A recommended approach for intelligent asset auditing and compliance analysis

Abstract

This invention discloses a recommended method for intelligent asset auditing and compliance analysis, relating to the fields of asset management, audit analysis, and compliance assessment. The method includes: acquiring asset data related to asset management from multiple source data sources within an enterprise, and cleaning and preprocessing the asset data to obtain consistent quality asset data; using machine learning algorithms to perform audit analysis on the asset data, identifying anomalies in asset usage efficiency and maintenance costs, and potential audit risks, thereby obtaining audit analysis results; and based on laws, regulations, and enterprise management standards, using natural language processing technology to perform compliance analysis, obtaining asset compliance analysis results. This invention enables enterprises to automate and intelligently manage assets, thereby improving audit accuracy and compliance, reducing operational risks, and optimizing asset usage and maintenance strategies.

Description

Technical Field

[0001] This invention relates to the fields of asset management, audit analysis, and compliance assessment, and more specifically, to a recommended method for intelligent asset auditing and compliance analysis. Background Technology

[0002] In asset management, auditing and compliance analysis are crucial steps in ensuring that asset use and management comply with laws, regulations, and corporate management standards. As companies grow and their operations diversify, the complexity of asset management increases, and traditional auditing methods face challenges such as inefficiency, insufficient accuracy, and difficulty in adapting to rapidly changing compliance requirements.

[0003] Current asset auditing processes largely rely on manual operations, which are not only time-consuming and labor-intensive but also susceptible to subjective judgment, leading to biased audit results. Meanwhile, compliance analysis requires professional knowledge and a deep understanding of regulations to ensure that companies remain compliant in an ever-changing legal environment. However, with increasingly sophisticated regulations, companies face enormous pressure in compliance management.

[0004] There are currently no effective solutions to the problems in the relevant technologies. Summary of the Invention

[0005] In response to the problems in related technologies, this invention proposes a recommended method for intelligent asset auditing and compliance analysis to overcome the aforementioned technical problems existing in existing related technologies.

[0006] Therefore, the specific technical solution adopted by the present invention is as follows:

[0007] A recommended method for intelligent asset auditing and compliance analysis includes:

[0008] Asset data related to asset management is obtained from multiple sources of data from the enterprise, and the asset data is cleaned and preprocessed to obtain asset data of consistent quality.

[0009] Machine learning algorithms are used to audit and analyze asset data, identify abnormal asset usage efficiency, maintenance costs and potential audit risks, and obtain audit analysis results.

[0010] Based on laws, regulations, and corporate management standards, and using natural language processing technology for compliance analysis, the asset compliance analysis results are obtained.

[0011] Based on the audit and compliance analysis results, customized improvement suggestions and action plans are provided to asset managers;

[0012] By leveraging learning mechanisms to adaptively optimize based on new data and user feedback, audit and compliance analysis standards are dynamically adjusted.

[0013] Furthermore, asset data related to asset management is obtained from multiple source data sources within the enterprise, and this data is cleaned and preprocessed to obtain consistent quality asset data, including:

[0014] Based on multiple data sources for the enterprise, a set of data sources is defined, and the same asset data is queried through the database to ensure the automatic acquisition and initial preparation of asset data.

[0015] Data cleaning rules are used to remove duplicates, correct formats, and fill in missing values ​​in the synchronized asset data to obtain cleaned asset data.

[0016] Based on consistency check rules, logical, numerical, temporal and format consistency checks are performed on asset data from different data sources to ensure that asset data remains accurate after multi-source fusion.

[0017] Based on the cleaned and consistent asset data, the numerical attributes of the asset data are processed by weighted fusion, and the categorical attributes of the asset data are processed by majority voting mechanism to obtain asset data with consistent quality.

[0018] Furthermore, the formula for weighted fusion is:

[0019] ;

[0020] In the formula, S This represents the result value of the merged numerical attribute. oh i Indicates the first i The weight of each data source; c i Indicates the first i Confidence level of each data source; Indicates the first i Numerical attribute values ​​that have been cleaned, formatted, and preprocessed from a data source; n This indicates the total number of data sources participating in the integration.

[0021] Furthermore, machine learning algorithms are used to conduct audit analysis on asset data to identify anomalies in asset utilization efficiency and maintenance costs, as well as potential audit risks. The audit analysis results include:

[0022] Based on asset management data, a feature matrix is ​​constructed and standardized to obtain feature data of a uniform scale.

[0023] By using feature engineering methods, key nonlinear features are extracted from asset data, and the data dimensionality is reduced.

[0024] An anomaly detection model is constructed and trained using extracted key nonlinear features. The reconstruction error of each key nonlinear feature is calculated, and the anomaly threshold is determined.

[0025] Based on asset utilization efficiency indicators, time series analysis is used to predict asset utilization trends and obtain asset efficiency classification results.

[0026] Based on maintenance cost indicators and cost-benefit ratios, determine whether there are any abnormalities in the maintenance costs of the assets;

[0027] By combining maintenance costs and usage efficiency, a risk scoring function is defined to calculate the risk score of an asset, thereby obtaining the identification results of potential audit risks.

[0028] Furthermore, based on asset utilization efficiency indicators, time series analysis is used to predict asset utilization trends, resulting in asset efficiency classifications including:

[0029] Based on the asset usage data, the monthly usage for the year is obtained, and time series analysis is performed to obtain the annual baseline value of asset usage.

[0030] Using the annual usage baseline, calculate the monthly efficiency status ratio to obtain the asset's efficiency status, and classify the efficiency based on the comparison between the efficiency status and the annual usage baseline.

[0031] Based on the obtained efficiency status ratio, a three-level classification standard is established, and assets are divided into high-efficiency, normal, and low-efficiency categories, thus obtaining the efficiency classification results of assets.

[0032] Furthermore, the formula for calculating the annual usage benchmark value of the asset is as follows:

[0033] ;

[0034] In the formula, B Indicates the benchmark value for the annual usage of the asset; U t This indicates the monthly usage.

[0035] Furthermore, the formula for calculating the monthly efficiency status ratio is as follows:

[0036] ;

[0037] In the formula, S t This indicates the percentage of efficiency status each month; U t This indicates the monthly usage. B This represents the baseline value for the annual usage of an asset.

[0038] Furthermore, based on laws, regulations, and corporate management standards, and utilizing natural language processing technology for compliance analysis, the asset compliance analysis results include:

[0039] A legal knowledge base is constructed, which collects a set of texts of laws, regulations and corporate management standards related to asset management. The text set is preprocessed to obtain structured legal text data.

[0040] Natural language processing technology is used to extract keywords representing compliance rules from legal and regulatory texts;

[0041] Compliance rules are presented as a series of conditional statements. Each conditional statement defines the compliance standards for asset use and management, and a weight is assigned to each compliance rule.

[0042] Asset management data is used as a set of feature vectors, and based on data featureization, asset management data is transformed into the form of feature vectors to ensure that each asset management feature vector can comprehensively describe its use and management.

[0043] Based on the matching degree between the weights of compliance rules and asset feature vectors, a compliance scoring model is constructed to calculate the compliance score for each asset feature vector, thus obtaining the compliance score for each asset feature vector.

[0044] Set a compliance score threshold. If the compliance score of an asset feature vector is less than the threshold, it is judged as non-compliant; otherwise, it is judged as compliant. For each asset feature vector, output its compliance score and decision result, and generate a compliance report.

[0045] Furthermore, a regulatory knowledge base is constructed, collecting a text set of laws, regulations, and corporate management standards related to asset management. This text set is preprocessed to obtain structured regulatory text data, including:

[0046] A legal knowledge base is constructed, which collects a set of texts of laws, regulations and corporate management standards related to asset management. Redundant characters are removed, and the texts are standardized in terms of encoding and capitalization to obtain cleaned legal texts.

[0047] Natural language processing technology is used to segment legal and regulatory texts, optimize the segmentation of professional terms, and remove general and domain stop words to obtain high-quality legal and regulatory word segmentation results.

[0048] Based on the obtained legal and regulatory word segmentation results, part-of-speech tagging tools are used to annotate grammatical attributes, extract key phrases and semantic information, and obtain structured legal and regulatory text data.

[0049] Furthermore, the formula for calculating the compliance score for each asset feature vector is as follows:

[0050] ;

[0051] In the formula, m i Indicates the first i The weight of each rule; r i , f ) represents the asset feature vector f and rules r i The degree of matching.

[0052] The beneficial effects of this invention are as follows:

[0053] 1. This invention improves data quality by integrating and cleaning multi-source asset data; it intelligently identifies asset usage efficiency, maintenance cost anomalies, and compliance issues by combining machine learning and natural language processing; it provides customized improvement suggestions based on audit and compliance analysis results; and it enhances the system's adaptability and accuracy by dynamically optimizing analysis standards through a learning mechanism, thereby comprehensively improving the intelligence and compliance level of asset management. This enables enterprises to automate and intelligently manage assets, thereby improving the accuracy and compliance of audits, reducing operational risks, and optimizing asset usage and maintenance strategies.

[0054] 2. This invention improves data accuracy and integrity through multi-source asset data cleaning and consistency detection; it effectively solves data conflict and heterogeneity problems by adopting weighted fusion and majority voting mechanisms, ensuring the uniformity and reliability of asset data quality; and it provides high-quality data support for subsequent intelligent auditing and compliance analysis, enhancing the accuracy of analysis results and decision support capabilities.

[0055] 3. This invention accurately extracts asset usage trends through time series analysis based on asset utilization efficiency indicators; it scientifically classifies assets into high-efficiency, normal, and low-efficiency categories by combining efficiency status ratios and annual benchmark values, thereby improving the accuracy of asset operation status identification; it provides strong support for asset management optimization and resource allocation decisions, and enhances management efficiency.

[0056] 4. This invention extracts compliance rules by constructing a regulatory knowledge base and applying natural language processing technology; it quantifies the level of asset compliance by combining asset feature vectors with rule matching, accurately identifies non-compliant assets, and automatically generates compliance analysis reports, thereby improving the efficiency of compliance detection and decision support capabilities in asset management and assisting enterprises in risk control. Attached Figure Description

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0058] Figure 1 This is a flowchart of a recommended method for intelligent asset auditing and compliance analysis according to an embodiment of the present invention. Detailed Implementation

[0059] To further illustrate the various embodiments, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. These drawings are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these drawings, those skilled in the art should be able to understand other possible implementation methods and the advantages of the present invention.

[0060] According to embodiments of the present invention, a recommended method for intelligent asset auditing and compliance analysis is provided.

[0061] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, the recommended method for intelligent asset auditing and compliance analysis according to an embodiment of the present invention includes:

[0062] S1. Obtain asset data related to asset management from multiple source data sources of the enterprise, and clean and preprocess the asset data to obtain asset data of consistent quality.

[0063] Specifically, it automatically collects asset management-related data from multiple data sources within the enterprise, including but not limited to asset purchase records, depreciation information, maintenance logs, usage status, and historical transactions. It features data cleaning and preprocessing capabilities to ensure the quality and consistency of the input data.

[0064] S2. Use machine learning algorithms to audit and analyze asset data, identify abnormal asset usage efficiency, maintenance costs and potential audit risks, and obtain audit analysis results.

[0065] Specifically, machine learning algorithms are used to conduct in-depth analysis of the collected asset data to identify asset utilization efficiency, maintenance costs, and potential audit risks. This enables the automatic detection of data anomalies and inconsistencies, providing a basis for subsequent compliance analysis.

[0066] S3. Based on laws, regulations, and corporate management standards, and using natural language processing technology, compliance analysis is conducted to obtain asset compliance analysis results;

[0067] Specifically, by combining the latest laws, regulations, and corporate management standards, the use and management of assets are assessed to determine whether they comply with the prescribed standards. Natural language processing technology is used to understand and interpret complex regulatory texts and apply them to actual asset management activities.

[0068] S4. Based on the audit analysis results and compliance analysis results, provide asset managers with customized improvement suggestions and action plans;

[0069] Specifically, based on the results of intelligent auditing and compliance analysis, customized improvement suggestions and action plans are provided to asset managers. These suggestions may include asset reallocation, maintenance plan adjustments, and compliance risk mitigation measures. User feedback is also taken into account to continuously optimize the relevance and accuracy of the recommendations.

[0070] S5. Utilize learning mechanisms to adaptively optimize based on new data and user feedback, and dynamically adjust audit and compliance analysis standards.

[0071] Specifically, through self-learning and optimization mechanisms, the standards for auditing and compliance analysis can be continuously adjusted based on new data and user feedback, ensuring the continued effectiveness and adaptability of the recommended methods.

[0072] In this optional embodiment, asset data related to asset management is obtained from multiple source data sources of the enterprise, and the asset data is cleaned and preprocessed to obtain asset data of consistent quality, including:

[0073] Based on multiple data sources for the enterprise, a set of data sources is defined, and the same asset data is queried through the database to ensure the automatic acquisition and initial preparation of asset data.

[0074] Data cleaning rules are used to remove duplicates, correct formats, and fill in missing values ​​in the synchronized asset data to obtain cleaned asset data.

[0075] Based on consistency check rules, logical, numerical, temporal and format consistency checks are performed on asset data from different data sources to ensure that asset data remains accurate after multi-source fusion.

[0076] Based on the cleaned and consistent asset data, the numerical attributes of the asset data are processed by weighted fusion, and the categorical attributes of the asset data are processed by majority voting mechanism to obtain asset data with consistent quality.

[0077] Specifically, multi-source data automatic synchronization and preprocessing algorithms include:

[0078] 1) Data source identification and synchronization:

[0079] Define data source collection D ={ D 1,D 2, ..., D n},in D i Representing the i One data source.

[0080] Use API calls or database queries to periodically synchronize data from each data source.

[0081] 2) Data cleaning:

[0082] For each data source D i Define data cleaning rules R i This includes removing duplicate records, correcting formatting errors, and filling in missing values.

[0083] Application rules R i To data D i Obtain the cleaned data The formula is:

[0084] .

[0085] 3) Data consistency check:

[0086] The design uses consistency rules, a set of predefined logical constraints, to ensure that identical or related data from different data sources remain consistent across logical, numerical, and temporal dimensions. Its design goal is to automatically detect and resolve conflicts between multi-source data, thereby guaranteeing the accuracy and reliability of the merged data.

[0087] Field value consistency rule: Key fields of the same entity should be completely consistent across different data sources.

[0088] Logical consistency rules stipulate that the logical relationships between data should conform to business rules or mathematical formulas.

[0089] The time consistency rule stipulates that time-related data should conform to the time sequence or timeliness.

[0090] The format consistency rule stipulates that the format of the same field should be consistent across all data sources.

[0091] Inconsistent data will be resolved according to predefined priority rules or through manual review.

[0092] If data conflicts can be resolved using predefined rules (such as the authority of the data source, the priority of timestamps, etc.), then the priority rule should be used first. This approach is suitable for data conflicts that can be clearly identified, such as when some data sources are more reliable or certain fields have stricter formatting rules.

[0093] If data conflicts cannot be resolved through priority rules, or if more complex judgments are required (such as those involving subjective judgments or requiring further verification), then manual review should be used to resolve them.

[0094] 4) Data fusion:

[0095] For numerical attributes (such as asset valuation and maintenance costs), the weighted fusion formula is as follows:

[0096] .

[0097] In the formula, S This represents the final value of the merged numerical attribute result, which is the result of weighted calculations from multiple data sources. oh i Indicates the first i The weight of each data source indicates its credibility or importance; the higher the weight, the greater its contribution to the final result. c i Indicates the first i The confidence level of a data source reflects the quality or reliability of the data source; Indicates the first i Numerical attribute values ​​that have been cleaned, formatted, and preprocessed from a data source; n This indicates the total number of data sources participating in the integration.

[0098] For categorical attributes in asset management (such as asset status, equipment type, etc.), data conflict issues are resolved through multi-source data fusion. This is combined with data source weights (…). oh i ) and confidence level ( c i We will design a dynamically adjusted majority voting mechanism to ensure that the fusion results reflect both the consistency of the majority of data sources and the authority of high-weight and high-confidence data sources.

[0099] From each data source D i Extract the cleaned and formatted categorical attribute values v i ,verify v i Does it conform to the predefined set of legal values? V Illegal values ​​are marked as "unknown" and excluded from voting.

[0100] Calculate the voting weight of each data source. W i Due to its basic weight oh i and confidence level c i Joint decision:

[0101] .

[0102] Perform weighted majority voting for each candidate value. v ∈ V Calculate the total voting weight and select the candidate with the highest value as the final result:

[0103] .

[0104] In this optional embodiment, machine learning algorithms are used to perform audit analysis on asset data to identify abnormal asset utilization efficiency, maintenance costs, and potential audit risks. The audit analysis results include:

[0105] Based on asset management data, a feature matrix is ​​constructed and standardized to obtain feature data of a uniform scale.

[0106] By using feature engineering methods, key nonlinear features are extracted from asset data, and the data dimensionality is reduced.

[0107] An anomaly detection model is constructed and trained using extracted key nonlinear features. The reconstruction error of each key nonlinear feature is calculated, and the anomaly threshold is determined.

[0108] Based on asset utilization efficiency indicators, time series analysis is used to predict asset utilization trends and obtain asset efficiency classification results.

[0109] Based on maintenance cost indicators and cost-benefit ratios, determine whether there are any abnormalities in the maintenance costs of the assets;

[0110] By combining maintenance costs and usage efficiency, a risk scoring function is defined to calculate the risk score of an asset, thereby obtaining the identification results of potential audit risks.

[0111] In this optional embodiment, based on asset utilization efficiency indicators, time series analysis is used to predict asset utilization trends, resulting in asset efficiency classification results including:

[0112] Based on the asset usage data, the monthly usage for the year is obtained, and time series analysis is performed to obtain the annual baseline value of asset usage.

[0113] Using the annual usage baseline, calculate the monthly efficiency status ratio to obtain the asset's efficiency status, and classify the efficiency based on the comparison between the efficiency status and the annual usage baseline.

[0114] Based on the obtained efficiency status ratio, a three-level classification standard is established, and assets are divided into high-efficiency, normal, and low-efficiency categories, thus obtaining the efficiency classification results of assets.

[0115] Specifically, intelligent asset management audit algorithms based on anomaly detection include:

[0116] 1) Data representation:

[0117] Asset management data is represented as a feature matrix. X ∈ ,in m It refers to the number of data points (asset records), such as usage frequency and maintenance costs.

[0118] The input data is standardized to eliminate differences in dimensions.

[0119] 2) Feature engineering:

[0120] By applying feature engineering methods, key nonlinear features in asset management data are extracted through an autoencoder, and the data dimensionality is reduced, providing highly informative feature representations for subsequent anomaly detection and risk assessment.

[0121] 3) Anomaly detection model:

[0122] Train an anomaly detection model, such as an Isolation Forest, a One-Class SVM, or an Autoencoder.

[0123] High-information feature matrix output using feature engineering X ∈ (e.g., dimensionality reduction features extracted by an autoencoder).

[0124] Standardize the feature matrix (e.g., Z-score standardization) to ensure model convergence and stability.

[0125] Training data: The model is trained using only normal data (unlabeled) so that it learns the distribution pattern of normal data.

[0126] Calculate the reconstruction error:

[0127] For each sample X i Calculate the MSE of its input and output.

[0128] Determine the outlier threshold based on the reconstruction error distribution of normal samples (e.g., the 95th percentile). i .

[0129] Model evaluation:

[0130] Accuracy: The proportion of correctly identified outlier samples.

[0131] Recall: The proportion of actual anomalous samples that are correctly identified.

[0132] Use a labeled test set (containing known outliers) to validate the model performance.

[0133] 4) Usage efficiency analysis:

[0134] Define usage efficiency metrics, such as the ratio of asset usage frequency to expected usage frequency.

[0135] Time series analysis is used to comprehensively predict asset usage trends and identify inefficient assets.

[0136] Data preparation: Collect the most recent data from the assets. t Period (default) t Monthly usage data (annual monthly usage) for 12 months:

[0137] .

[0138] In the formula, U This indicates the total amount of assets used. U t This indicates the monthly usage.

[0139] To calculate the annual benchmark value (annual usage benchmark value of the asset), the benchmark value is extended to a 12-month horizontal linear series. The formula for calculating the annual benchmark value (annual usage benchmark value of the asset) is as follows:

[0140] .

[0141] In the formula, B Indicates the benchmark value for the annual usage of the asset; U t This indicates the monthly usage.

[0142] Efficiency registration classification: Calculate the proportion of efficiency states S t Establish a three-level classification standard and calculate the monthly efficiency status ratio (calculate the efficiency status ratio). S t The formula for ) is:

[0143] .

[0144] In the formula, S t This indicates the percentage of efficiency status each month; U t This indicates the monthly usage. B This represents the baseline value for the annual usage of an asset.

[0145] High efficiency (high efficiency category): Efficiency level ≥ 80%.

[0146] Normal (Normal Class): Efficiency State ≤ 50% < 80%.

[0147] Inefficient (Inefficient category): Efficiency level <50%.

[0148] 5) Maintenance cost analysis:

[0149] Define maintenance cost metrics M For example, the ratio of asset maintenance costs to asset value.

[0150] Cost-benefit analysis is used to determine whether the maintenance costs of the assets are within a reasonable range.

[0151] Input data:

[0152] Maintenance cost data: cost per maintenance session, total annual maintenance cost (from maintenance log).

[0153] Asset value data: original value of the asset, current valuation, and residual value (from asset acquisition records and depreciation information).

[0154] Key performance indicator calculation:

[0155] Maintenance cost ratio ( M ):

[0156] .

[0157] Cost-benefit ratio ( CBR ):

[0158] .

[0159] Determination of reasonable range:

[0160] If the current assets M If the value exceeds ±2 standard deviations of the cluster center, it is marked as an anomaly.

[0161] Reasonable settings CBR Threshold (e.g.) CBR ≤1 indicates that the benefit is greater than or equal to the cost, which is considered reasonable.

[0162] like CBRIf the value is greater than 1, it is determined that "maintenance costs are uneconomical".

[0163] 6) Risk assessment:

[0164] Define risk scoring function R ( x Risk scores are calculated based on asset utilization efficiency, maintenance costs, and other relevant factors.

[0165] Design a threshold i A risk score higher than i The assets were marked as potential audit risks.

[0166] In this optional embodiment, based on laws, regulations, and corporate management standards, and utilizing natural language processing technology for compliance analysis, the asset compliance analysis results include:

[0167] A legal knowledge base is constructed, which collects a set of texts of laws, regulations and corporate management standards related to asset management. The text set is preprocessed to obtain structured legal text data.

[0168] Natural language processing technology is used to extract keywords representing compliance rules from legal and regulatory texts;

[0169] Compliance rules are presented as a series of conditional statements. Each conditional statement defines the compliance standards for asset use and management, and a weight is assigned to each compliance rule.

[0170] Asset management data is used as a set of feature vectors, and based on data featureization, asset management data is transformed into the form of feature vectors to ensure that each asset management feature vector can comprehensively describe its use and management.

[0171] Based on the matching degree between the weights of compliance rules and asset feature vectors, a compliance scoring model is constructed to calculate the compliance score for each asset feature vector, thus obtaining the compliance score for each asset feature vector.

[0172] Set a compliance score threshold. If the compliance score of an asset feature vector is less than the threshold, it is judged as non-compliant; otherwise, it is judged as compliant. For each asset feature vector, output its compliance score and decision result, and generate a compliance report.

[0173] In this optional embodiment, a regulatory knowledge base is constructed, collecting a text set of laws, regulations, and corporate management standards related to asset management. The text set is preprocessed to obtain structured regulatory text data, including:

[0174] A legal knowledge base is constructed, which collects a set of texts of laws, regulations and corporate management standards related to asset management. Redundant characters are removed, and the texts are standardized in terms of encoding and capitalization to obtain cleaned legal texts.

[0175] Natural language processing technology is used to segment legal and regulatory texts, optimize the segmentation of professional terms, and remove general and domain stop words to obtain high-quality legal and regulatory word segmentation results.

[0176] Based on the obtained legal and regulatory word segmentation results, part-of-speech tagging tools are used to annotate grammatical attributes, extract key phrases and semantic information, and obtain structured legal and regulatory text data.

[0177] Specifically, compliance analysis algorithms based on rules and machine learning:

[0178] 1) Construction of a legal knowledge base:

[0179] Create a regulatory knowledge base containing a collection of texts related to laws, regulations, and corporate management standards relevant to asset management.

[0180] 2) Text preprocessing:

[0181] Preprocessing of text in the legal knowledge base includes word segmentation, stop word removal, and part-of-speech tagging. Natural language processing includes:

[0182] Remove redundant characters, HTML tags, special symbols (such as semicolons), and illegal characters.

[0183] Unified encoding and case sensitivity: Convert text to UTF-8 encoding and standardize it to lowercase (for English scenarios) or retain case sensitivity (for Chinese scenarios).

[0184] Use open-source tools (such as jieba) and a custom dictionary to optimize the segmentation of technical terms.

[0185] General stop word list (such as NLTK's stopwords) + domain-specific stop words (such as redundant legal terms like "the", "article", "according to").

[0186] Use the POS annotation function of LTP, HanLP, or jieba.

[0187] 3) Feature extraction:

[0188] Keyword extraction involves extracting keywords from regulatory texts that can characterize compliance rules.

[0189] Phrase and Clause Number Extraction: Extract phrases and clause numbers from the regulatory text.

[0190] 4) Compliance rules are expressed as follows:

[0191] The compliance rules are represented as a series of conditional statements, each of which defines the compliance standards for the use and management of assets.

[0192] Represent the rule as a "condition → action" form, for example:

[0193] Condition: The asset's depreciation period exceeds 10 years.

[0194] Action: Asset disposal must be carried out.

[0195] Assign a weight to each rule; the weight can be determined based on the priority of the regulations or historical audit data.

[0196] 5) Asset management data representation:

[0197] Asset management data is represented as a set of feature vectors, where each vector... x i It describes the use and management of an asset.

[0198] Data characterization, which transforms asset management data into feature vectors, includes:

[0199] Define a set of features for asset management data, for example:

[0200] Asset ID, purchase date, depreciation information, and maintenance records.

[0201] The frequency of asset use, maintenance costs, and current status.

[0202] Asset management data is mapped to feature vectors to ensure that the feature vector of each asset can comprehensively describe its usage and management.

[0203] 6) Compliance scoring model:

[0204] Construct a compliance scoring model S:X->[0,1], where the model is for each asset management feature vector. x i Calculate a compliance score.

[0205] The formula for defining the compliance score (calculated for each asset feature vector) is as follows:

[0206] .

[0207] In the formula, m i Indicates the first i The weight of each rule; r i , f ) represents the asset feature vector fand rules r i The degree of matching.

[0208] 7) Compliance decisions:

[0209] A compliance threshold τ is set. If the compliance score of an asset is less than τ, the asset management activity is considered non-compliant.

[0210] For each asset, output its compliance score and decision result (compliant / non-compliant).

[0211] Generate a compliance report detailing the reasons for non-compliant assets and providing recommendations for improvement.

[0212] In summary, by utilizing the above-mentioned technical solutions of this invention, the present invention improves data accuracy and integrity through multi-source asset data cleaning and consistency detection; it effectively resolves data conflicts and heterogeneity issues by employing weighted fusion and majority voting mechanisms, ensuring the unified and reliable quality of asset data; and it provides high-quality data support for subsequent intelligent auditing and compliance analysis, enhancing the accuracy of analysis results and decision support capabilities. This invention accurately extracts asset usage trends through time-series analysis based on asset usage efficiency indicators; it scientifically classifies assets into efficient, normal, and inefficient categories by combining efficiency status ratios and annual benchmark values, improving the accuracy of asset operation status identification; and it provides strong support for asset management optimization and resource allocation decisions, improving management efficiency. This invention extracts compliance rules by constructing a regulatory knowledge base and applying natural language processing technology; it quantifies asset compliance levels by combining asset feature vectors with rule matching, accurately identifying non-compliant assets; and it automatically generates compliance analysis reports, improving the efficiency of compliance detection and decision support capabilities in asset management, and assisting enterprises in risk control.

[0213] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A recommended method for intelligent asset auditing and compliance analysis, characterized in that, include: Asset data related to asset management is obtained from multiple source data sources of the enterprise, and the asset data is cleaned and preprocessed to obtain asset data of consistent quality. Machine learning algorithms are used to perform audit analysis on asset data to identify abnormal asset usage efficiency, maintenance costs, and potential audit risks, and to obtain audit analysis results. Based on laws, regulations, and corporate management standards, and using natural language processing technology for compliance analysis, the asset compliance analysis results are obtained. Based on the audit and compliance analysis results, customized improvement suggestions and action plans are provided to asset managers; Utilize learning mechanisms to adaptively optimize based on new data and user feedback, and dynamically adjust audit and compliance analysis standards. The use of machine learning algorithms to audit and analyze asset data identifies abnormal asset utilization efficiency, maintenance costs, and potential audit risks, and the audit analysis results include: Based on asset management data, a feature matrix is ​​constructed and standardized to obtain feature data of a uniform scale. By using feature engineering methods, key nonlinear features are extracted from asset data, and the data dimensionality is reduced. An anomaly detection model is constructed and trained using extracted key nonlinear features. The reconstruction error of each key nonlinear feature is calculated, and the anomaly threshold is determined. Based on asset utilization efficiency indicators, time series analysis is used to predict asset utilization trends and obtain asset efficiency classification results. Based on maintenance cost indicators and cost-benefit ratios, determine whether there are any abnormalities in the maintenance costs of the assets; By combining maintenance costs and usage efficiency, a risk scoring function is defined to calculate the risk score of an asset, thereby obtaining the identification results of potential audit risks.

2. The recommendation method for intelligent asset auditing and compliance analysis according to claim 1, characterized in that, The process of acquiring asset management-related asset data from multiple source data sources within an enterprise, and then cleaning and preprocessing this asset data to obtain consistent quality asset data includes: Based on multiple data sources for the enterprise, a set of data sources is defined, and the same asset data is queried through the database to ensure the automatic acquisition and initial preparation of asset data. Data cleaning rules are used to remove duplicates, correct formats, and fill in missing values ​​in the synchronized asset data to obtain cleaned asset data. Based on consistency check rules, logical, numerical, temporal and format consistency checks are performed on asset data from different data sources to ensure that asset data remains accurate after multi-source fusion. Based on the cleaned and consistent asset data, the numerical attributes of the asset data are processed by weighted fusion, and the categorical attributes of the asset data are processed by majority voting mechanism to obtain asset data with consistent quality.

3. The recommendation method for intelligent asset auditing and compliance analysis according to claim 2, characterized in that, The formula for the weighted fusion is: ； In the formula, S This represents the result value of the merged numerical attribute. ω i Indicates the first i The weight of each data source; c i Indicates the first i Confidence level of each data source; Indicates the first i Numerical attribute values ​​that have been cleaned, formatted, and preprocessed from a data source; n This indicates the total number of data sources participating in the integration.

4. The recommendation method for intelligent asset auditing and compliance analysis according to claim 1, characterized in that, The asset efficiency index, based on time series analysis, predicts asset usage trends and yields asset efficiency classification results, including: Based on the asset usage data, the monthly usage for the year is obtained, and time series analysis is performed to obtain the annual baseline value of asset usage. Using the annual usage baseline, calculate the monthly efficiency status ratio to obtain the asset's efficiency status, and classify the efficiency based on the comparison between the efficiency status and the annual usage baseline. Based on the obtained efficiency status ratio, a three-level classification standard is established, and assets are divided into high-efficiency, normal, and low-efficiency categories, resulting in the asset efficiency classification results.

5. The recommendation method for intelligent asset auditing and compliance analysis according to claim 4, characterized in that, The formula for calculating the annual usage benchmark value of the asset is as follows: ； In the formula, B Indicates the benchmark value for the annual usage of the asset; U t This indicates the monthly usage.

6. The recommendation method for intelligent asset auditing and compliance analysis according to claim 4, characterized in that, The formula for calculating the monthly efficiency status ratio is as follows: ； In the formula, S t This indicates the percentage of efficiency status each month; U t This indicates the monthly usage. B This represents the baseline value for the annual usage of an asset.

7. The recommendation method for intelligent asset auditing and compliance analysis according to claim 1, characterized in that, The compliance analysis results obtained, based on laws, regulations, and corporate management standards, and utilizing natural language processing technology, include: A legal knowledge base is constructed, which collects a set of texts of laws, regulations and corporate management standards related to asset management. The text set is preprocessed to obtain structured legal text data. Natural language processing technology is used to extract keywords representing compliance rules from legal and regulatory texts; Compliance rules are presented as a series of conditional statements. Each conditional statement defines the compliance standards for asset use and management, and a weight is assigned to each compliance rule. Asset management data is used as a set of feature vectors, and based on data featureization, asset management data is transformed into the form of feature vectors to ensure that each asset management feature vector can comprehensively describe its use and management. Based on the matching degree between the weights of compliance rules and asset feature vectors, a compliance scoring model is constructed to calculate the compliance score for each asset feature vector, thus obtaining the compliance score for each asset feature vector. Set a compliance score threshold. If the compliance score of an asset feature vector is less than the threshold, it is judged as non-compliant; otherwise, it is judged as compliant. For each asset feature vector, output its compliance score and decision result, and generate a compliance report.

8. The recommendation method for intelligent asset auditing and compliance analysis according to claim 7, characterized in that, The construction of the regulatory knowledge base involves collecting a text set of laws, regulations, and corporate management standards related to asset management. This text set is then preprocessed to obtain structured regulatory text data, including: A legal knowledge base is constructed, which collects a set of texts of laws, regulations and corporate management standards related to asset management. Redundant characters are removed, and the texts are standardized in terms of encoding and capitalization to obtain cleaned legal texts. Natural language processing technology is used to segment legal and regulatory texts, optimize the segmentation of professional terms, and remove general and domain stop words to obtain high-quality legal and regulatory word segmentation results. Based on the obtained legal and regulatory word segmentation results, part-of-speech tagging tools are used to annotate grammatical attributes, extract key phrases and semantic information, and obtain structured legal and regulatory text data.

9. The recommendation method for intelligent asset auditing and compliance analysis according to claim 7, characterized in that, The formula for calculating the compliance score for each asset feature vector is as follows: ； In the formula, μ i Indicates the first i The weight of each rule; ( r i , f ) represents the asset feature vector f With rules r i The degree of matching.

Images