upgrade standard.

[0067] Specifically, the upgrade quantization metric is compared with the preset quantization metric. If the quantization metric is greater than or equal to the preset value, an upgrade decision result of "recommended upgrade" can be returned; otherwise, an upgrade decision result of "not recommended upgrade" is returned.

[0068] Through quantization calculation, the quantization matching module 14 can accurately evaluate the matching degree between the upgrade package and the enterprise requirements, reduce manual intervention, and avoid errors caused by subjective judgment.

[0069] In some embodiments, for the upgrade matching quantization calculation according to the upgrade package profile and the enterprise data profile, the quantization matching module 14 includes: An overlap matching degree calculation unit, configured to perform cosine similarity matching on the function tag word vector set of the upgrade package profile and the usage heat set of the enterprise data profile, and output a function overlap matching degree; a correlation matching degree calculation unit, configured to perform fault correlation identification on the key repair module set of the upgrade package profile and the faulty module set of the enterprise data profile, and output a fault correlation matching degree; a quantization metric generation unit, configured to calculate the function overlap matching degree and the fault correlation matching degree, and output an upgrade quantization metric based on the upgrade installation package.

[0070] Specifically, by performing an upgraded matching quantitative calculation on the upgrade package portrait and the enterprise data portrait, intelligent upgrade decision-making in the enterprise environment is realized. Among them, the overlapping matching degree refers to the similarity between the function tag word vector set of the upgrade package and the usage heat set of the enterprise data, which is usually calculated by cosine similarity. The fault correlation matching degree refers to the matching degree between the key repair module set of the upgrade package and the fault module set of the enterprise data, which can be calculated by set intersection or similarity.

[0071] Specifically, taking the function tag word vector set (from the upgrade package portrait) and the usage heat set (from the enterprise data portrait) as inputs, the similarity between the two sets (i.e., the function overlapping matching degree) is calculated by cosine similarity to measure the overlapping degree between the upgrade package function and the enterprise high-frequency usage module. Among them, the value range of the function overlapping matching degree is [0, 1], and the larger the value, the higher the matching degree.

[0072] At the same time, taking the key repair module set (from the upgrade package portrait) and the fault module set (from the enterprise data portrait) as inputs, module association recognition is performed through methods such as module ID matching, function tag similarity, or call relationship graph analysis, and the output result is used as the fault correlation matching degree to reflect the potential value of the upgrade package in improving the current system stability of the enterprise.

[0073] Exemplarily, the fault correlation matching degree of the two sets is calculated by set intersection, and the formula for the fault correlation matching degree is: Furthermore, a fusion calculation is performed on the above two matching degree results to output an upgrade priority or a value quantitative score. For example, a weighted sum of the function overlapping matching degree and the fault correlation matching degree is performed to generate an upgrade quantitative index, which is used to guide whether to perform an upgrade, upgrade priority ranking, or upgrade resource allocation.

[0074] Through the above multiple units, it can be realized: based on the actual business usage situation and fault risk, intelligently evaluate the adaptability and value of the upgrade package in the enterprise environment; avoid ineffective upgrades or low-value upgrades, and improve the usage efficiency of upgrade resources; support personalized upgrade recommendations and the formulation of differentiated operation and maintenance strategies; and improve the system stability and function coverage effect.

[0075] In some embodiments, when performing the upgrade matching quantitative calculation according to the upgrade package portrait and the enterprise data portrait, the quantitative matching module 14 further includes: A data analysis unit is used to analyze the multi-source historical operation data to obtain a set of function module time factors, where the function module time factor is characterized by the time interval since the last upgrade of each function module; a weight calculation and matching degree update unit is used to calculate the weights of the modules with overlapping functions according to the set of function module time factors, update the overlapping function matching degree, calculate the weights of the modules associated with faults according to the set of function module time factors, and update the fault association matching degree; an index update unit is used to recalculate the updated overlapping function matching degree and the updated fault association matching degree, and update the upgrade quantization index.

[0076] Specifically, on the basis of the upgrade matching quantization calculation according to the upgrade package portrait and the enterprise data portrait, a set of function module time factors is further introduced. This set of function module time factors is used to introduce information on the upgrade time interval dimension, so as to adjust the score of the system that has not been upgraded for a long time, and improve the rationality and timeliness of upgrade recommendations.

[0077] Specifically, first, extract the time dimension information of the function modules from the enterprise's multi-source historical operation data to construct a set of function module time factors. This includes extracting the last upgrade time of each function module in the current enterprise system (such as through the system maintenance log), comparing it with the current time, and calculating the upgrade time interval of each module as the function module time factor. The larger this time factor is, the longer the module has not been upgraded, and there may be security risks, technical debts or compatibility problems, and it may be more inclined to update it. Correspondingly, if a certain module has been manually cancelled or rolled back several times in a row during updates, it means that the current version of this module may be the most suitable / stable / version that conforms to the target enterprise process. At this time, the time factor may be extremely large.

[0078] Furthermore, adjust the matching degree based on the time factor, including: Adjustment of the overlapping function matching degree, which involves weighting the modules participating in the overlapping function matching in combination with their time factors. For example, perform a time-weighted average or weighted summation on the cosine similarity results; among them, the larger the function module time factor, the higher its weight in the matching degree calculation, reflecting its urgency to upgrade; Adjustment of the fault association matching degree, which involves weighting the modules participating in the fault association matching in combination with their time factors. Among them, if a certain module has both fault risks and has not been upgraded for a long time, its priority will be significantly improved; correspondingly, if a module has skipped updates several times in a row (corresponding to an extremely large function module time factor), its priority can be reduced to the lowest or set to not be automatically updated (manual review is required before update).

[0079] Further, based on the updated matching degree result, the index update unit recalculates and outputs the upgrade quantization index, including invoking the updated function overlap matching degree and fault association matching degree, and recalculating the upgrade score according to a preset weighting function, and outputting the final upgrade quantization index to support upgrade decision ranking, automated recommendation or manual approval.

[0080] By introducing a time factor, the above units and corresponding processes can achieve the following technical effects: dynamically identifying modules that have not been upgraded for a long time, improving system security and technology update efficiency; avoiding ignoring the upgrade of old modules due to "low function overlap" or "no current faults".

[0081] In some implementation manners, the upgrade quantization index is compared with a preset quantization index to return an upgrade decision result. The execution steps of the quantization matching module 14 include: If the upgrade quantization index is less than the preset quantization index, cache the upgrade installation package; if the upgrade quantization index is greater than or equal to the preset quantization index, perform a system upgrade on the enterprise management software system according to the upgrade installation package.

[0082] Specifically, the upgrade quantization index is a numerically calculated value used to represent the matching degree between the upgrade package and enterprise requirements. The preset quantization index is a predefined threshold used to determine whether the upgrade quantization index meets the upgrade standard.

[0083] Specifically, if the upgrade quantization index is less than the preset quantization index, the upgrade installation package is temporarily stored in the system for possible subsequent use (when the subsequent upgrade quantization index is greater than the preset quantization index); if the upgrade quantization index is greater than or equal to the preset quantization index, it can be considered that the current upgraded version has a high matching degree with the current system, and the enterprise management software system can be directly upgraded according to the upgrade installation package to ensure that the enterprise software system can be updated in a timely manner and its functions and performance can be improved.

[0084] Optionally, the upgrade threshold can be configured to support dynamic adjustment, and thus can automatically change based on business peak periods, budget cycles or operation and maintenance strategies.

[0085] Optionally, the upgrade decision result can be output to the approval process interface and supplemented with a hybrid strategy of "manual confirmation + automatic execution". At the same time, a multi-level threshold mechanism can also be introduced to implement various strategies such as "forced upgrade", "recommended upgrade", and "delayed upgrade".

[0086] Optionally, a regular re-evaluation mechanism is set for the cached upgrade packages to clean up unnecessary upgrade packages in a timely manner or streamline the upgrade packages, only retaining the incremental / changed parts, to avoid waste of system resources caused by long-term backlogs.

[0087] In some implementations, when returning the upgrade decision result, the quantization matching module 14 further includes: An enterprise management cloud center establishment unit, configured to establish an enterprise management cloud center, where the enterprise management cloud center is connected to multi-terminal enterprise management software systems, and one enterprise user corresponds to each enterprise management software system; a multi-terminal upgrade decision and execution unit, configured to obtain, according to the enterprise management cloud center, multiple upgrade decision results returned by the multi-terminal enterprise management software systems, and upgrade the multi-terminal enterprise management software systems with the multiple upgrade decision results.

[0088] Specifically, first, a unified enterprise management cloud platform is constructed through the enterprise management cloud center establishment unit to centrally manage the upgrade tasks of multiple enterprise management software systems. Exemplarily, it includes: establishing an enterprise management cloud center and connecting multiple enterprise management software systems deployed on different terminals or in different enterprise environments, and one independent enterprise management software system instance corresponds to each enterprise user; then, through the unified data collection, index calculation, upgrade package distribution, and upgrade policy control capabilities of the enterprise management cloud center, information such as the running status, upgrade history, and functional modules of each enterprise system is centrally managed.

[0089] Furthermore, through the multi-terminal upgrade decision and execution unit, based on the enterprise management cloud center, the upgrade decision and execution processes of multiple terminal systems are coordinated uniformly. Specifically, it includes: obtaining the upgrade quantization indicators and upgrade decision results respectively returned by multiple enterprise management software systems from the enterprise management cloud center, and then, according to the unified threshold set by the cloud center or the enterprise-customized threshold, combining the upgrade decision results of each system, and performing the upgrade action; for example, if the upgrade quantization indicator of an enterprise system meets the upgrade condition, the corresponding upgrade installation package is automatically pushed to the system and the upgrade is executed; if the condition is not met, the upgrade package is cached in the system and waits for subsequent re-evaluation or administrator confirmation.

[0090] Through the above process, centralized upgrade management of multiple enterprises and multi-terminal systems can be realized, meeting the multi-tenant upgrade requirements of group enterprises or large service providers.

[0091] In summary, the enterprise management intelligent decision-making system integrating big data analysis provided by the present invention has the following technical effects: The installation package receiving and parsing module is used to connect to the enterprise management software system and receive and parse the upgrade installation package when the system triggers an upgrade; the installation package portrait construction module uses NLP technology to identify semantic tags for the upgrade installation package and construct an upgrade package portrait containing a functional tag word vector set and a key repair module set; the data portrait construction module extracts multi-source historical operation data from multiple heterogeneous software systems of the target enterprise and constructs an enterprise data portrait through analysis, including the usage popularity set and the failure module set of each functional module; the quantization matching module performs quantization calculations for upgrade matching based on the upgrade package portrait and the enterprise data portrait, generates an upgrade quantization index, compares the index with a preset quantization index, and finally outputs an upgrade decision result, so as to achieve the technical effects of optimizing upgrade adaptability, enhancing the independent analysis ability and the active decision-making ability.

[0092] It should be understood that the disclosed embodiments of the present invention and the above descriptions enable those skilled in the art to implement the present invention using the present invention. At the same time, the present invention is not limited to the above-mentioned part of the embodiments. It should be understood that those of ordinary skill in the art can still modify the technical solutions recorded in the foregoing embodiments or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. An enterprise management intelligent decision-making system integrating big data analysis, characterized by: include: An installation package receiving and parsing module, used to connect to the enterprise management software system, and receive and parse the upgrade installation package when the enterprise management software system triggers an upgrade; An installation package portrait construction module, used to perform semantic label recognition on the upgrade installation package based on NLP technology to construct an upgrade package portrait, wherein the upgrade package portrait includes a function label word vector set and a key repair module set of the upgrade package; A data profile building module is used to extract multi-source historical operation data from multiple heterogeneous software systems of a target enterprise, analyze the multi-source historical operation data, and build an enterprise data profile, wherein the enterprise data profile includes a usage heat set of a functional module and a fault module set; The quantitative matching module is used to perform upgrade matching quantitative calculation according to the upgrade package portrait and the enterprise data portrait, output upgrade quantitative indicators, compare the upgrade quantitative indicators with preset quantitative indicators, and return the upgrade decision result.

2. The enterprise management intelligent decision-making system integrating big data analysis as claimed in claim 1, characterized in that: Performing upgrade matching quantitative calculation according to the upgrade package profile and the enterprise data profile, the quantitative matching module includes: An overlapping matching degree calculation unit, used for calling the function label word vector set of the upgrade package portrait and the usage heat set of the enterprise data portrait to perform cosine similarity matching, and outputting the function overlapping matching degree; A correlation matching degree calculation unit, used for calling the key repair module set of the upgrade package portrait and the fault module set of the enterprise data portrait to perform fault correlation identification and output the fault correlation matching degree; The quantitative index generating unit is used to calculate the function overlap matching degree and the fault correlation matching degree, and output the upgrade quantitative index based on the upgrade installation package.

3. The enterprise management intelligent decision-making system integrating big data analysis as claimed in claim 2, characterized in that: Performing upgrade matching quantitative calculation according to the upgrade package portrait and the enterprise data portrait, the quantitative matching module further includes: A data analysis unit, configured to analyze the multi-source historical operation data to obtain a set of function module time factors, wherein the function module time factors do not represent the interval time from the last upgrade of each function module; A weight calculation and matching degree updating unit, configured to perform weight calculation on modules with overlapping functions according to the function module time factor set, update the function overlapping matching degree, perform weight calculation on modules with fault association according to the function module time factor set, and update the fault association matching degree; The index updating unit is used to recalculate the updated function overlap matching degree and the updated fault correlation matching degree, and update the upgrade quantitative index.

4. The enterprise management intelligent decision-making system integrating big data analysis as claimed in claim 1, characterized in that: The upgrade quantitative index is compared with the preset quantitative index, and the upgrade decision result is returned. The execution steps of the quantitative matching module include: If the upgrade quantitative index is less than the preset quantitative index, caching the upgrade installation package; If the upgrade quantitative index is greater than or equal to the preset quantitative index, the enterprise management software system is upgraded according to the upgrade installation package.

5. The enterprise management intelligent decision-making system integrating big data analysis as claimed in claim 1, characterized in that: Return the upgrade decision result. The quantitative matching module also includes: An enterprise management cloud center establishment unit, used to establish an enterprise management cloud center, wherein the enterprise management cloud center is connected to multiple enterprise management software systems, wherein each enterprise management software system corresponds to one enterprise user; A multi-terminal upgrade decision and execution unit is used to obtain multiple upgrade decision results returned by the multi-terminal enterprise management software system according to the enterprise management cloud center, and upgrade the multi-terminal enterprise management software system with the multiple upgrade decision results.

6. The enterprise management intelligent decision-making system integrating big data analysis as claimed in claim 1, characterized in that: Obtaining a function label word vector set, the execution steps of the installation package portrait construction module include: Performing text processing on the upgrade installation package based on NLP technology to output an upgrade installation text; Define function keyword samples, and obtain context word vectors of each function keyword in the upgrade installation text; Aggregate the context word vectors to obtain a functional label representing each keyword, and output a functional label word vector set.

7. The enterprise management intelligent decision-making system integrating big data analysis as claimed in claim 6, characterized in that: Obtaining a set of key repair modules, the execution steps of the installation package image building module include: Define repair action keyword samples, identify related modules of each repair action keyword in the upgrade installation text, and output a key repair module set.

8. The enterprise management intelligent decision-making system integrating big data analysis as claimed in claim 1, characterized in that: Obtaining the usage heat set, the execution steps of the data profile building module include: Analyze the multi-source historical operation data, wherein the multi-source historical operation data includes call logs, operation frequency, exception records, and response delay; Get the calling frequency, continuous calling cycle and number of operating users of each functional module; The calling frequency, continuous calling cycle and number of operating users of each functional module are normalized to obtain the usage heat set.

9. The enterprise management intelligent decision-making system integrating big data analysis as claimed in claim 8, characterized in that: Obtaining a fault module set, the execution steps of the data profile building module include: Analyze the multi-source historical operation data to obtain the fault frequency, fault level and fault impact range of each functional module; The fault frequency, fault level and fault impact range of each functional module are normalized to obtain a fault module set.

Images