Knowledge graph generation method for audit documents

By performing data processing and relational model construction on audit documents and generating knowledge graphs, the problem of inefficient processing of traditional audit documents is solved, and more efficient electronic document query and analysis is achieved.

CN113868421BActive Publication Date: 2024-12-06HANZHONG AUDIT BUREAU +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111180977.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-10-11
Publication Date
2024-12-06
Estimated Expiration
2041-10-11

AI Technical Summary

Technical Problem

The processing of traditional audit documents is inefficient and cannot be effectively electronicized, which affects the query and analysis of electronic documents.

Method used

By stating the paper audit documents, extracting keyword groups and classifying them, establishing a relationship model of people, identities, events, and objects, generating a knowledge graph, correlating relevant data, constructing a transitive relationship, and forming a knowledge graph of audit documents.

Benefits of technology

It has improved the electronic processing efficiency of audit documents and enhanced the query and analysis capabilities of documents.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113868421B_ABST
    Figure CN113868421B_ABST
Patent Text Reader

Abstract

The present invention discloses a knowledge graph generation method for audit documents, which performs data processing on paper audit documents to obtain the document content after identification; classifies according to the keyword group of the document content after identification, and obtains the classification data of people, identities, events, and items respectively; takes the classification data of people, identities, events, and items as the main column, and establishes the relationship model of people-identity pointing to identity value, judging people-identity pointing to identity-identity value, people-identity pointing to identity-identity value, pointing to event-event value, people-identity pointing to identity-identity value, pointing to event-event value pointing to item-item value in the corresponding people, identities, events, and items; associates the classification data of people, identities, events, and items according to the relationship model and transitive relationship, and obtains the knowledge graph of the audit document. The present invention generates a knowledge graph from the electronic document of the audit document for subsequent query and analysis, thereby improving work efficiency and analysis ability.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of audit document generation graphs, and specifically relates to a knowledge graph generation method for audit documents. Background Art

[0002] Audit documents refer to the large amount of documentary materials generated during the audit process. In the case where the audit documents cannot be preserved in electronic form or the electronic documents are preserved by manual page turning, photographing, or scanning, audit documents need to be able to be processed and used effectively and efficiently.

[0003] The actual audit process requires manual extraction of the contents of people, identities, events, and objects in the documents for manual judgment by auditors. The disadvantages of manual processing are large workload and low efficiency.

[0004] To sum up, the problems and defects of the existing technology are as follows: the traditional audit document processing cannot effectively and efficiently digitize the audit documents, which affects the query and analysis of electronic documents. Summary of the invention

[0005] In view of this, the main purpose of the present invention is to provide a knowledge graph generation method for audit documents.

[0006] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0007] An embodiment of the present invention provides a method for generating a knowledge graph for audit documents, the method comprising:

[0008] Digitalize paper audit documents to obtain the identified document contents;

[0009] Classify the recognized document content according to the keyword group to obtain person, identity, event, and item classification data respectively;

[0010] Taking the classified data of people, identities, events, and items as the main columns, establish the corresponding relationship models of people-identity pointing to identity value, judging people-identity pointing to identity-identity value, people-identity pointing to identity-identity value pointing to event-event value, people-identity pointing to identity-identity value, pointing to event-event value pointing to item-item value;

[0011] According to the relational model and the transitive relationships among people, identities, events and objects, the classification data of people, identities, events and objects are associated to obtain a knowledge graph of the audit document.

[0012] In the above scheme, the classification data of people, identities, events, and items are used as the main columns to establish corresponding relationship models among people, identities, events, and items, in which person-identity points to identity value, person-identity points to identity-identity value, person-identity points to identity-identity value, person-identity points to identity-identity value, and event-event value points to item-item value. Specifically, it at least includes the definition of person-identity gender, person-identity age, person-identity title, person-identity nationality, person-identity ID number, identity age-identity value age, identity gender-identity value gender, event noun-event value, event phrase-event value, item shape-item value, item noun-item value.

[0013] In the above scheme, the transitive relationship between the person, identity, event, and object is specifically as follows: create an entity pair for the person, identity, event, and object, and then construct the context keyword group of each entity pair: identity gender, identity age, identity title, identity nationality, identity ID number, event noun, event phrase, object shape, object noun, and finally construct the transitive relationship between the person, identity, event, and object.

[0014] In the above scheme, the context keyword groups for each entity pair in constructing person-identity, person-event, person-item, item-identity, item-event, and event-identity are specifically: the context keyword groups of the entity pairs, identity gender, identity age, identity title, identity nationality, identity ID number, event nouns, event phrases, item shapes, and item nouns are represented by a four-dimensional matrix, the rows and columns of the matrix correspond to entities respectively, the depth of the matrix corresponds to the context keyword groups, and the transitivity of the matrix corresponds to the file names to which the person, identity, event, and item belong.

[0015] In the above scheme, the knowledge graph of the audit document includes the relationship graph of the generator, the relationship graph of the generator, the relationship graph of the generator and the identity, the relationship graph of the generator, the identity and the event, and the relationship graph of the generator, the identity, the event and the object.

[0016] In the above scheme, the generation of the relationship map of people is specifically as follows: judging whether person-identity points to the meaning value of identity, if so, returning person, identity, and the meaning value of identity, that is, generating the relationship map of people; if not, cyclically judging whether person-identity points to the identity value until the judgment of pointing is successful.

[0017] In the above scheme, the relationship graph between person and identity is generated specifically as follows: according to the person-identity path, it is judged whether the person-identity points to the identity-identity value, whether it points to the identity-identity value, and whether it points to the identity; if so, and the name of the person is equal to the name of the owner of the identity, and the ID number of the person is equal to the ID number of the owner of the identity, then the person, identity, gender, age, job title, nationality, and ID number are returned, that is, the relationship graph between person and identity is generated; if not, it is judged in a loop that the person-identity points to the identity-identity value until the judgment of pointing is successful.

[0018] In the above scheme, the relationship map of person, identity and event is generated specifically as follows: according to the person-identity, person-event, identity-event and person-identity-event, it is judged whether the person-identity points to the identity-identity value, whether it points to the event noun-event value, whether it points to the event phrase-event value and whether it points to the object event sentence-event value; if so, and the person name of the event is equal to the person, the identity name and the person's ID number is equal to the ID number of the person to whom the event belongs, then the person, identity, event phrase, event noun, event sentence, event phrase value, event noun value and event sentence value are returned, i.e., the relationship map of person, identity and event is generated; if not, it is judged repeatedly whether the person-event points to the event value until the judgment of pointing is successful.

[0019] In the above scheme, the relationship map of person, identity, event and object is generated specifically as follows: according to the person-identity, person-event, person-object, object-identity, object-event, event-identity and person-identity-event-object, it is judged whether the person-identity points to the identity-identity value, whether it points to the event-event value, whether it points to the object shape-object value and whether it points to the object noun-object value; if so, and the person name is equal to the name of the person attached to the object, the name of the person to whom the event belongs is equal to the name of the person attached to the object and the person's ID number is equal to the identity ID number of the person attached to the object, then the person, identity, event phrase, event noun, event value, object shape, object noun, shape value and noun value are returned, i.e., the relationship map of person, identity, event and object is generated; if not, the relationship between person-object pointing to object value is judged cyclically until the pointing is judged successfully.

[0020] Compared with the prior art, the present invention generates a knowledge graph from the electronic documents of audit documents for subsequent query and analysis, thereby improving work efficiency and analysis capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The drawings described herein are used to disclose a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0022] Figure 1A flowchart of a method for generating a knowledge graph for audit documents is provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0024] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, article or device. In the absence of more restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, article or device including the element.

[0025] The embodiment of the present invention provides a method for generating a knowledge graph for audit documents, such as Figure 1 As shown, the method is implemented by the following steps:

[0026] Step 101: digitize the paper audit documents to obtain the identified document contents;

[0027] Step 102: Classify the recognized document content according to the keyword group to obtain classification data of people, identities, events, and items respectively;

[0028] Step 103: With the classified data of person, identity, event, and item as the main column, establish the corresponding relationship model of person-identity pointing to identity value, judging person-identity pointing to identity-identity value, person-identity pointing to identity-identity value pointing to event-event value, person-identity pointing to identity-identity value, pointing to event-event value pointing to item-item value;

[0029] Specifically, it at least includes the definition of person-identity gender, person-identity age, person-identity title, person-identity nationality, person-identity ID number, identity age-identity value age, identity gender-identity value gender, event noun-event value, event phrase-event value, item shape-item value, item noun-item value.

[0030] Step 104: According to the relationship model and the transitive relationship between people, identities, events and objects, the classification data of the people, identities, events and objects are associated to obtain the knowledge graph of the audit document.

[0031] Specifically, the transitive relationship between the person, identity, event, and object is as follows: create an entity pair for the person, identity, event, and object, and then construct the context keyword group of each entity pair: identity gender, identity age, identity title, identity nationality, identity ID number, event noun, event phrase, object shape, object noun, and finally construct the transitive relationship between the person, identity, event, and object.

[0032] The transitive relationship is the path between the phrases of people, identities, events, and objects, the paths between the meaning phrases they represent, and the paths between the meaning phrases and the file names, which are integrated into a path of phrases of people, identities, events, and objects→meaning phrases→file names.

[0033] The described construction of context keyword groups for each entity pair in person-identity, person-event, person-item, item-identity, item-event, and event-identity is specifically as follows: the context keyword groups of the entity pairs, identity gender, identity age, identity title, identity nationality, identity ID number, event nouns, event phrases, item shapes, and item nouns are represented by a four-dimensional matrix, the rows and columns of the matrix correspond to entities respectively, the depth of the matrix corresponds to the context keyword groups, and the transitivity of the matrix corresponds to the file names to which the person, identity, event, and item belong.

[0034] The knowledge graph is: by combining the theories and methods of applied mathematics, graphics, information visualization technology, information science and other disciplines with methods such as metrology citation analysis, and using a visual graph to vividly display the theories of the disciplines, the knowledge graph presents the associated people, identities, events, and object data of the search content (people, identities, events, and objects) in the background data storage database in the form of a dynamic and draggable relationship graph.

[0035] The knowledge graph of the audit document includes the relationship graph of the generator, the relationship graph of the generator, the relationship graph between the generator and the identity, the relationship graph between the generator, the identity and the event, and the relationship graph between the generator, the identity, the event and the object.

[0036] The generation of the relationship map of people is specifically as follows: determine whether person-identity (identity includes gender, age, job title, nationality, and ID number) points to the meaning value of identity. If so, return the meaning value of person, identity, and identity, that is, generate the relationship map of people; if not, loop to determine whether person-identity points to identity value until the judgment is successful.

[0037] For example, Zhang San-gender points to male (match), Zhang San-gender points to 13 (not match).

[0038] The generated relationship graph between person and identity is specifically as follows: according to the person-identity path described in claim 5, determine whether person-identity (identity includes gender, age, job title, nationality, and ID number) points to identity (age)-identity value (age), whether it points to identity (gender)-identity value (gender), and whether it points to identity. If so, and the name of the person is equal to the name of the person to whom the identity belongs, and the ID number of the person is equal to the ID number of the person to whom the identity belongs, then return person, identity, gender, age, job title, nationality, and ID number, that is, generate a relationship graph between person and identity; if not, loop to determine whether person-identity points to identity-identity value until the determination of pointing is successful.

[0039] The generated relationship map of person, identity and event is specifically as follows: according to claim 5, the person-identity, person-event, identity-event and person-identity-event judge whether the person-identity points to the identity-identity value, whether it points to the event noun-event value, whether it points to the event phrase-event value and whether it points to the object event sentence-event value. If so, and the person name of the event is equal to the person, the identity name and the person's ID number are equal to the ID number of the person to whom the event belongs, then return the person, identity, event phrase, event noun, event sentence, event phrase value, event noun value and event sentence value, i.e. generate the relationship map of person, identity and event; if not, loop to judge whether the person-event points to the event value until the judgment of pointing is successful.

[0040] The generated relationship map of person, identity, event and object is specifically as follows: the person-identity, person-event, person-object, object-identity, object-event, event-identity, person-identity-event-object judges whether person-identity points to identity-identity value, whether it points to event-event value, whether it points to object shape-object value, whether it points to object noun-object value; if so, and the person name is equal to the name of the person attached to the object, the name of the person to whom the event belongs is equal to the name of the person attached to the object, and the person's ID number is equal to the identity ID number of the person attached to the object, then return person, identity, event phrase, event noun, event value, object shape, object noun, shape value, noun value, that is, generate the relationship map of person, identity, event and object; if not, loop to judge the relationship between person-object pointing to object value until the judgment of pointing is successful.

[0041] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention.

Claims

1. A knowledge graph generation method for audit documents, characterized in that: The method is: Digitalize paper audit documents to obtain the identified document contents; Classify the recognized document content according to the keyword group to obtain person, identity, event, and item classification data respectively; Taking the classified data of people, identities, events, and items as the main columns, establish the corresponding relationship models of people-identity pointing to identity value, judging people-identity pointing to identity-identity value, people-identity pointing to identity-identity value pointing to event-event value, people-identity pointing to identity-identity value, pointing to event-event value pointing to item-item value; According to the relationship model and the transitive relationship between people, identities, events and objects, the classification data of the people, identities, events and objects are associated to obtain the knowledge graph of the audit document; The knowledge graph of the audit document includes a relationship graph of the generator, a relationship graph of the generator and the identity, a relationship graph of the generator, the identity and the event, and a relationship graph of the generator, the identity, the event and the object; The generating of the relationship map of the person is specifically as follows: determining whether the person-identity points to the meaning value of the identity, if so, returning the person, the identity, and the meaning value of the identity, that is, generating the relationship map of the person; if not, cyclically determining whether the person-identity points to the identity value, until the determination of pointing is successful; The generating of the relationship graph between the person and the identity is specifically as follows: judging whether the person-identity points to the identity-identity value, whether it points to the identity-identity value, and whether it points to the identity according to the person-identity path; if so, and the name of the person is equal to the name of the person to whom the identity belongs, and the ID number of the person is equal to the ID number of the person to whom the identity belongs, then returning the person, identity, gender, age, job title, nationality, and ID number, i.e. generating the relationship graph between the person and the identity; if not, cyclically judging whether the person-identity points to the identity-identity value until the judgment of pointing is successful; The generating of the relationship graph of person, identity and event is specifically as follows: judging whether person-identity points to identity-identity value, whether it points to event noun-event value, whether it points to event phrase-event value, whether it points to object event sentence-event value according to person-identity, person-event, identity-event, person-identity-event; if so, and the person name of the event is equal to the person, identity name, and the person's ID number is equal to the ID number of the person to whom the event belongs, then returning person, identity, event phrase, event noun, event sentence, event phrase value, event noun value, event sentence value, i.e. generating the relationship graph of person, identity and event; if not, cyclically judging whether person-event points to event value until the judgment of pointing is successful; The generated relationship map of person, identity, event and object is specifically as follows: the person-identity, person-event, person-object, object-identity, object-event, event-identity, person-identity-event-object judges whether person-identity points to identity-identity value, whether it points to event-event value, whether it points to object shape-object value, whether it points to object noun-object value; if so, and the person name is equal to the name of the person attached to the object, the name of the person to whom the event belongs is equal to the name of the person attached to the object, and the person's ID number is equal to the identity ID number of the person attached to the object, then return person, identity, event phrase, event noun, event value, object shape, object noun, shape value, noun value, that is, generate the relationship map of person, identity, event and object; if not, loop to judge the relationship between person-object pointing to object value until the judgment of pointing is successful.

2. The knowledge graph generation method for audit documents according to claim 1 is characterized in that: The described relationship model of person, identity, event, and item classification data as the main column is established to establish corresponding person, identity, event, and item person-identity pointing to identity value, judging person-identity pointing to identity-identity value, person-identity pointing to identity-identity value pointing to event-event value, person-identity pointing to identity-identity value, and pointing to event-event value pointing to item-item value, specifically: at least including the definition of person-identity gender, person-identity age, person-identity title, person-identity nationality, person-identity ID number, identity age-identity value age, identity gender-identity value gender, event noun-event value, event phrase-event value, item shape-item value, item noun-item value.

3. The knowledge graph generation method for audit documents according to claim 2 is characterized in that: The transitive relationship between the person, identity, event, and object is specifically as follows: create an entity pair for the person, identity, event, and object, then construct the context keyword group of each entity pair: identity gender, identity age, identity title, identity ethnicity, identity ID number, event noun, event phrase, object shape, object noun, and finally construct the transitive relationship between the person, identity, event, and object.

4. The knowledge graph generation method for audit documents according to claim 3 is characterized in that: The described construction of context keyword groups for each entity pair in person-identity, person-event, person-item, item-identity, item-event, and event-identity is specifically as follows: the context keyword groups of the entity pairs, identity gender, identity age, identity title, identity nationality, identity ID number, event nouns, event phrases, item shapes, and item nouns are represented by a four-dimensional matrix, the rows and columns of the matrix correspond to entities respectively, the depth of the matrix corresponds to the context keyword groups, and the transitivity of the matrix corresponds to the file names to which the person, identity, event, and item belong.