CN115794803B - Engineering audit problem monitoring method and system based on big data AI technology - Google Patents
Engineering audit problem monitoring method and system based on big data AI technology Download PDFInfo
- Publication number
- CN115794803B CN115794803B CN202310045164.5A CN202310045164A CN115794803B CN 115794803 B CN115794803 B CN 115794803B CN 202310045164 A CN202310045164 A CN 202310045164A CN 115794803 B CN115794803 B CN 115794803B
- Authority
- CN
- China
- Prior art keywords
- audit
- data
- information
- target data
- extracted
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The invention provides an engineering audit problem monitoring method and system based on big data AI technology, comprising the following steps: determining audit requirements according to the collected audit information; inputting the audit requirements into a pre-constructed audit association model to obtain audit target data; inputting the audit target data into a pre-constructed audit autonomous supervision model to obtain an audit monitoring result. The invention can widen the auditing range and promote the overall development of auditing supervision through the application of the data identification and data mining technology in the auditing project, and the data mining technology can realize the monitoring of auditing problems through the data description, deduction and mining of deep and hidden relations in the audited unit database, and the mining and extraction of unknown data operation modes.
Description
Technical Field
The invention relates to the technical field of audit problem monitoring, in particular to an engineering audit problem monitoring method and system based on a big data AI technology.
Background
At present, due to the fact that materials in engineering project construction processes are various and huge in quantity, but electronic file management systems in China are incomplete, poor phenomena such as confusion and incomplete information of electronic file management exist in the engineering construction field, even a large number of traditional paper files cannot be converted into electronic file management, current management requirements cannot be met, and effective grippers are lacked in real-time supervision of engineering construction overall process compliance. Along with the continuous promotion of informatization degree, although some enterprises manage cost, finance, contract and the like through an intelligent management system, the whole data integration is poor, information is split, the data quality is low, and effective data extraction and utilization are difficult. The project participants such as design, construction, owners, suppliers and the like can not unify the interfaces of data, so that associable main data is lacking, the enterprise has no structured data to be used, the supervision quality of the whole engineering construction process is low, and high management risks exist.
Disclosure of Invention
The invention provides an engineering audit problem monitoring method and system based on big data AI technology, comprising the following steps:
determining audit requirements according to the collected audit information;
inputting the audit requirements into a pre-constructed audit association model to obtain audit target data;
inputting the audit target data into a pre-constructed audit autonomous supervision model to obtain an audit monitoring result;
the audit association model construction comprises the following steps:
feature extraction is carried out according to the audit demand, and features to be extracted are determined;
feature selection is carried out according to the audit demand, and the category to be extracted is determined;
based on a preset audit database, carrying out data extraction according to the characteristics to be extracted and the categories to be extracted, and carrying out data cleaning and data duplication elimination on the extracted data to obtain associated data;
and carrying out association degree calculation on the association data and the audit demands, screening the association data with association degree calculation results larger than a preset threshold value as audit target data, and constructing an audit association model.
Preferably, the constructing of the audit autonomous supervision model includes:
performing multi-mode self-supervision learning aiming at the audit target data, and marking problem data;
and carrying out text classification on the problem data according to a preset problem attribute value to obtain an audit monitoring result.
Preferably, the performing multi-mode self-supervised learning on the audit target data, labeling problem data, includes:
encoding the mask data of the audit target data to obtain primary encoded data;
based on the audit demand, carrying out weight setting on the audit target data to obtain weight setting information;
encoding the unmasked data in the audit target data to obtain secondary encoded data;
and determining problem data based on the primary coding data, the secondary coding data and the weight setting information, and marking the problem data in the audit target data.
Preferably, the determining the audit requirement according to the collected audit information includes:
collecting audit information, and carrying out structural conversion on the audit information to obtain primary conversion information;
carrying out multi-scale feature fusion on the primary conversion information to obtain fusion information;
extracting key information according to the fusion information, and determining that the audit information corresponds to an audit requirement;
wherein the audit information includes: audit objectives, audit scope, and audit content.
Preferably, the feature to be extracted includes: text features and symbolic features; the category to be extracted comprises: text part of speech and text attributes.
Based on the same inventive concept, the invention also provides an engineering audit problem monitoring system based on big data AI technology, comprising:
the demand determination module: the audit information acquisition module is used for acquiring audit information of the user;
a target data determining module: the audit target data are obtained by inputting the audit requirements into a pre-constructed audit association model;
audit monitoring module: the audit target data are input into a pre-constructed audit autonomous supervision model to obtain an audit monitoring result;
the construction of the audit association model in the target data determination module comprises the following steps:
feature extraction is carried out according to the audit demand, and features to be extracted are determined;
feature selection is carried out according to the audit demand, and the category to be extracted is determined;
based on a preset audit database, carrying out data extraction according to the characteristics to be extracted and the categories to be extracted, and carrying out data cleaning and data duplication elimination on the extracted data to obtain associated data;
and carrying out association degree calculation on the association data and the audit demands, screening the association data with association degree calculation results larger than a preset threshold value as audit target data, and constructing an audit association model.
Preferably, the constructing an audit autonomous supervision model in the audit monitoring module includes:
performing multi-mode self-supervision learning aiming at the audit target data, and marking problem data;
and carrying out text classification on the problem data according to a preset problem attribute value to obtain an audit monitoring result.
Preferably, the audit monitoring module performs multi-mode self-supervised learning for the audit target data, annotates problem data, and includes:
encoding the mask data of the audit target data to obtain primary encoded data;
based on the audit demand, carrying out weight setting on the audit target data to obtain weight setting information;
encoding the unmasked data in the audit target data to obtain secondary encoded data;
and determining problem data based on the primary coding data, the secondary coding data and the weight setting information, and marking the problem data in the audit target data.
Preferably, the requirement determining module is specifically configured to:
collecting audit information, and carrying out structural conversion on the audit information to obtain primary conversion information;
carrying out multi-scale feature fusion on the primary conversion information to obtain fusion information;
extracting key information according to the fusion information, and determining that the audit information corresponds to an audit requirement;
wherein the audit information includes: audit objectives, audit scope, and audit content.
Preferably, the target data determining module is configured to determine a feature to be extracted, including: text features and symbolic features; the category to be extracted comprises: text part of speech and text attributes.
Compared with the closest prior art, the invention has the following beneficial effects:
the invention provides an engineering audit problem monitoring method and system based on big data AI technology, comprising the following steps: determining audit requirements according to the collected audit information; inputting the audit requirements into a pre-constructed audit association model to obtain audit target data; inputting the audit target data into a pre-constructed audit autonomous supervision model to obtain an audit monitoring result; the audit association model construction comprises the following steps: feature extraction is carried out according to the audit demand, and features to be extracted are determined; feature selection is carried out according to the audit demand, and the category to be extracted is determined; based on a preset audit database, carrying out data extraction according to the characteristics to be extracted and the categories to be extracted, and carrying out data cleaning and data duplication elimination on the extracted data to obtain associated data; and carrying out association degree calculation on the association data and the audit demands, screening the association data with association degree calculation results larger than a preset threshold value as audit target data, and constructing an audit association model. The invention promotes the innovation of the electric auditing operation means through data identification and data mining, has wide coverage service range and large data volume; through the application of the data analysis technology in the audit project, the electric power audit means is changed from the traditional tool cultivation fire to the information technology high-tech means; the audit range is widened, the audit supervision is promoted to be fully developed, and the data mining technology is used for mining and extracting unknown data operation modes by describing, deducting and mining deep and hidden relations in the audited unit database to generate audit points. The original unexpected audit problem even without the unexpected audit problem can be found out by utilizing the big data monitoring technology, so that the problem of missing by an audit unit, even the artificial problem, can greatly enrich the audit theory and experience, enlarge the audit range, increase the audit depth, improve the audit quality, reduce the audit risk and enable the comprehensive audit to be possible; and the unknown business rule can be found out, the unknown business rule is found out in the massive power data, and an audit trail hidden in the massive data is found out through the business rule, so that data support service is provided for audit personnel to develop audit projects.
Additional features of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure particularly pointed out in the written description and drawings.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:
FIG. 1 is a schematic flow chart of an engineering audit problem monitoring method based on big data AI technology in an embodiment of the invention;
FIG. 2 is a schematic diagram of an audit flow scheme in accordance with an embodiment of the present invention;
FIG. 3 is a schematic flow chart of a data mining modeling technique employed in an embodiment of the present invention;
FIG. 4 is a schematic diagram of a decision tree model obtained by using a data mining modeling technique in an embodiment of the present invention;
fig. 5 is a schematic diagram of the structural composition of an engineering audit problem monitoring system based on big data AI technology in an embodiment of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, it being understood that the preferred embodiments described herein are for illustration and explanation of the present invention only, and are not intended to limit the present invention.
It should be noted that in this document, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Example 1
The embodiment of the invention provides an engineering audit problem monitoring method based on a big data AI technology, which is shown in a figure 1 and comprises the following steps:
determining audit requirements according to the collected audit information;
inputting the audit requirements into a pre-constructed audit association model to obtain audit target data;
inputting the audit target data into a pre-constructed audit autonomous supervision model to obtain an audit monitoring result;
the audit association model construction comprises the following steps:
feature extraction is carried out according to the audit demand, and features to be extracted are determined;
feature selection is carried out according to the audit demand, and the category to be extracted is determined;
based on a preset audit database, carrying out data extraction according to the characteristics to be extracted and the categories to be extracted, and carrying out data cleaning and data duplication elimination on the extracted data to obtain associated data;
and carrying out association degree calculation on the association data and the audit demands, screening the association data with association degree calculation results larger than a preset threshold value as audit target data, and constructing an audit association model.
The audit autonomous supervision model construction comprises the following steps:
performing multi-mode self-supervision learning aiming at the audit target data, and marking problem data;
and carrying out text classification on the problem data according to a preset problem attribute value to obtain an audit monitoring result.
The multi-mode self-supervision learning is performed on the audit target data, and the problem data is marked, including:
encoding the mask data of the audit target data to obtain primary encoded data;
based on the audit demand, carrying out weight setting on the audit target data to obtain weight setting information;
encoding the unmasked data in the audit target data to obtain secondary encoded data;
and based on the primary coding data, the secondary coding data and the weight setting information, marking the problem data in the audit target data.
Preferably, the determining the audit requirement according to the collected audit information includes:
collecting audit information, and carrying out structural conversion on the audit information to obtain primary conversion information;
carrying out multi-scale feature fusion on the primary conversion information to obtain fusion information;
extracting key information according to the fusion information, and determining that the audit information corresponds to an audit requirement;
wherein the audit information includes: audit objectives, audit scope, and audit content.
The feature to be extracted comprises: text features and symbolic features; the category to be extracted comprises: text part of speech and text attributes.
An embodiment is used for explaining the engineering audit problem monitoring method based on the big data AI technology, as shown in figure 2, comprising the following steps:
step 101: the collection and storage of audit data are carried out by first deeply knowing the basic condition of the audited unit and determining the specific audit requirement by analyzing the audit target, audit range and audit content. Collecting a large amount of unstructured data related to the audited units is a key step in audit data analysis, according to the explicit audit requirements. Firstly, data from the outside of an enterprise can be acquired by adopting a web crawler technology, OCR, API and other modes; on the other hand, unstructured data may be collected through various effective data access interfaces. In order to ensure the integrity and the authenticity of the audit data, a strict and standard system needs to be established, and the collected unstructured data is scientifically and safely managed;
step 102: the text mining of audit data comprises three parts, namely mining core content analyzed by feature extraction and feature selection, extracting words or phrases which can represent feature information of audit texts and can distinguish and identify the types of the audit texts by a vector model by the feature extraction, establishing a text representation model, and converting unstructured audit texts into expression forms which can be processed by a computer; the feature selection preferably adopts the feature with strong expressive power on the audit text content. The auditor further selects the text features extracted by the features through a feature selection method for audit purposes, so that junk data is avoided, and the effectiveness of the data is ensured. And secondly, mining audit analysis data through text abstracts, association analysis and text classification, and acquiring the abstracts of the audit texts through a mode of obtaining a central document by the text abstracts. Audit staff improves audit data analysis efficiency through a convenient browsing mode and a quick audit trail query method, and correlation analysis technology is mainly used for correlating knowledge in text mining. And carrying out association analysis on the audit unstructured text, researching a certain hidden relation possibly existing between audit documents, acquiring essential relation between audit matters and audit targets, and classifying the text to be supervised learning. The text clustering has the advantages that supervised learning is not needed, simulation is not needed through training, and some information which is difficult to find by a conventional audit data analysis method and is hidden in big data in a special form can be made up. After the audit texts are clustered, specific analysis, comparison and summarization can be carried out on each type of text according to the types, and the reasons for the existence of abnormal data are analyzed, so that the workload of audit personnel for audit data analysis is greatly reduced. Thirdly, the complex audit data is expressed in a manner which can be understood by a user. Knowledge acquired by text mining can be clearly displayed through visual symbols (network diagram, tree diagram, venn diagram, coordinates and the like), auditors can analyze, explain and evaluate visual results according to audit targets and evaluation indexes, and audit data are deeply observed and comprehensively multidimensional analyzed from different angles;
step 103: summarizing and publishing, the auditor and the technician jointly screen, classify, arrange and deeply analyze the results displayed by the visual analysis in the previous stage, summarize valuable and effective audit knowledge and standardize the valuable and effective audit knowledge, and form audit experience and audit clues for the auditor to analyze and evidence.
As shown in figure 3, the invention also adopts a data mining modeling technology to monitor engineering audit problems, and the electricity property authentication is essentially a classification problem, so that a classification algorithm in data mining can be used for pattern recognition. Because the decision tree model has the advantage of model result interpretability, a decision tree algorithm is selected for data mining. The modeling indexes mainly considered in the modeling process include the running capacity of a user, the time, the electricity consumption proportion of each day, the electricity consumption proportion of each hour and the electricity consumption property, the modeling data use the screened more accurate data of the electricity consumption property, and finally a decision tree algorithm is used for building an electricity consumption property identification model. The mining model is then used to predict those users that are uncertain, and identify the actual electricity usage properties of the users. Those models are predicted to be industrial or commercial, and users actually labeled as residents are used as audit trail users.
The method specifically comprises the following steps:
step 201: the idea of algorithms, decision tree learning, is an example-based generalized learning algorithm that focuses on reasoning out classification rules of the decision tree representation from a set of unordered, irregular cases, typically used to form classifiers and predictive models, which can classify or predict unknown data, data mining, etc. The method comprises the steps of establishing and refining a decision tree by using a training sample set, and establishing a decision tree model. This process is actually a process of acquiring knowledge from data and performing machine learning. The built decision tree classifies the new data. Among the various algorithms for decision tree classification, the ID3 algorithm proposed by Quinlan is the most influential, and based on the ID3 algorithm, he proposes a C4.5 algorithm.
The model adopts a classical C4.5 algorithm in the construction of a decision tree, the algorithm is proposed by quinlan.J.R in 1993, the model is developed from the algorithm, and the ID3 (Iterative Dichotomizer 3) algorithm is proposed by Quinlan in 1986. It is representative of decision tree algorithms, and most decision tree algorithms are realized by improving on the basis of the decision tree algorithms. When the attribute is selected on each level of nodes of the decision tree, the information gain is used as the selection standard of the attribute, so that the maximum category information about the tested record can be obtained when testing is carried out on each non-leaf node. Detecting all the attributes, selecting the attribute with the maximum information gain to generate a decision tree node, establishing branches by different values of the attribute, and recursively calling subsets of each branch to establish the branches of the decision tree node until all the subsets only contain data of the same category. Finally, a decision tree is obtained, which can classify the new sample.
The C4.5 algorithm inherits the advantages of the ID3 algorithm and introduces new methods and functions:
(1) the concept of the information gain rate is used, and the defect that the attribute is biased to multiple values when the attribute is selected by the information gain is overcome;
(2) pruning is carried out in the tree construction process so as to avoid overfitting of the tree;
(3) discretization processing of continuous attributes can be performed;
(4) a training sample set with missing attribute values may be processed;
(5) incomplete data can be processed;
(6) k, cross-validation;
(7) rule generation formula.
The C4.5 algorithm reduces the computational complexity and enhances the computational efficiency. An important improvement to the ID3 algorithm is the use of the information gain rate to select attributes. Theory and experiment show that the information gain rate is better than the information gain rate, and the attribute that the selection bias of the ID3 method is more is overcome. The C4.5 algorithm also processes the data of the continuous value attribute, and overcomes the defect that the ID3 algorithm can only process the data of the discrete value attribute.
Step 202: and (5) constructing a model. The method is used for summarizing the electricity consumption of the first month of certain city and characterizing the data by using five attributes, wherein the electricity consumption is running capacity, the electricity consumption per day of 1 month, the electricity consumption per day is the proportion of the electricity consumption of the whole month, and the electricity consumption of the whole point is the proportion of the electricity consumption of the whole month. The C4.5 algorithm is used for constructing a decision tree, and the decision tree model shown in figure 4 is obtained according to the electricity consumption data and the electricity consumption property of the statistics of the month of January,
from the root node of the tree to each leaf node, a classification rule is formed, and the partial classification rule represented by the above graph is shown in table 1:
TABLE 1 rule List for decision Tree model
| Numbering device | Rule condition | Class of electricity | Reliability of rule (%) |
| 1 | 24-point electric quantity duty ratio<= 0.029565529941723567 and No. 27 electric quantity duty ratio> 0.028783556206058748 and No. 30 electric quantity duty ratio> 0.013059723327690499 | Commercial business | 96.42857 |
| 2 | 7-point electric quantity duty ratio> 0.014347999362346564 and 18-point electric quantity duty ratio<= 0.05558296635932825 and No. 30 electric quantity duty ratio<= 0.013059723327690499 | Industrial process | 91.04478 |
| 3 | 24-point electric quantity duty ratio> 0.029565529941723567 and number 1 electric quantity duty ratio> 0.03092121905804584 and No. 30 electric quantity duty ratio> 0.013059723327690499 | Resident | 87.7193 |
| 4 | 24-point electric quantity duty ratio<= 0.029565529941723567 and No. 27 electric quantity duty ratio<= 0.028783556206058748 and No. 30 electric quantity duty ratio> 0.013059723327690499 | General industry | 83.33333 |
| 5 | 7-point electric quantity duty ratio<= 0.014347999362346564 and 18-point electric quantity duty ratio<= 0.05558296635932825 and No. 30 electric quantity duty ratio<= 0.013059723327690499 | General industry | 75 |
Step 203: model result analysis, namely testing the power consumption data of a certain city in 1 month on a decision tree classification model to detect the effectiveness of the model on a training data set, and selecting an overall evaluation index: correct classification rate, kappa statistics; the target evaluation indexes are as follows: accuracy, coverage, F measure. The overall accuracy of the model reaches 86.3%, and the accuracy of each target is higher.
Example 2
The embodiment of the invention provides an engineering audit problem monitoring system based on a big data AI technology, which is shown in a figure 5 and comprises the following steps:
the demand determination module: the audit information acquisition module is used for acquiring audit information of the user;
a target data determining module: the audit target data are obtained by inputting the audit requirements into a pre-constructed audit association model;
audit monitoring module: the audit target data are input into a pre-constructed audit autonomous supervision model to obtain an audit monitoring result;
the construction of the audit association model in the target data determination module comprises the following steps:
feature extraction is carried out according to the audit demand, and features to be extracted are determined;
feature selection is carried out according to the audit demand, and the category to be extracted is determined;
based on a preset audit database, carrying out data extraction according to the characteristics to be extracted and the categories to be extracted, and carrying out data cleaning and data duplication elimination on the extracted data to obtain associated data;
and carrying out association degree calculation on the association data and the audit demands, screening the association data with association degree calculation results larger than a preset threshold value as audit target data, and constructing an audit association model.
The construction of the audit autonomous supervision model in the audit monitoring module comprises the following steps:
performing multi-mode self-supervision learning aiming at the audit target data, and marking problem data;
and carrying out text classification on the problem data according to a preset problem attribute value to obtain an audit monitoring result.
The audit monitoring module carries out multi-mode self-supervision learning aiming at the audit target data and marks problem data, and the method comprises the following steps:
encoding the mask data of the audit target data to obtain primary encoded data;
based on the audit demand, carrying out weight setting on the audit target data to obtain weight setting information;
encoding the unmasked data in the audit target data to obtain secondary encoded data;
and based on the primary coding data, the secondary coding data and the weight setting information, marking the problem data in the audit target data.
The demand determining module is specifically configured to:
collecting audit information, and carrying out structural conversion on the audit information to obtain primary conversion information;
carrying out multi-scale feature fusion on the primary conversion information to obtain fusion information;
extracting key information according to the fusion information, and determining that the audit information corresponds to an audit requirement;
wherein the audit information includes: audit objectives, audit scope, and audit content.
The feature to be extracted in the target data determining module comprises the following steps: text features and symbolic features; the category to be extracted comprises: text part of speech and text attributes.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, magnetic disk storage, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (6)
1. The engineering audit problem monitoring method based on the big data AI technology is characterized by comprising the following steps:
determining audit requirements according to the collected audit information;
inputting the audit requirements into a pre-constructed audit association model to obtain audit target data;
inputting the audit target data into a pre-constructed audit autonomous supervision model to obtain an audit monitoring result;
the audit association model construction process comprises the following steps:
feature extraction is carried out according to the audit demand, and features to be extracted are determined;
feature selection is carried out according to the audit demand, and the category to be extracted is determined;
based on a preset audit database, carrying out data extraction according to the characteristics to be extracted and the categories to be extracted, and carrying out data cleaning and data duplication elimination on the extracted data to obtain associated data;
performing association degree calculation on the association data and the audit demands, screening the association data with association degree calculation results larger than a preset threshold value as audit target data, and constructing an audit association model;
the audit autonomous supervision model construction comprises the following steps:
performing multi-mode self-supervision learning aiming at the audit target data, and marking problem data;
text classification is carried out on the problem data according to a preset problem attribute value, and an audit monitoring result is obtained;
the multi-mode self-supervision learning is performed on the audit target data, and the problem data is marked, including:
encoding the mask data of the audit target data to obtain primary encoded data;
based on the audit demand, carrying out weight setting on the audit target data to obtain weight setting information;
encoding the unmasked data in the audit target data to obtain secondary encoded data;
and determining problem data based on the primary coding data, the secondary coding data and the weight setting information, and marking the problem data in the audit target data.
2. The method of claim 1, wherein determining audit requirements based on the collected audit information comprises:
collecting audit information, and carrying out structural conversion on the audit information to obtain primary conversion information;
carrying out multi-scale feature fusion on the primary conversion information to obtain fusion information;
extracting key information according to the fusion information, and determining that the audit information corresponds to an audit requirement;
wherein the audit information includes: audit objectives, audit scope, and audit content.
3. The method of claim 1, wherein the feature to be extracted comprises: text features and symbolic features; the category to be extracted comprises: text part of speech and text attributes.
4. An engineering audit problem monitoring system based on big data AI technology, which is characterized by comprising:
the demand determination module: the audit information acquisition module is used for acquiring audit information of the user;
a target data determining module: the audit target data are obtained by inputting the audit requirements into a pre-constructed audit association model;
audit monitoring module: the audit target data are input into a pre-constructed audit autonomous supervision model to obtain an audit monitoring result;
the construction of the audit association model in the target data determination module comprises the following steps:
feature extraction is carried out according to the audit demand, and features to be extracted are determined;
feature selection is carried out according to the audit demand, and the category to be extracted is determined;
based on a preset audit database, carrying out data extraction according to the characteristics to be extracted and the categories to be extracted, and carrying out data cleaning and data duplication elimination on the extracted data to obtain associated data;
performing association degree calculation on the association data and the audit demands, screening the association data with association degree calculation results larger than a preset threshold value as audit target data, and constructing an audit association model;
the construction of the audit autonomous supervision model in the audit monitoring module comprises the following steps:
performing multi-mode self-supervision learning aiming at the audit target data, and marking problem data;
text classification is carried out on the problem data according to a preset problem attribute value, and an audit monitoring result is obtained;
the audit monitoring module carries out multi-mode self-supervision learning aiming at the audit target data and marks problem data, and the method comprises the following steps:
encoding the mask data of the audit target data to obtain primary encoded data;
based on the audit demand, carrying out weight setting on the audit target data to obtain weight setting information;
encoding the unmasked data in the audit target data to obtain secondary encoded data;
and determining problem data based on the primary coding data, the secondary coding data and the weight setting information, and marking the problem data in the audit target data.
5. The system of claim 4, wherein the demand determination module is specifically configured to:
collecting audit information, and carrying out structural conversion on the audit information to obtain primary conversion information;
carrying out multi-scale feature fusion on the primary conversion information to obtain fusion information;
extracting key information according to the fusion information, and determining that the audit information corresponds to an audit requirement;
wherein the audit information includes: audit objectives, audit scope, and audit content.
6. The system of claim 4, wherein the target data determining the feature to be extracted in the module comprises: text features and symbolic features; the category to be extracted comprises: text part of speech and text attributes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310045164.5A CN115794803B (en) | 2023-01-30 | 2023-01-30 | Engineering audit problem monitoring method and system based on big data AI technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310045164.5A CN115794803B (en) | 2023-01-30 | 2023-01-30 | Engineering audit problem monitoring method and system based on big data AI technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115794803A CN115794803A (en) | 2023-03-14 |
| CN115794803B true CN115794803B (en) | 2023-05-09 |
Family
ID=85429207
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310045164.5A Active CN115794803B (en) | 2023-01-30 | 2023-01-30 | Engineering audit problem monitoring method and system based on big data AI technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115794803B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116843488A (en) * | 2023-06-28 | 2023-10-03 | 广东铭太信息科技有限公司 | Audit model building method and system |
| CN117151479A (en) * | 2023-11-01 | 2023-12-01 | 国网浙江省电力有限公司金华供电公司 | Engineering audit risk tracing method, device and medium based on fault tree model |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114998670A (en) * | 2022-04-14 | 2022-09-02 | 哈尔滨工业大学重庆研究院 | Multi-mode information pre-training method and system |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2897683A1 (en) * | 2015-07-16 | 2017-01-16 | Caseware International Inc. | Method, software, and device for displaying a graph visualizing audit risk data |
| CN109284319A (en) * | 2018-10-25 | 2019-01-29 | 九江学院 | A kind of auditing system based on big data visualization technique |
| CN109598484A (en) * | 2018-12-04 | 2019-04-09 | 广东电网有限责任公司 | A kind of project under construction turns fixed assets number auditing method and device |
| CN115357719B (en) * | 2022-10-20 | 2023-01-03 | 国网天津市电力公司培训中心 | Power audit text classification method and device based on improved BERT model |
-
2023
- 2023-01-30 CN CN202310045164.5A patent/CN115794803B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114998670A (en) * | 2022-04-14 | 2022-09-02 | 哈尔滨工业大学重庆研究院 | Multi-mode information pre-training method and system |
Non-Patent Citations (1)
| Title |
|---|
| 大数据环境下基于文本挖掘的审计数据分析框架;张志恒;成雪娇;;会计之友(第16期);第11--122页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115794803A (en) | 2023-03-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021184630A1 (en) | Method for locating pollutant discharge object on basis of knowledge graph, and related device | |
| CN110223168B (en) | Label propagation anti-fraud detection method and system based on enterprise relationship map | |
| CN115794803B (en) | Engineering audit problem monitoring method and system based on big data AI technology | |
| CN106845717B (en) | Energy efficiency evaluation method based on multi-model fusion strategy | |
| CN113590698B (en) | Artificial intelligence technology-based data asset classification modeling and hierarchical protection method | |
| CN114048870A (en) | Power system abnormity monitoring method based on log characteristic intelligent mining | |
| CN114003791B (en) | Depth map matching-based automatic classification method and system for medical data elements | |
| CN113779272A (en) | Data processing method, device and equipment based on knowledge graph and storage medium | |
| CN111126865B (en) | Technology maturity judging method and system based on technology big data | |
| CN115641162A (en) | Prediction data analysis system and method based on construction project cost | |
| Graf et al. | A risk analysis of file formats for preservation planning | |
| CN117151870B (en) | Portrait behavior analysis method and system based on guest group | |
| CN112363996A (en) | Method, system, and medium for building a physical model of a power grid knowledge graph | |
| Masud et al. | Cluster survival model of concept drift in load profile data | |
| CN110597796A (en) | Big data real-time modeling method and system based on full life cycle | |
| CN112506930B (en) | Data insight system based on machine learning technology | |
| CN115619539A (en) | Pre-loan risk evaluation method and device | |
| CN111680572B (en) | Dynamic judgment method and system for power grid operation scene | |
| CN114820074A (en) | Target user group prediction model construction method based on machine learning | |
| KR101629178B1 (en) | Apparatus for technology life analysis using multiple patent indicators | |
| CN114282875A (en) | Flow approval certainty rule and semantic self-learning combined judgment method and device | |
| CN113379093A (en) | Energy consumption analysis and optimization method for oil gas gathering and transportation system | |
| CN108564422A (en) | A kind of system based on matrimony vine data analysis | |
| CN112836926B (en) | Enterprise operation condition evaluation method based on electric power big data | |
| CN117076454B (en) | Engineering quality acceptance form data structured storage method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |