CN117932074A - Audit knowledge mapping system based on digital audit platform - Google Patents
Audit knowledge mapping system based on digital audit platform Download PDFInfo
- Publication number
- CN117932074A CN117932074A CN202311681774.0A CN202311681774A CN117932074A CN 117932074 A CN117932074 A CN 117932074A CN 202311681774 A CN202311681774 A CN 202311681774A CN 117932074 A CN117932074 A CN 117932074A
- Authority
- CN
- China
- Prior art keywords
- knowledge graph
- audit
- intelligent
- data
- knowledge
- 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.)
- Pending
Links
- 238000012550 audit Methods 0.000 title claims abstract description 117
- 238000013507 mapping Methods 0.000 title abstract description 7
- 230000007246 mechanism Effects 0.000 claims abstract description 57
- 238000005516 engineering process Methods 0.000 claims abstract description 41
- 238000010276 construction Methods 0.000 claims abstract description 29
- 238000005457 optimization Methods 0.000 claims abstract description 24
- 238000010801 machine learning Methods 0.000 claims abstract description 15
- 238000000034 method Methods 0.000 claims description 34
- 238000000605 extraction Methods 0.000 claims description 28
- 238000012549 training Methods 0.000 claims description 27
- 238000004458 analytical method Methods 0.000 claims description 21
- 230000008451 emotion Effects 0.000 claims description 19
- 230000006870 function Effects 0.000 claims description 17
- 230000011218 segmentation Effects 0.000 claims description 15
- 238000012795 verification Methods 0.000 claims description 11
- 238000003860 storage Methods 0.000 claims description 10
- 230000008859 change Effects 0.000 claims description 9
- 238000013136 deep learning model Methods 0.000 claims description 9
- 238000013461 design Methods 0.000 claims description 9
- 238000002372 labelling Methods 0.000 claims description 9
- 238000013135 deep learning Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 7
- 230000001360 synchronised effect Effects 0.000 claims description 7
- 238000012544 monitoring process Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000002787 reinforcement Effects 0.000 claims description 6
- 238000004140 cleaning Methods 0.000 claims description 4
- 238000007418 data mining Methods 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 4
- 238000000638 solvent extraction Methods 0.000 claims description 4
- 238000007619 statistical method Methods 0.000 claims description 4
- 239000013598 vector Substances 0.000 claims description 4
- 239000012634 fragment Substances 0.000 claims description 3
- 210000001503 joint Anatomy 0.000 claims description 3
- 230000007935 neutral effect Effects 0.000 claims description 3
- 238000010845 search algorithm Methods 0.000 claims description 3
- 238000012942 design verification Methods 0.000 claims description 2
- 230000003044 adaptive effect Effects 0.000 abstract description 6
- 238000003058 natural language processing Methods 0.000 description 12
- 238000007726 management method Methods 0.000 description 10
- 238000012545 processing Methods 0.000 description 8
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 4
- 238000013480 data collection Methods 0.000 description 4
- 230000010354 integration Effects 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 4
- 238000002790 cross-validation Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000004806 packaging method and process Methods 0.000 description 3
- 125000004122 cyclic group Chemical group 0.000 description 2
- 238000013499 data model Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000012854 evaluation process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011068 loading method Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/253—Grammatical analysis; Style critique
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/103—Workflow collaboration or project management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- Databases & Information Systems (AREA)
- Strategic Management (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Resources & Organizations (AREA)
- Life Sciences & Earth Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Economics (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Animal Behavior & Ethology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
An audit knowledge mapping system based on a digital audit platform comprises four modules: knowledge graph construction and updating, intelligent recommendation and search engine, collaborative editing and sharing, intelligent learning and adaptive optimization. The knowledge graph construction module adopts NLP and graph database technology to analyze audit text and extract key information, so that semantic understanding is ensured, and an automatic updating mechanism keeps real-time synchronization. The intelligent recommendation and search engine module is used for realizing full-text retrieval and fuzzy search by analyzing task characteristics and intelligent recommendation related knowledge and search engine technology. The collaborative editing and sharing module ensures non-tamper property by using a blockchain technology, and participates in updating and sharing together to reflect actual operation and experience. The intelligent learning and adaptive optimization module utilizes a machine learning algorithm to adaptively adjust the knowledge graph structure and content according to feedback and application conditions. The system improves the knowledge acquisition efficiency of auditors and optimizes the auditing flow.
Description
Technical Field
The invention relates to a digital auditing platform, in particular to an auditing knowledge mapping system based on the digital auditing platform.
Background
The digital audit platform refers to a platform for comprehensively monitoring, analyzing and evaluating financial and business activities of enterprises, organizations or individuals by utilizing digital technologies and information systems, and comprises digital audit, information system audit, risk assessment and the like. The digital audit is a product of the combination of the traditional audit method and the digital technology, and improves the audit efficiency and accuracy by means of data mining, data analysis, model establishment and the like.
The digital audit platform mainly comprises links such as data acquisition, data processing, model analysis, report generation and the like. The data acquisition stage is to acquire data of each layer of the enterprise, including financial data, operation data, risk data and the like. The data processing stage utilizes data cleaning, converting, loading (ETL) and other technologies to convert the acquired data into an analyzable format. The model analysis stage adopts methods such as statistical analysis, machine learning and the like to deeply mine data, so as to find potential problems and risks. And finally, the report generation stage provides clear audit results for auditors in a visual and summarized mode.
At present, a digital audit platform has become an integral part of enterprise management and audit work. Along with the continuous development of technologies such as big data, artificial intelligence and the like, a digital audit platform has remarkably progressed in the aspects of data processing, model building, intelligent analysis and the like. The modern digital audit platform can realize real-time monitoring, intelligent analysis and automatic reporting, and greatly improves audit efficiency and accuracy.
However, digital audit platforms still present some challenges in knowledge management. The auditing work relates to a plurality of complex business rules, legal laws, enterprise internal processes and the like, and auditors have rich knowledge systems. The current digital audit platform has a relatively limited problem in the aspect of knowledge management, and lacks a systematic knowledge graph, so that auditors are difficult to quickly acquire relevant knowledge when processing complex audit tasks, and the efficiency and quality of audit work are affected.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides an audit knowledge mapping system based on a digital audit platform. The system organically integrates various kinds of knowledge required by auditors by constructing an audit knowledge graph to form a clear knowledge map. On the basis of a digital audit platform, the automatic management, retrieval and application of audit knowledge are realized through the establishment of a knowledge graph, the knowledge acquisition efficiency of auditors is improved, and the audit flow is optimized.
The audit knowledge mapping system based on the digital audit platform comprises a knowledge map construction and updating module, an intelligent recommendation and search engine module, a collaborative editing and sharing module and an intelligent learning and adaptability optimization module.
The knowledge graph construction and updating module analyzes the audit text to extract key information by utilizing an NLP and graph database technology, ensures semantic understanding by utilizing the NLP, introduces entity relation extraction to establish association relation, and maintains real-time performance by an automatic updating mechanism to synchronously audit the knowledge graph and service environment change in time; the intelligent recommending and searching engine module comprises an intelligent recommending system and a high-efficiency searching engine; by analyzing the auditing task characteristics and personnel working habits, the system intelligently recommends related knowledge graph information, and the search engine technology realizes full-text retrieval and fuzzy search functions and rapidly positions knowledge nodes; the collaborative editing and sharing module establishes a collaborative editing and sharing mechanism of a knowledge graph, adopts a blockchain technology to ensure non-tamper property, and auditors participate in updating and maintaining together to share experience and insight and reflect actual operation and experience; the intelligent learning and adaptability optimizing module performs intelligent learning on the audit knowledge graph by utilizing a machine learning algorithm, and the system adaptively adjusts the structure and the content of the knowledge graph by analyzing the working history, the preference and the task completion condition according to the feedback and the application condition of the auditor.
The NLP algorithm of the knowledge graph construction and update module comprises word segmentation, part-of-speech tagging and grammar analysis; the word segmentation stage uses jieba and StanfordCoreNLP to realize the accurate segmentation of Chinese and English texts; part-of-speech tagging is performed through an HMM or CRF model, and an accurate part-of-speech tag is obtained; in the grammar analysis stage, sentence structure is known by means of dependency syntax or phrase structure syntax analysis, a syntax tree is constructed, and word dependency relationship is identified; and carrying out emotion analysis by adopting a deep learning method, and training a model to realize automatic judgment of emotion polarity of the text, wherein the judgment comprises positive emotion, negative emotion or neutral emotion.
The graph database technology of the knowledge graph construction and updating module further defines the data structure of nodes and edges, wherein the data structure comprises the accurate expression audit knowledge elements and association relations, the node definition comprises entities such as audit tasks and legal rules, and the edge definition relates to the relation of connecting nodes and comprises weights and relation types; and establishing proper indexes and using query language of the graph database, performing access control design, setting authority control mechanism, and data partitioning and slicing strategy, so as to improve the expandability of the graph database.
The knowledge graph construction and updating module further adopts an entity relation extraction algorithm to ensure accurate extraction of entities and relations in complex texts, prepares a training data set, collects labeling samples from audit related texts, trains a model through a machine learning or deep learning method by manually labeling or semi-supervised learning to obtain entity and relation labeling, improves accuracy and robustness through optimization algorithm training, integrates the trained model into a knowledge graph construction flow, integrates the model with a graph database module, realizes accurate extraction and storage of nodes and relations, and is applied to an online environment through online deployment to respond and process large-scale audit text data in real time.
The knowledge graph construction and updating module further combines an event triggering mechanism with a timing task, automatically updates the knowledge graph based on business events and timing checking text data, detects newly added, modified or deleted data by using a text comparison algorithm, and only synchronizes the changed part through a synchronization strategy, thereby avoiding total synchronization performance overhead; the update log recording and rollback mechanism establishes an update log by recording each update operation including content and time stamp, and the rollback mechanism is designed to restore the knowledge graph to the previous state when the update is wrong, thereby ensuring the stability of the system.
The intelligent recommendation system of the intelligent recommendation and search engine module collects data from audit task description, legal text of regulations and historical work records of auditors to establish a training data set of a deep learning model, preprocesses the data, comprises word segmentation, part-of-speech tagging and feature extraction, acquires semantic information and key features, selects a deep learning model structure, and designs an input layer, a hidden layer and an output layer to realize intelligent recommendation of relevant information in a knowledge graph; selecting proper loss function and optimization algorithm, and continuously adjusting model parameters through back propagation to improve performance; the API interface is designed to be in butt joint with the digital audit platform, so that the model is ensured to accept task description and return recommended results, the model is deployed to a high-performance server, a monitoring mechanism is arranged to monitor performance in real time, and the model is continuously optimized through user feedback information.
The intelligent recommendation and search engine module further comprises the step of carrying out full-text indexing on nodes and edges in the knowledge graph by adopting inverted indexes, so that a user is supported to quickly acquire related information through keywords and phrases; integrating a fuzzy search algorithm, analyzing fuzzy queries such as spelling errors and synonyms, improving search fault tolerance, and carrying out semantic similarity matching by using a similarity algorithm to improve accuracy; and proper index fragments and copy numbers are configured to improve the searching performance and usability, and the query efficiency of the search engine is optimized by adjusting the parameters of the query algorithm.
The collaborative editing and sharing module selects a distributed database suitable for collaborative editing in a collaborative editing mechanism, and configures parameters to meet real-time synchronous and high concurrent editing requirements; the CRDT algorithm is applied to ensure real-time synchronization of multi-user editing operation, decoupling editing operation is atomic operation, a synchronization mechanism and a conflict resolution strategy are designed to ensure consistency, a permission mechanism is designed to ensure different editing permissions of different users, user identity verification and permission classification are implemented, and dynamic permission adjustment is realized to ensure system flexibility and security.
The collaborative editing and sharing module further comprises a blockchain platform suitable for the knowledge graph, intelligent contracts are applied, editing operation and authority rules are definitely defined by the intelligent contracts, operation legitimacy is guaranteed, each time of knowledge graph editing record is packaged into a block to form a tamper-proof chain record, an auditor is called by a design verification algorithm to verify legal editing of the knowledge graph through the intelligent contracts, and a credibility report is generated to help judge credibility of the knowledge graph.
The intelligent learning and adaptability optimizing module further comprises the steps of acquiring working history, task completion condition and user feedback data of auditors through a digital audit platform and related systems, cleaning and preprocessing the data and storing the data in a proper database, defining characteristics related to audit service and knowledge graph requirements, extracting the characteristics by adopting a data mining technology and a statistical analysis method to form characteristic vectors, constructing a training set by utilizing a supervised learning algorithm, including input characteristics and output labels, and ensuring model generalization performance by adopting a cross verification method; adopting a reinforcement learning algorithm, embedding an online learning module, receiving user feedback and application data in real time, and adjusting the structure and the content of a knowledge graph in real time; and (3) periodically adopting a performance index to evaluate the learning model, and adjusting parameters or update frequency of the online learning algorithm according to the evaluation result.
The beneficial effects of the invention are as follows:
By establishing the audit knowledge graph, the system organically integrates various kinds of knowledge required by auditors, and a clear knowledge map is formed. This helps the audit personnel to know and apply audit knowledge more fully, improves work efficiency.
The knowledge graph construction and updating module utilizes NLP and graph database technology to realize semantic understanding and establishment of association relation of audit texts. The system keeps real-time synchronization of the knowledge graph and the business environment through an automatic updating mechanism, and ensures that the management of audit knowledge is always kept in an up-to-date state.
The intelligent recommending and searching engine module intelligently recommends relevant knowledge graph information by analyzing auditing task characteristics and personnel working habits. The search engine technology realizes full-text retrieval and fuzzy search functions, helps auditors to quickly locate required knowledge nodes, and improves information retrieval efficiency.
The collaborative editing and sharing module introduces a blockchain technology to ensure the non-tamper property of the knowledge graph. Audit staff can participate in updating and maintaining the knowledge graph together, and share experience and insight, so that actual operation and experience are reflected, and the knowledge graph is more practical and adaptive.
The intelligent learning and adaptability optimization module utilizes a machine learning algorithm to adaptively adjust the structure and the content of the knowledge graph according to feedback and application conditions of auditors. The mechanism is helpful for continuously improving the quality and adaptability of the knowledge graph and meeting the change of the actual demands of auditors.
According to the application, through functions of integration, management, recommendation, search, collaborative editing, intelligent learning and the like, more intelligent and efficient knowledge graph support is provided for a digital audit platform, so that the efficiency and quality of audit work are improved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate and together with the description serve to explain the application, if necessary:
Fig. 1 is a system configuration diagram of the present invention.
Detailed Description
The present invention will now be described in detail with reference to the drawings and the specific embodiments thereof, wherein the exemplary embodiments and the description are for the purpose of illustrating the invention only and are not to be construed as limiting the invention.
The audit knowledge mapping system based on the digital audit platform comprises a knowledge map construction and updating module, an intelligent recommendation and search engine module, a collaborative editing and sharing module and an intelligent learning and adaptability optimization module.
(1) Knowledge graph construction and updating module
The knowledge graph construction and updating module technology combines advanced Natural Language Processing (NLP) and graph database technology to construct an intelligent audit knowledge graph. Through NLP technology, the system can more accurately understand semantic information in complex texts, so that the knowledge graph is more intelligent. The introduction of entity relation extraction technology further enhances the relevance among knowledge elements and provides more comprehensive knowledge support for auditors.
Technical implementation of the knowledge graph construction and updating module comprises NLP text processing, graph database construction and management, entity relation extraction technology and an automatic updating mechanism. Firstly, advanced NLP algorithm is adopted to segment words, part of speech tagging and grammar analysis on the audit related text so as to extract entities and keywords in the text. Semantic emotion in a text is evaluated through emotion analysis and other technologies, and meaning behind the text is better understood. Next, a graph database, such as Neo4j or ArangoDB, suitable for knowledge graph is selected for storing and managing audit knowledge graphs. And defining the data structure of the nodes and the edges in the graph database, and ensuring that each element of the audit knowledge and the association relation of each element can be accurately expressed.
Further, the entity relation extraction algorithm is used for extracting the nodes and the association relation of the audit knowledge graph from the text. And through a machine learning or deep learning method, the entity relation extraction model is continuously optimized, and the accuracy and the robustness are improved. And finally, designing a periodic or real-time auditing knowledge graph updating mechanism to ensure that the knowledge graph is synchronous with the change of the service environment. And automatically detecting the update of the text data by using an event triggering mechanism or a timing task, and triggering the corresponding update of the knowledge graph.
The technology can deeply understand text information, intelligent association relation, automatic updating mechanism, feasibility and reliability and improve knowledge retrieval efficiency. Deep mining of relevant text of the examination is achieved through an NLP technology, and knowledge patterns are more intelligent and accurate through an intelligent association relation extraction technology. The automatic updating mechanism can track the service environment change in real time and keep the knowledge graph synchronous with the actual service. Finally, knowledge related to the auditing task can be retrieved and acquired more rapidly through intelligent recommendation and search engine technology, and auditing efficiency is improved. Compared with the traditional method, the technical scheme processes the text information related to the audit more comprehensively and intelligently, builds a more accurate and deep knowledge graph, and provides more powerful knowledge management support for the digital audit platform. Specific embodiments are as follows:
In order to realize the advanced processing of the relevant text of the examination, an advanced NLP algorithm is adopted, including word segmentation, part-of-speech tagging and grammar analysis. Firstly, through classical word segmentation algorithms, such as jieba word segmentation and Stanford CoreNLP word segmentation, chinese or English text is accurately segmented, and word-level text representation is obtained. For Chinese, jieba words are segmented by adopting a prefix dictionary and an HMM model, and StanfordCoreNLP provides advanced English word segmentation and part-of-speech tagging functions.
And then, part-of-speech tagging is carried out on the word segmentation result to obtain the grammar class of each word in the sentence. In the step, models such as HMM or CRF are adopted, and accurate classification of various words is realized through training the models, so that more accurate part-of-speech labels are obtained.
Then, through grammar analysis, the sentence structure is deeply known, grammar components such as main guests and the like are identified, and a syntax tree is constructed. In this process, text is deeply parsed to identify dependencies between words by means of dependency syntax analysis or phrase structure syntax analysis, e.g., using StanfordParser.
Finally, emotion analysis is performed to evaluate the recognition emotion tendencies of the text, and the semantic meaning of the text is better understood. In the emotion analysis stage, an emotion analysis model is constructed by adopting a deep learning method, such as a Convolutional Neural Network (CNN) or a Recurrent Neural Network (RNN). Through model training, the automatic judgment of emotion polarity in the text is realized, including positive, negative or neutral emotion.
Through the technical implementation of NLP text processing, the auditing related text can accurately divide words, label parts of speech, analyze grammar and analyze emotion. The method provides basic data for subsequent entity relation extraction and knowledge graph construction, provides clear operation guidance for technicians, and ensures that the method can be successfully implemented in practical application.
In order to achieve efficient storage and management of knowledge maps, a graph database, such as Neo4j or ArangoDB, suitable for knowledge maps is selected, and the data structures of nodes and edges in the graph database are defined. Firstly, in the stage of selecting a graph database, according to project requirements and performance requirements, neo4j is selected as a graph-based database, and excellent graph processing performance is achieved; or select ArangoDB a multimodal database supporting comprehensive application of the graph database, document database, and key value database. The selection ensures efficient storage and retrieval of complex associations according to specific circumstances.
Further, in terms of data structure definition of nodes and edges, each element and association relation of each element can be ensured to be accurately expressed. The definition of a node involves each entity, such as audit tasks, legal laws, internal processes of an enterprise, etc., designing a corresponding node type, and defining node attributes including, but not limited to, names, descriptions, keywords, etc. The definition of an edge then involves the relationships connecting the nodes, defining attributes of the edge, such as weights, relationship types, etc., according to the particular relationship type.
In the aspects of graph database indexing and query optimization, the improvement of the query efficiency of the graph database is key. And adopting an index strategy to establish proper indexes for the attributes of the nodes and the edges so as to improve the query performance. By using the query language of the graph database, the optimized query statement is written, and the advantages of the graph database are fully utilized to perform efficient graph query.
In addition, in order to ensure that the graph database has good safety and expandability in the use process, safety and expandability design is carried out. In the aspect of access control, a proper authority control mechanism is set, the safety of data is ensured, and the access authority of a user is limited. In terms of data partitioning and slicing, proper data partitioning and slicing strategies are designed aiming at a large-scale knowledge graph so as to improve the expandability of the graph database.
Through the technical implementation of the construction and management of the graph database, the efficient storage and retrieval of the knowledge graph are ensured, and reliable basic support is provided for the follow-up entity relation extraction and intelligent recommendation. The detailed implementation of the technical means provides clear operation guidance for technicians, and ensures that the technical means can be successfully implemented in practical applications.
In order to extract the nodes and the association relations of the audit knowledge graph from the text, an entity relation extraction algorithm is adopted, and an entity relation extraction model is continuously optimized through a machine learning or deep learning method. In an embodiment, first, an entity relationship extraction algorithm suitable for auditing knowledge-graphs is selected, including rule matching methods and machine learning based methods, such as Support Vector Machines (SVMs) or Conditional Random Fields (CRFs). This ensures accurate extraction of entities and relationships in complex text.
Secondly, preparing a training data set for the training entity relation extraction model, and collecting sample data with entity and relation labels from audit related texts. And labeling the entities and the relations in the training data set by a manual labeling or semi-supervised learning method, so as to ensure enough training data.
Model training and optimization is then performed by machine learning or deep learning methods. And selecting a proper entity relation extraction model, such as a Convolutional Neural Network (CNN) or a cyclic neural network (RNN), and training model parameters by using a labeling data set through optimization algorithms such as gradient descent and the like, so that the accuracy and the robustness of the model parameters are improved. And evaluating and optimizing the model by a cross-validation method and the like, and enhancing the generalization capability of the model on unlabeled data.
And finally, integrating the trained entity relation extraction model into the whole knowledge graph construction flow, and carrying out practical application. The integration of the model and the graph database construction module is ensured, and the accurate extraction and storage of the nodes and the relations of the knowledge graph are ensured. By online deployment, the model is applied to an online environment, and large-scale audit text data is responded and processed in real time.
Through the detailed implementation of the entity relation extraction technology, the capability of efficiently extracting the entity and the association relation from the audit related text is ensured, and the accuracy and the robustness of the entity and the association relation are improved through continuous optimization of the model. The detailed implementation of the technical means provides clear operation guidance for technicians, and ensures that the technical means can be successfully implemented in practical application.
In order to ensure the synchronous update of the knowledge graph and the service environment, a periodic or real-time audit knowledge graph update mechanism is designed. First, in the design phase of the update trigger mechanism, a mode of combining an event trigger mechanism and a timing task is adopted. Designing an event triggering mechanism based on business events, such as completion of audit tasks or legal change of regulations; meanwhile, a timing task is set, and the updating condition of the text data is periodically checked to automatically trigger the updating of the knowledge graph.
Secondly, in terms of data change detection and synchronization, a text comparison algorithm is utilized to detect the new addition, modification or deletion of text data, only the changed part is ensured to be synchronized through a synchronization strategy, and the performance overhead caused by full synchronization is avoided.
Further, an incremental update and full update strategy is formulated. Incremental updating is adopted, and only the changed part is updated, so that the updating efficiency is improved; and carrying out full-scale updating when necessary to ensure the integrity and consistency of the knowledge graph.
Finally, in terms of the update log recording and rollback mechanism, an update log is established by recording the update operation of each knowledge graph, including the changed content, the time stamp and other information. A rollback mechanism is designed, and when an update error occurs, the knowledge graph can be restored to the previous state through rollback, so that the stability of the system is ensured.
Through the detailed implementation scheme of the automatic updating mechanism, the knowledge graph can timely and accurately reflect the change of the service environment, and the instantaneity and the reliability of the system are improved.
(2) Intelligent recommendation and search engine module
And the intelligent recommending and searching engine module is used for realizing the technical scheme of intelligent recommending and efficient searching by analyzing the auditing task characteristics and the working habits of auditors. By combining the auditing task characteristics and the working habits of auditors, the dual functions of intelligent recommendation and efficient searching are realized. The intelligent recommendation is used for understanding the requirements of auditors through a deep learning model, so that intelligent pushing of relevant information in the knowledge graph is realized; the high-efficiency search engine combines full text search and fuzzy search technology to provide quick and accurate information search function.
In a technical embodiment, first is an intelligent recommendation system. And training a recommendation model by utilizing the deep learning model and analyzing information such as text description, related regulations, history records and the like of the auditing task. The model can understand the working demands of auditors and realize intelligent recommendation of relevant information in the knowledge graph. The implementation steps comprise data collection, data preprocessing, model construction and model training, and finally the data collection, the data preprocessing, the model construction and the model training are integrated into a digital audit platform, so that a real-time intelligent recommendation function is realized.
Second, it is an efficient search engine. And combining full-text retrieval and fuzzy search technology to construct a search engine index, so as to realize rapid retrieval of the knowledge graph. The implementation steps include establishing a full text index of the knowledge graph, selecting an applicable search engine technology, realizing a full text retrieval function, and improving the performance of a search engine by optimizing an index structure and a query algorithm.
The intelligent recommendation system improves the working efficiency of auditors through the characteristics of personalized recommendation and instantaneity. The efficient search engine meets the requirement of auditors for quickly acquiring knowledge information through quick and accurate information retrieval and fault tolerance design.
The intelligent recommendation and search engine module fully considers the specificity of the auditing task, realizes intelligent recommendation through a deep learning model, and improves the efficiency and fault tolerance of the search engine through full text retrieval and fuzzy search technology. Compared with the traditional search engine and recommendation system, the scheme can better understand the context of the audit task, so that recommendation is more personalized, and meanwhile, a strategy of combining full text search and fuzzy search is adopted in the aspect of the search engine, so that the requirement of audit personnel for quickly acquiring knowledge information is better met.
The specific implementation of the intelligent recommendation and search engine module is as follows:
In order to realize the intelligent recommendation system, necessary data are collected from audit task descriptions, legal text of regulations and audit personnel historical work records in the data collection and preparation stage. Including audit task description data, legal rules, and historical work records, to build training data sets for deep learning models.
When the data preprocessing is carried out, the collected data is processed, including word segmentation and part-of-speech tagging, so as to obtain richer semantic information. And simultaneously extracting the characteristics, and extracting key characteristics in task description, such as keywords, entity recognition results and the like, for constructing an input characteristic vector. The deep learning model construction stage selects a proper model structure, such as a cyclic neural network (RNN) or a Transformer, and designs an input layer, a hidden layer and an output layer to understand the work requirements of auditors and realize intelligent recommendation of relevant information in a knowledge graph.
In the model training and optimizing stage, proper loss functions, such as cross entropy loss, are selected for measuring the difference between the model output and the actual label. And the model parameters are continuously adjusted through a back propagation algorithm by using optimization algorithms such as gradient descent and the like, so that the model performance is improved. And adjusting the super parameters of the model by a cross-validation method and the like to find the optimal model configuration.
And finally, in the stage of integrating to the audit platform, designing an API interface which is in butt joint with the digital audit platform, so as to ensure that the model can accept audit task description and return a recommended result. The model is deployed to a high performance server to ensure real-time response to user requests. And setting a monitoring mechanism, monitoring the performance of the model in real time, and collecting user feedback information so as to continuously optimize the model.
Through the implementation scheme, smooth development and integration of the intelligent recommendation system are ensured, a powerful knowledge recommendation function is provided for the digital audit platform, and the working efficiency of auditors is improved.
In order to realize an efficient search engine, in the full-text indexing stage of establishing a knowledge graph, full-text indexing is carried out on nodes (such as entities and attributes) and edges (relations) in the knowledge graph, so that all aspects of the knowledge graph are ensured to be covered. At the same time, a proper index structure, such as inverted index, is adopted to improve the searching efficiency and reduce the storage space of the index.
In selecting search engine technology suitable for knowledge-graph, such as elastiscearch integration technology. Version adaptation is key, and proper search engine versions are selected according to the characteristics of the knowledge graph, so that the knowledge graph can support graph data better.
In the stage of realizing the full text retrieval function, keyword retrieval and phrase retrieval are important functions. By realizing the function of supporting keyword retrieval, the user is ensured to be able to quickly acquire related information by inputting keywords. Meanwhile, phrase retrieval functions are integrated to support the search requirements of a user for specific phrases.
When the fuzzy search algorithm is integrated, fuzzy queries input by a user, such as misspellings, synonyms and the like, are analyzed to improve search fault tolerance. And carrying out semantic similarity matching on the query keywords by using a similarity algorithm so as to improve the accuracy of searching.
And finally, in the stage of optimizing the index structure and the query algorithm, configuring proper index fragments and the number of copies, and improving the searching performance and usability. The query efficiency of the search engine is optimized by adjusting query algorithm parameters, such as Boolean query, weight distribution and the like.
Through the implementation scheme, the high-efficiency search engine can quickly and accurately search information of the knowledge graph in the digital audit platform, so that strong support is provided for auditors, and information search efficiency is improved.
(3) Collaborative editing and sharing module
And establishing a collaborative editing and sharing mechanism of the audit knowledge graph. Through the mechanism, auditors can participate in updating and maintaining the knowledge graph together, and share experience and insight of each other. The block chain technology is adopted to ensure the non-tamper property of the knowledge graph and increase the credibility of the information. The collaborative editing mechanism can reflect the actual operation and experience of different auditors, so that the knowledge graph is more practical and adaptive.
The collaborative editing and sharing module is used for establishing a collaborative editing and sharing mechanism of the audit knowledge graph, and combining the collaborative editing and sharing mechanism and the blockchain technology to realize the common maintenance of the knowledge graph and the trusted sharing of information. The collaborative editing mechanism enables different auditors to participate in updating of the knowledge graph in real time, and the blockchain technology guarantees the non-tamper property of the knowledge graph and the credibility of information.
In the implementation scheme of the collaborative editing mechanism, a distributed database and a real-time synchronization algorithm are adopted to realize collaborative editing of the knowledge graph by multiple users. A distributed database suitable for collaborative editing is selected, such as Firebase RealtimeDatabase, and the CRDT algorithm is used to handle conflicts that may arise from multi-user simultaneous editing. The authority control is necessary, and an authority control mechanism is designed to ensure that different users have different editing authorities so as to prevent illegal tampering.
In an embodiment of a blockchain technology application, an appropriate blockchain platform is selected, such as Ethereum or HyperledgerFabric, to determine the public or private chain as desired. By writing intelligent contracts, editing operation and authority rules of the knowledge graph are defined, and the legality of the editing operation is ensured. And packaging each knowledge graph editing record into a block, and storing the block on a block chain to form a tamper-proof chain record. And a verification mechanism is provided, and an auditor can verify whether a specific part of the knowledge graph is legally edited or not, so that the credibility of the information is increased.
Compared with the traditional knowledge graph editing mechanism, the scheme introduces collaborative editing and blockchain technology, so that the editing of the knowledge graph is more real-time, safe and reliable. The traditional method is difficult to solve the conflict problem possibly caused by simultaneous editing of multiple users, and the collaborative editing mechanism effectively solves the problem through CRDT algorithm. The introduction of the blockchain technology ensures the non-tamper property of the editing record and provides a verifiable editing history for auditors.
The specific embodiment is as follows:
In the collaborative editing mechanism, attention is paid to selection of a distributed database, and the distributed database suitable for collaborative editing is selected so as to meet the requirements of real-time synchronization and high concurrency editing operation. Taking FirebaseRealtime Database as an example, ensuring that the system can be stably connected to a distributed database, configuring database parameters such as data storage space and read-write permission to adapt to the requirement of collaborative editing. Meanwhile, a data model suitable for the knowledge graph is designed, and the data model comprises a structure of nodes and edges, so that the graph data can be effectively stored and synchronized.
And secondly, a real-time synchronization algorithm is applied, and Conflict-FREEREPLICATEDDATATYPE (CRDT) algorithm is adopted, so that real-time synchronization of multi-user editing operation is ensured, and conflicts and inconsistent data are avoided. And CRDT algorithm suitable for knowledge graph editing is selected, the editing operation is decoupled into atomic operation, a synchronization mechanism is designed to ensure that the editing operation of each user can be perceived by other users in real time, and meanwhile, a conflict resolution strategy is formulated, such as a final writing winning strategy is adopted, so that editing consistency is ensured.
Finally, paying attention to authority control, designing an authority control mechanism, and ensuring different users to have different editing authorities so as to prevent illegal tampering. The user identity verification mechanism is implemented, the identity of the user is verified through the identity verification system, the authority classification is formulated, the editing authorities are divided according to the roles of the user, and dynamic authority adjustment is realized, so that an administrator can flexibly adjust the user authorities as required, and the flexibility and the safety of the system are ensured.
Through the implementation scheme, the collaborative editing mechanism can realize real-time synchronization in a distributed environment, and the efficiency and consistency of collaborative editing are ensured. The CRDT algorithm effectively solves the problem of conflict possibly generated by simultaneous editing of multiple users, and the authority control mechanism guarantees the legality and safety of editing operation.
In the application of blockchain technology, attention is paid to the selection of a blockchain platform, and the blockchain platform applicable to a knowledge graph, such as Ethereum or HyperledgerFabric, is selected to ensure that the platform can meet requirements. And carrying out platform assessment, selecting a platform meeting the requirements of the knowledge graph, and determining whether to use a public chain or a private chain or adopting a combination of mixed chains according to the security and privacy requirements. Building a corresponding blockchain environment, and ensuring that the storage and verification requirements of the knowledge graph are met.
Secondly, applying the intelligent contract, writing the intelligent contract, and definitely defining the editing operation and authority rule of the knowledge graph to ensure the legality of the editing operation. The smart contract is an automated contract that executes on the blockchain that ensures that operations are performed without a centralized authority. The edit operation is explicitly defined, the design smart contract specifies the rights and access controls for the edit operation, and the applicable smart contract language is selected, such as Solidity or Chaincode.
Then, paying attention to the edit record storage, packaging each knowledge graph edit record into a block, and storing the block on a block chain to form a tamper-proof chain record. Defining a data structure for storing editing records in a block, packaging each editing record into a transaction, writing the transaction into a block chain, and optimizing the storage structure to ensure high efficiency.
Finally, information credibility verification is provided, a verification algorithm is designed, and an auditor can verify whether a specific part of the knowledge graph is legally edited or not through the invocation of the intelligent contract. And providing a credibility report generation mechanism, and presenting the verification result to an auditor in an understandable manner to help the auditor judge the credibility of the knowledge graph. Traceability and authenticity of information is achieved through non-tamper-ability of intelligent contracts and blockchains.
Through the implementation scheme, the collaborative editing mechanism can realize collaborative editing of the knowledge graph by multiple users in a distributed environment, and simultaneously guarantee real-time synchronization and authority control. The CRDT algorithm effectively solves the conflict possibly caused by concurrent editing, and the authority control mechanism guarantees the safety and the legality of the knowledge graph.
(4) Intelligent learning and adaptability optimization module
The intelligent learning and adaptability optimization module utilizes a machine learning algorithm to enable the intelligent learning module to flexibly adapt to the needs of auditors and the changes of service environments through intelligent learning of the audit knowledge graph. According to feedback and actual application conditions of auditors, the self-adaptive adjustment of the knowledge graph is realized, the quality and the adaptability of the knowledge graph are continuously improved, and the auditing requirements are met more flexibly.
In the aspect of the technical implementation scheme, the intelligent learning and adaptability optimization module adopts machine learning algorithms such as supervised learning, reinforcement learning and the like. Firstly, a data set of a learning model is constructed by collecting data such as working history, task completion condition and feedback of auditors. From these data, the auditor's working characteristics and preferences are then extracted as input to the machine learning model. The preliminary training of the model adopts a supervised learning algorithm to simulate a sample marked manually. In practical application, online learning is performed through a reinforcement learning algorithm, and the structure and the content of the knowledge graph are continuously adjusted according to practical feedback. The performance of the learning model is periodically evaluated and its effect in adaptive optimization is examined.
The scheme can dynamically adapt to the working requirements of auditors and the change of service environments. More importantly, the method can provide personalized knowledge support according to the work history and preference of individual auditors, realize continuous learning and optimization of the knowledge graph through a machine learning algorithm, and improve the quality and practicability of the knowledge graph.
Specific embodiments are as follows:
First, the data collection phase is to build a base stone for intelligent learning and adaptive optimization. And acquiring data such as working history, task completion condition and user feedback of auditors through a digital audit platform and related systems. These data are subjected to careful cleaning and preprocessing to handle missing values and outliers to ensure the quality and integrity of the data. The processed data is then stored in an appropriate database, providing the basis for subsequent feature extraction and model training.
In the feature extraction stage, the technical details comprise defining a series of features related to audit business and knowledge graph requirements, such as working time periods, task types, task difficulties, audit objects and the like. These features are extracted from the collected data by data mining techniques and statistical analysis methods to form feature vectors. Feature engineering processes such as normalization and normalization are further performed to ensure feature comparability and consistency.
Following the model training phase, a suitable supervised learning algorithm, such as a Support Vector Machine (SVM) or Deep Neural Network (DNN), is selected. The training set is constructed to include input features and corresponding output labels, such as user feedback or task completion. And adopting methods such as cross validation and the like to ensure the generalization performance of the model.
The online learning stage introduces a reinforcement learning algorithm, and selects an algorithm suitable for online learning, such as a Q-learning or deep reinforcement learning algorithm. An online learning module is embedded in the digital audit platform, user feedback and actual application data are received in real time, and the structure and the content of the knowledge graph are adjusted in real time through an algorithm. And establishing a closed loop system for user feedback, and ensuring that the online learning effect can be timely reflected in the optimization of the knowledge graph.
Finally, the model evaluation stage periodically adopts proper performance indexes to evaluate the learning model, such as accuracy, recall rate, F1 value and the like. And adjusting parameters or update frequency of the online learning algorithm according to the evaluation result so as to improve the performance and adaptability of the model. Meanwhile, the evaluation process and the result are documented to form a technical document, so that the technical personnel can perform further optimization and maintenance conveniently. The series of technical embodiments provide intelligent learning and adaptability optimization capability for the digital auditing platform, so that the knowledge graph can better meet the requirements and business changes of auditors.
The foregoing description is only of the preferred embodiments of the invention, and all changes and modifications that come within the meaning and range of equivalency of the structures, features and principles of the invention are therefore intended to be embraced therein.
Claims (10)
1. An audit knowledge graph system based on a digital audit platform is characterized in that: the audit knowledge graph system comprises a knowledge graph construction and updating module, an intelligent recommendation and search engine module, a collaborative editing and sharing module and an intelligent learning and adaptability optimization module; the knowledge graph construction and updating module analyzes the audit text to extract key information by utilizing an NLP and graph database technology, ensures semantic understanding by utilizing the NLP, introduces entity relation extraction to establish association relation, and maintains real-time performance by an automatic updating mechanism to synchronously audit the knowledge graph and service environment change in time; the intelligent recommending and searching engine module comprises an intelligent recommending system and a high-efficiency searching engine; by analyzing the auditing task characteristics and personnel working habits, the system intelligently recommends related knowledge graph information, and the search engine technology realizes full-text retrieval and fuzzy search functions and rapidly positions knowledge nodes; the collaborative editing and sharing module establishes a collaborative editing and sharing mechanism of a knowledge graph, adopts a blockchain technology to ensure non-tamper property, and auditors participate in updating and maintaining together to share experience and insight and reflect actual operation and experience; the intelligent learning and adaptability optimizing module performs intelligent learning on the audit knowledge graph by utilizing a machine learning algorithm, and the system adaptively adjusts the structure and the content of the knowledge graph by analyzing the working history, the preference and the task completion condition according to the feedback and the application condition of the auditor.
2. An audit knowledge graph system based on a digital audit platform as claimed in claim 1, wherein: the NLP algorithm of the knowledge graph construction and update module comprises word segmentation, part-of-speech tagging and grammar analysis; the word segmentation stage uses jieba and Stanford CoreNLP to realize the accurate segmentation of Chinese and English texts; part-of-speech tagging is performed through an HMM or CRF model, and an accurate part-of-speech tag is obtained; in the grammar analysis stage, sentence structure is known by means of dependency syntax or phrase structure syntax analysis, a syntax tree is constructed, and word dependency relationship is identified; and carrying out emotion analysis by adopting a deep learning method, and training a model to realize automatic judgment of emotion polarity of the text, wherein the judgment comprises positive emotion, negative emotion or neutral emotion.
3. An audit knowledge graph system based on a digital audit platform as claimed in claim 1 or 2, wherein: the graph database technology of the knowledge graph construction and updating module further defines the data structure of nodes and edges, wherein the data structure comprises the accurate expression audit knowledge elements and association relations, the node definition comprises entities such as audit tasks and legal rules, and the edge definition relates to the relation of connecting nodes and comprises weights and relation types; and establishing proper indexes and using query language of the graph database, performing access control design, setting authority control mechanism, and data partitioning and slicing strategy, so as to improve the expandability of the graph database.
4. An audit knowledge graph system based on a digital audit platform as claimed in claim 1, wherein: the knowledge graph construction and updating module further adopts an entity relation extraction algorithm to ensure accurate extraction of entities and relations in complex texts, prepares a training data set, collects labeling samples from audit related texts, trains a model through a machine learning or deep learning method by manually labeling or semi-supervised learning to obtain entity and relation labeling, improves accuracy and robustness through optimization algorithm training, integrates the trained model into a knowledge graph construction flow, integrates the model with a graph database module, realizes accurate extraction and storage of nodes and relations, and is applied to an online environment through online deployment to respond and process large-scale audit text data in real time.
5. An audit knowledge graph system based on a digital audit platform as claimed in claim 1, wherein: the knowledge graph construction and updating module further combines an event triggering mechanism with a timing task, automatically updates the knowledge graph based on business events and timing checking text data, detects newly added, modified or deleted data by using a text comparison algorithm, and only synchronizes the changed part through a synchronization strategy, thereby avoiding total synchronization performance overhead; the update log recording and rollback mechanism establishes an update log by recording each update operation including content and time stamp, and the rollback mechanism is designed to restore the knowledge graph to the previous state when the update is wrong, thereby ensuring the stability of the system.
6. An audit knowledge graph system based on a digital audit platform as claimed in claim 1, wherein: the intelligent recommendation system of the intelligent recommendation and search engine module collects data from audit task description, legal text of regulations and historical work records of auditors to establish a training data set of a deep learning model, preprocesses the data, comprises word segmentation, part-of-speech tagging and feature extraction, acquires semantic information and key features, selects a deep learning model structure, and designs an input layer, a hidden layer and an output layer to realize intelligent recommendation of relevant information in a knowledge graph; selecting proper loss function and optimization algorithm, and continuously adjusting model parameters through back propagation to improve performance; the API interface is designed to be in butt joint with the digital audit platform, so that the model is ensured to accept task description and return recommended results, the model is deployed to a high-performance server, a monitoring mechanism is arranged to monitor performance in real time, and the model is continuously optimized through user feedback information.
7. An audit knowledge graph system based on a digital audit platform as claimed in claim 1 or 6, wherein: the intelligent recommendation and search engine module further comprises the step of carrying out full-text indexing on nodes and edges in the knowledge graph by adopting inverted indexes, so that a user is supported to quickly acquire related information through keywords and phrases; integrating a fuzzy search algorithm, analyzing fuzzy queries such as spelling errors and synonyms, improving search fault tolerance, and carrying out semantic similarity matching by using a similarity algorithm to improve accuracy; and proper index fragments and copy numbers are configured to improve the searching performance and usability, and the query efficiency of the search engine is optimized by adjusting the parameters of the query algorithm.
8. An audit knowledge graph system based on a digital audit platform as claimed in claim 1, wherein: the collaborative editing and sharing module selects a distributed database suitable for collaborative editing in a collaborative editing mechanism, and configures parameters to meet real-time synchronous and high concurrent editing requirements; the CRDT algorithm is applied to ensure real-time synchronization of multi-user editing operation, decoupling editing operation is atomic operation, a synchronization mechanism and a conflict resolution strategy are designed to ensure consistency, a permission mechanism is designed to ensure different editing permissions of different users, user identity verification and permission classification are implemented, and dynamic permission adjustment is realized to ensure system flexibility and security.
9. An audit knowledge graph system based on a digital audit platform as claimed in claim 1 or 8, wherein: the collaborative editing and sharing module further comprises a blockchain platform suitable for the knowledge graph, intelligent contracts are applied, editing operation and authority rules are definitely defined by the intelligent contracts, operation legitimacy is guaranteed, each time of knowledge graph editing record is packaged into a block to form a tamper-proof chain record, an auditor is called by a design verification algorithm to verify legal editing of the knowledge graph through the intelligent contracts, and a credibility report is generated to help judge credibility of the knowledge graph.
10. An audit knowledge graph system based on a digital audit platform as claimed in claim 1, wherein: the intelligent learning and adaptability optimizing module further comprises the steps of acquiring working history, task completion condition and user feedback data of auditors through a digital audit platform and related systems, cleaning and preprocessing the data and storing the data in a proper database, defining characteristics related to audit service and knowledge graph requirements, extracting the characteristics by adopting a data mining technology and a statistical analysis method to form characteristic vectors, constructing a training set by utilizing a supervised learning algorithm, including input characteristics and output labels, and ensuring model generalization performance by adopting a cross verification method; adopting a reinforcement learning algorithm, embedding an online learning module, receiving user feedback and application data in real time, and adjusting the structure and the content of a knowledge graph in real time; and (3) periodically adopting a performance index to evaluate the learning model, and adjusting parameters or update frequency of the online learning algorithm according to the evaluation result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311681774.0A CN117932074A (en) | 2023-12-08 | 2023-12-08 | Audit knowledge mapping system based on digital audit platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311681774.0A CN117932074A (en) | 2023-12-08 | 2023-12-08 | Audit knowledge mapping system based on digital audit platform |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117932074A true CN117932074A (en) | 2024-04-26 |
Family
ID=90749719
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311681774.0A Pending CN117932074A (en) | 2023-12-08 | 2023-12-08 | Audit knowledge mapping system based on digital audit platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117932074A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118296164A (en) * | 2024-06-06 | 2024-07-05 | 新立讯科技集团股份有限公司 | Automatic agricultural product information acquisition and updating method and system based on knowledge graph |
| CN118377479A (en) * | 2024-06-26 | 2024-07-23 | 宁波沃尔斯软件有限公司 | Low-code software application system with reusable model |
| CN118656379A (en) * | 2024-08-20 | 2024-09-17 | 广州中南网络技术有限公司 | Automatic round operation circulation increasing and decreasing method based on numbers in server |
-
2023
- 2023-12-08 CN CN202311681774.0A patent/CN117932074A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118296164A (en) * | 2024-06-06 | 2024-07-05 | 新立讯科技集团股份有限公司 | Automatic agricultural product information acquisition and updating method and system based on knowledge graph |
| CN118377479A (en) * | 2024-06-26 | 2024-07-23 | 宁波沃尔斯软件有限公司 | Low-code software application system with reusable model |
| CN118656379A (en) * | 2024-08-20 | 2024-09-17 | 广州中南网络技术有限公司 | Automatic round operation circulation increasing and decreasing method based on numbers in server |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021103492A1 (en) | Risk prediction method and system for business operations | |
| Choi et al. | Using machine learning to revisit the diversification–performance relationship | |
| Hu et al. | Matching large ontologies: A divide-and-conquer approach | |
| CN112612902A (en) | Knowledge graph construction method and device for power grid main device | |
| KR101114023B1 (en) | Content propagation for enhanced document retrieval | |
| CN117932074A (en) | Audit knowledge mapping system based on digital audit platform | |
| CN113377850A (en) | Big data technology platform of cognitive Internet of things | |
| EP3671526B1 (en) | Dependency graph based natural language processing | |
| US8954360B2 (en) | Semantic request normalizer | |
| Bifulco et al. | An intelligent system for focused crawling from Big Data sources | |
| CN113392227A (en) | Metadata knowledge map engine system facing rail transit field | |
| CN108089843A (en) | A kind of intelligentized banking establishments grade requirement management systems | |
| US20200356726A1 (en) | Dependency graph based natural language processing | |
| CN112036842B (en) | Intelligent matching device for scientific and technological service | |
| Dong | Exploration on web usage mining and its application | |
| Zhang et al. | SAKA: an intelligent platform for semi-automated knowledge graph construction and application | |
| US20220374401A1 (en) | Determining domain and matching algorithms for data systems | |
| Morrison et al. | Business process integration: Method and analysis | |
| Salam et al. | Distributed framework for political event coding in real-time | |
| CN115982429B (en) | Knowledge management method and system based on flow control | |
| CN117371940A (en) | Holographic intelligent control method and system for financial credit and debit management | |
| Abbott et al. | Population statistics without a Census or register | |
| Liu et al. | A preliminary approach of constructing a knowledge graph-based enterprise informationized audit platform | |
| Li et al. | Automatic classification algorithm for multisearch data association rules in wireless networks | |
| Seidler et al. | MOSAIC: Criminal network analysis for multi-modal surveillance and decision support |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |