A method and system for semantic association of airworthiness regulations structure

Abstract

The present invention discloses a method and system for semantic association of airworthiness regulations structure. The method comprises: formatting the airworthiness regulations text; constructing a clause level feature rule template according to the hierarchical structure features of the airworthiness regulations text after formatting; identifying the clause level of the current row unit according to the clause level feature rule template, extracting the basic clause unit content as a node, and assigning a corresponding attribute value to each basic clause unit; the attribute value is the title level; judging whether there is a semantic connection relationship between each node pair according to the edge structure judgment mechanism; and performing structural conversion on the airworthiness regulations text according to the judgment result. The present invention converts the special clause structure of the airworthiness regulations into a general type of natural language structure with complete semantic information and syntactic structure, eliminating the adverse effects of the special clause structure of the airworthiness regulations text on natural language processing related tasks, and facilitating subsequent data intelligent, automated analysis and processing.

Description

Technical Field

[0001] The present invention relates to the field of airworthiness safety technology, and in particular to an airworthiness regulation structure semantic association method and system. Background Art

[0002] Natural language processing technology integrates the knowledge framework of linguistics and computer science. Its purpose is to understand, analyze and apply human language through computers, thereby improving the effectiveness of communication and control between people and machines. At present, the objects of natural language processing are mainly texts in general language form, such as: "Yao Ming's height is 226 cm." This type of text data does not have a special format, and its semantic information and grammatical structure are complete, which is suitable for natural language processing related tasks. The safety information contained in the airworthiness regulations is organized and stored in the special article form structure of laws and regulations. It has a hierarchical structure and the text structure is obviously different from the general type of language form. The special text structure makes it difficult to directly apply existing natural language processing technology to airworthiness regulations texts.

[0003] Graph theory belongs to applied mathematics. This theory represents things with points and the connections between things with lines. Graph-based natural language processing methods, such as rhetorical structure theory and dependency syntax analysis, can solve the problems of paragraph analysis and syntax analysis of general types of texts, but they cannot be applied to airworthiness regulations. The main reasons include the following two points: 1) All article units of airworthiness regulations (in terms of articles, clauses, items, and items) have the problem of incomplete syntactic structure, that is, semantic information. They must be combined with the corresponding superior basic article units to form a complete syntactic structure and express complete semantic information; 2) The hierarchical structure relationship of legal articles leads to multiple and multiple nested relationships between entities, which increases the complexity of natural language processing tasks such as syntax analysis, paragraph-level relationship extraction, and knowledge graph construction. For the special article hierarchical structure of airworthiness regulations, there is still a lack of corresponding text structure processing and semantic connection methods. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for semantic association of airworthiness regulations structure, which can convert the special article structure of airworthiness regulations into a general type of natural language structure with complete semantic information and syntactic structure according to the inherent hierarchical characteristics of the article structure of airworthiness regulations text, eliminate the adverse effects of the special article structure of airworthiness regulations text on natural language processing related tasks, and facilitate subsequent intelligent and automated data analysis and processing. The present invention can promote the application of natural language processing technology in the field of airworthiness safety, and further promote the automated and intelligent development of airworthiness safety analysis and certification.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for semantic association of airworthiness regulation structures, comprising:

[0007] Clean up the format of airworthiness regulations text;

[0008] According to the hierarchical structure characteristics of the airworthiness regulations text after format cleaning, a rule template for the hierarchical characteristics of the clauses is constructed;

[0009] According to the article level feature rule template, the article level of the current row unit is identified, the basic article unit content is extracted as a node, and a corresponding attribute value is assigned to each basic article unit; the attribute value is the title level;

[0010] According to the edge structure determination mechanism, determining whether there is a semantic connection relationship between adjacent nodes;

[0011] The airworthiness regulations text is structurally transformed based on the judgment results.

[0012] Optionally, after performing structural conversion on the airworthiness regulations text according to the judgment result, the method further includes: performing post-processing on the converted airworthiness regulations text including removing duplicate lines and correcting punctuation.

[0013] Optionally, judging whether there is a semantic connection relationship between adjacent nodes according to an edge structure judgment mechanism specifically includes:

[0014] If the attribute value of the current node is greater than the attribute value of the previous adjacent node, it is judged that there is a semantic connection relationship between the current node and the previous adjacent node, the two nodes form a directed link, and the corresponding basic clause unit contents have a semantic connection relationship;

[0015] If the attribute value of the current node is equal to the attribute value of the previous adjacent node, it is determined that there is no semantic connection relationship between the current node and the previous adjacent node, and the semantic connection node is re-determined;

[0016] If the attribute value of the current node is less than the attribute value of the previous adjacent node, it is determined that there is no semantic connection relationship between the current node and the previous adjacent node, and the semantic connection node is re-determined;

[0017] Optionally, the clause level feature rule template includes a first rule template and a second rule template;

[0018] The first-level title in the first rule template is the CCAR department number, the second-level title is the article, the third-level title is represented by brackets and lowercase English letters, the fourth-level title is represented by Arabic numerals in brackets, the fifth-level title is represented by Roman numerals in brackets, and the sixth-level title is represented by uppercase English letters in brackets;

[0019] The first-level title in the second rule template is the CCAR department number, the second-level title is the article, the third-level title is represented by Chinese numbers in brackets, the fourth-level title is represented by Arabic numerals and the symbol '.', and the fifth-level title is represented by Arabic numerals in brackets.

[0020] The present invention also provides an airworthiness regulations structure semantic association system, comprising:

[0021] Format cleaning module, used to clean the format of airworthiness regulations text;

[0022] The template construction module is used to construct a rule template for clause hierarchical characteristics according to the hierarchical structure characteristics of the airworthiness regulations text after format cleaning;

[0023] The identification and extraction module is used to identify the level of the current line unit article according to the article level feature rule template, extract the basic article unit content as a node, and assign a corresponding attribute value to each basic article unit; the attribute value is the title level;

[0024] A judgment module, used to judge whether there is a semantic connection relationship between adjacent nodes according to an edge structure judgment mechanism;

[0025] The conversion module is used to perform structural conversion on the airworthiness regulations text according to the judgment results.

[0026] Optionally, it also includes:

[0027] The post-processing module is used to perform post-processing on the converted airworthiness regulations text, including removing duplicate lines and correcting punctuation.

[0028] Optionally, judging whether there is a semantic connection relationship between adjacent nodes according to an edge structure judgment mechanism specifically includes:

[0029] A first result determination unit is used to determine that, when the attribute value of the current node is greater than the attribute value of the previous adjacent node, there is a semantic connection relationship between the current node and the previous adjacent node, the two nodes form a directed link, and the corresponding basic clause unit contents have a semantic connection relationship;

[0030] A second result determination unit is used to determine that there is no semantic connection relationship between the current node and the previous adjacent node when the attribute value of the current node is equal to the attribute value of the previous adjacent node, and then re-determine the semantic connection node;

[0031] A third result determination unit, configured to determine that there is no semantic connection relationship between the current node and the previous adjacent node when the attribute value of the current node is less than the attribute value of the previous adjacent node, and then redetermine the semantic connection node;

[0032] Optionally, the clause level feature rule template includes a first rule template and a second rule template;

[0033] The first-level title in the first rule template is the CCAR department number, the second-level title is the article, the third-level title is represented by brackets and lowercase English letters, the fourth-level title is represented by Arabic numerals in brackets, the fifth-level title is represented by Roman numerals in brackets, and the sixth-level title is represented by uppercase English letters in brackets;

[0034] The first-level title in the second rule template is the CCAR department number, the second-level title is the article, the third-level title is represented by Chinese numbers in brackets, the fourth-level title is represented by Arabic numerals and the symbol '.', and the fifth-level title is represented by Arabic numerals in brackets.

[0035] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0036] The present invention discloses a method and system for semantic association of airworthiness regulations structure. The method comprises: formatting the airworthiness regulations text; constructing a clause hierarchical feature rule template according to the hierarchical structure features of the airworthiness regulations text after formatting; identifying the clause level of the current row unit according to the clause hierarchical feature rule template, extracting the basic clause unit content as a node, and assigning a corresponding attribute value to each basic clause unit; the attribute value is the title level; judging whether there is a semantic connection relationship between adjacent nodes according to the edge structure judgment mechanism; and performing structural conversion on the airworthiness regulations text according to the judgment result. The present invention converts the special clause structure of the airworthiness regulations into a general type of natural language structure with complete semantic information and syntactic structure based on the inherent hierarchical features of the clause structure of the airworthiness regulations text, eliminates the adverse effects of the special clause structure of the airworthiness regulations text on natural language processing related tasks, and facilitates subsequent data intelligent, automated analysis and processing. The present invention can promote the application of natural language processing technology in the field of airworthiness safety, and further promote the automated and intelligent development of airworthiness safety analysis and certification. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0038] Figure 1 The edge structure determination mechanism with semantic connection relationship in the embodiment of the present invention

[0039] Figure 2 The airworthiness regulations text attribute directed graph model of the embodiment of the present invention;

[0040] Figure 3 A flowchart of a method for semantic association of airworthiness regulations structure according to an embodiment of the present invention;

[0041] Figure 4 This is the content of the airworthiness regulation 23.609 of CCAR-23-R3 of the embodiment of the present invention;

[0042] Figure 5 The airworthiness regulations structure conversion and semantic connection results of the embodiment of the present invention;

[0043] Figure 6 This is a diagram of the structural conversion and semantic connection process for the airworthiness regulations of Article 23.609 of CCAR-23-R3 according to an embodiment of the present invention;

[0044] Figure 7 The airworthiness regulations text structure form of the Class I structure of the embodiment of the present invention and the airworthiness regulations text structure form of the Class I structure after being processed by the algorithm;

[0045] Figure 8 The airworthiness regulations text structure form of the Class II structure of the embodiment of the present invention and the airworthiness regulations text structure form of the Class II structure after being processed by the algorithm;

[0046] Fig. 9 It is the CPU time required for processing the airworthiness regulations text for Category I and Category II structures according to the embodiment of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] The purpose of the present invention is to provide a method and system for semantic association of airworthiness regulations structure, which can convert the special article structure of airworthiness regulations into a general type of natural language structure with complete semantic information and syntactic structure according to the inherent hierarchical characteristics of the article structure of airworthiness regulations text, eliminate the adverse effects of the special article structure of airworthiness regulations text on natural language processing related tasks, and facilitate subsequent intelligent and automated data analysis and processing. The present invention can promote the application of natural language processing technology in the field of airworthiness safety, and further promote the automated and intelligent development of airworthiness safety analysis and certification.

[0049] In order to solve the problem of structural conversion and semantic connection of airworthiness regulations, the present invention takes graph theory as the mathematical basis and provides a new semantic connection theory of airworthiness regulations. At the same time, in view of the hierarchical structure characteristics of the text of Chinese airworthiness regulations, a new algorithm for structural conversion and semantic connection of regulations text is provided. The present invention uses the hierarchical characteristics inherent in the structure of the Chinese airworthiness regulations as the attributes of the attribute directed graph, and uses the multi-level title form as the node attribute value. By setting the edge structure judgment mechanism under different conditions, semantically associated edges are constructed between node pairs with semantic connection relationships. Therefore, the problem of text structural conversion and semantic connection of Chinese airworthiness regulations is abstractly expressed as a graph theory problem, that is, the problem of edge set construction of the attribute directed graph model of airworthiness regulations text. For the attribute directed graph model of airworthiness regulations text, legal sentences with complete semantic information can be represented by the maximum directed link.

[0050] The present invention organizes and systematically analyzes the airworthiness regulations documents that are available for public download on the official website of the Civil Aviation Administration of China for the currently valid Chinese airworthiness regulations texts, and sorts out two different hierarchical structure presentation methods. According to the two different structural characteristics of the Chinese airworthiness regulations texts, regular expressions for basic clause unit extraction and node attribute value assignment are designed respectively. Finally, two sets of rule templates are formed that can extract the basic clause units of airworthiness regulations (as graph model nodes) based on the text structure characteristics, and assign corresponding attribute values ​​(corresponding to the title level) to the nodes.

[0051] The present invention provides a theory of semantic association of airworthiness regulations structure, which includes four elements, corresponding to definition, theorem, model and rule respectively.

[0052] Definition: attribute directed graph;

[0053] Model: Airworthiness regulations text attribute directed graph model;

[0054] Theorem: Graphability theorem of satisfaction degree sequence;

[0055] Rules: Edge structure determination mechanism with semantic associations.

[0056] The theory provided by the present invention is described in detail below in conjunction with airworthiness regulations.

[0057] Definition: (Attributed directed graph) Suppose there is a vertex set V = {v1, v2, ..., v n}, and attribute set VL = {vl1, vl2, ..., vl n}, each vertex has one or more specific attributes in the attribute set, and the corresponding attribute value set is expressed as L = {l1, l2, ..., l n}, then vertex v i Represented as v i (VLi (L i )). Assume that there exists an edge set E = {e1, e1, ..., e n}, where e i =(v i , v j )∈E represents an edge that satisfies the key attribute constraint. The parameter ψ is the edge structure determination mechanism, which is used to determine whether two nodes meet the key attribute constraint. If the vertex set and edge set meet the following four conditions:

[0058] (1) Each node has corresponding attributes and attribute values

[0059] (2) The degree sequence satisfies graphability

[0060] (3) Satisfying the unidirectional connectivity of the edge structure

[0061] (4) Determination function ψ(v i , v j ): A structure used to determine whether there is an edge between two adjacent nodes

[0062]

[0063] Among them, ψ = 1 indicates that there is a semantic connection relationship between the two nodes, and an edge should be constructed between the node pair, otherwise, the edge structure does not exist. Therefore, GL = (V (VL (L)), E, ​​ψ) is defined as an attribute directed graph, referred to as GL.

[0064] The basic idea of ​​the airworthiness regulations text structure conversion and semantic connection algorithm proposed in the present invention is to construct a directed graph model of airworthiness regulations text attributes based on the structural characteristics of airworthiness regulations. The model uses the basic article units of airworthiness regulations (the article content corresponding to each level of title) as nodes, and establishes edges between node pairs with semantic connection relationships through an edge structure judgment mechanism. Finally, the maximum directed link that can express complete semantic information is obtained, and the airworthiness regulations text article structure conversion and semantic connection tasks are realized, and the special article structure of the airworthiness regulations is converted into a general type of language form with complete semantic information and syntactic structure.

[0065] The present invention presents the hierarchical structure of airworthiness regulations through parameter L (attribute value set) and controls the hierarchical structure conversion process. The Chinese airworthiness regulations structure contains 6 different title levels (i.e., Level I-Level VI), corresponding to CCAR part number, article, clause, item, item, and sub-item respectively. The attribute value set L of the airworthiness regulations text is shown in Formula 2.

[0066] L = (1, 2, ... 6) (2)

[0067] By judging the title level of two adjacent nodes (i.e., the basic text units corresponding to adjacent text line units), the hierarchical transfer process is controlled, and an edge structure determination mechanism is designed. According to different situations, edges are constructed between node pairs with complete semantic information, thereby achieving the task of text structure conversion and semantic connection. The problem is defined as follows:

[0068]

[0069] The hierarchical transfer judgment process of airworthiness regulations structure conversion and semantic connection algorithm is shown in Formula 4.

[0070]

[0071] Among them, i and i-1 represent the row numbers corresponding to the current row and the previous row respectively, and i is greater than 1 and less than the maximum row value n of the current airworthiness regulations text. i and i-1 Respectively represent the title levels of the current row and the previous row. The level transfer process is controlled by calculating the difference between the title level of the current row and the title level of the previous row, and the structural conversion and semantic connection methods in the transfer process are determined. The determination method is shown in Formula 5.

[0072]

[0073] like Figure 1 As shown, different situations are explained:

[0074] Case 1: When F>0, then ψ(v i , v i-1 )=1. It indicates that the current node v i and the previous adjacent node v i-1 There is a semantic connection relationship between nodes v i Yes i-1 Therefore, there is an associated edge structure between the node pair, which is represented by e(v i-1 , v i ). The two nodes should form a directed link, and the corresponding basic article unit contents have a semantic connection relationship.

[0075] Case 2: When F = 0, then ψ(v i , v i-1 )=0, indicating that the current node v i and the previous adjacent node v i-1 There is no semantic connection relationship between nodes v i With v i-1 The two nodes have the same attribute value. Therefore, the current node v i Should be attached to the list later than the current node v iA node at a lower heading level, represented by: (v f , l i -1), where v f Represents node v i Before, and level is l i -1 node. That is, the node (v f , l i -1) should be the same as the node (v i , l i ) form a directed link, and the corresponding basic clause units have a semantic connection relationship. The edge structure with semantic connection and hierarchical affiliation relationship should exist at node v f With node v i Between, expressed as e(v f , v i ).

[0076] Case 3: When F<0, then ψ(v i , v i-1 )=0, the situation is similar to Case 2, and the edge structure with semantic connection and hierarchical affiliation is represented by e(v f , v i ). Node v i With v i-1 Both nodes have the same heading level, and v i Yes i-1 Therefore, the current node v i Should be attached to the list later than the current node v i A node at a lower heading level, represented by: (v f , l i -1). That is, node (v f , l i -1) should be the same as the node (v i , l i ) form a directed link, and the corresponding basic clause units have a semantic connection relationship. The edge structure with semantic connection and hierarchical affiliation relationship should exist at node v f With node v i between.

[0077] Therefore, the edge structure construction mechanism is expressed as:

[0078]

[0079] The attribute directed graph model of airworthiness regulations text is as follows: Figure 2 As shown, the nodes represent the basic content of the article units, the lines represent the edges with semantic associations, and the arrows represent the semantic associations between the article structures.

[0080] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0081] A method for semantic association of airworthiness regulation structures comprises the following steps:

[0082] Step 101: Format and clean the airworthiness regulations text.

[0083] Step 102: According to the hierarchical structure characteristics of the airworthiness regulations text after format cleaning, construct a rule template for article hierarchical characteristics. The rule template for article hierarchical characteristics includes a first rule template and a second rule template; the first-level title in the first rule template is the CCAR part number, the second-level title is the article, the third-level title is represented by brackets and lowercase English letters, the fourth-level title is represented by Arabic numerals in brackets, the fifth-level title is represented by Roman numerals in brackets, and the sixth-level title is represented by uppercase English letters in brackets; the first-level title in the second rule template is the CCAR part number, the second-level title is the article, the third-level title is represented by Chinese numerals in brackets, the fourth-level title is represented by Arabic numerals and the symbol '.', and the fifth-level title is represented by Arabic numerals in brackets.

[0084] Step 103: According to the article level feature rule template, identify the article level of the current row unit, extract the basic article unit content as a node, and assign a corresponding attribute value to each basic article unit; the attribute value is the title level.

[0085] Step 104: Determine whether there is a semantic connection relationship between adjacent nodes according to the edge structure determination mechanism.

[0086] Step 105: Perform structural transformation on the airworthiness regulations text according to the judgment result.

[0087] After step 105, the method further includes:

[0088] Step 106: Post-process the converted airworthiness regulations text, including removing duplicate lines and correcting punctuation.

[0089] Step 104 specifically includes:

[0090] If the attribute value of the current node is greater than the attribute value of the previous adjacent node, it is judged that there is a semantic connection relationship between the current node and the previous adjacent node, the two nodes form a directed link, and the corresponding basic clause unit contents have a semantic connection relationship;

[0091] If the attribute value of the current node is equal to the attribute value of the previous adjacent node, it is determined that there is no semantic connection relationship between the current node and the previous adjacent node, and the semantic connection node is re-determined;

[0092] If the attribute value of the current node is less than the attribute value of the previous adjacent node, it is determined that there is no semantic connection relationship between the current node and the previous adjacent node, and the semantic connection node is re-determined.

[0093] The present invention is described in detail using specific examples below:

[0094] The flow chart of this method is as follows Figure 3 As shown, the Chinese Airworthiness Regulations CCAR-23-R3 is cited as an embodiment of the present invention, see Figure 4 . The conversion results of the embodiment are shown in Figure 5 (a), 5(b), 5(c); Figure 5 (a) The result of identifying the hierarchical structure of the title of the airworthiness regulations in Section 23.609 of CCAR-23-R3. Figure 5 (b) The semantic connection between the text structure of Article 23.609 of the airworthiness regulations in CCAR-23-R3. Figure 5 (c) The structural transformation and semantic connection result of the airworthiness regulation 23.609 of CCAR-23-R3. The structural transformation and semantic connection process of the airworthiness regulation 23.609 of CCAR-23-R3 is as follows: Figure 6 shown.

[0095] (1) Obtain 119 currently valid airworthiness regulations that are publicly available for download on the website of the Civil Aviation Administration of China. Clean the format of the airworthiness regulations to eliminate problems such as typos, line breaks, and punctuation marks that are introduced due to format conversion, and ensure that the content of the airworthiness regulations is represented by text lines as basic text units. The airworthiness regulations conversion process cited in the embodiment is as follows: Figure 6 shown.

[0096] (2) The text structure characteristics of airworthiness regulations were analyzed, the hierarchical characteristics of the article structure were sorted out, and two sets of rule templates were constructed based on two different structural characteristics, which can extract basic article units (as vertices) and assign corresponding attribute values ​​(corresponding to title levels). The construction of rule templates was achieved by designing regular expressions combined with hierarchical characteristics. The two different hierarchical characteristics are shown in Tables 1 and 2. Tables 1 and 2 describe in detail the hierarchical structure characteristics of the two types of airworthiness regulations and the corresponding title levels, and give examples of the title forms corresponding to the specific hierarchical structures.

[0097] Table 1. Class I language structure of airworthiness regulations hierarchical feature rule template

[0098]

[0099]

[0100] Table 2. Class II language structure of airworthiness regulations hierarchical feature rule template

[0101]

[0102] For the first rule template, as can be seen from Table 1, the first-level title is the CCAR department number, the second-level title is the article, and the extraction rules are designed by recognizing Chinese characters, numbers, and the corresponding positions of the two. The third-level title is represented by brackets and lowercase English letters, the fourth-level title is represented by Arabic numerals in brackets, the fifth-level title is represented by Roman numerals in brackets, and the sixth-level title is represented by uppercase English letters in brackets. Based on the above features, the extraction rules are designed, and the article level recognition result of the embodiment is shown in 5(a).

[0103] For the second type of airworthiness regulations text, it can be seen from Table 2 that the highest title level in the airworthiness regulations text is level five. The first level is the CCAR number, and the second level is the article. The extraction rules can be designed by Chinese characters, numbers, and the corresponding character features of the two. The third level title is represented by Chinese numbers in brackets, the fourth level title is represented by Arabic numerals and the symbol '.', and the fifth level title is represented by Arabic numerals in brackets. Based on the above features, the extraction rules are designed.

[0104] (3) Once the first-level title (i.e., CCAR part number) is identified, the procedure begins and the airworthiness regulations text is read line by line.

[0105] (4) Identify the level of the current line unit article according to the article structure level extraction template, extract the basic article unit content as a node, and assign a corresponding attribute value, i.e., the title level, to each basic article unit.

[0106] (5) According to the edge structure judgment mechanism (such as Figure 1 ), according to three different situations, determine whether there is an edge with a semantic connection relationship between node pairs, that is, analyze the relationship between the attribute value of the basic article unit on the current row unit and the corresponding attribute value of the basic article unit on the previous row unit, so as to control the hierarchical transmission process.

[0107] If F>0, then ψ(v i , v i-1 )=1. It indicates that the current node v i and the previous adjacent node v i-1 There is a semantic connection between them (the sub-title is set to supplement the unfinished content of the upper-level title), node v i Yes i-1 Therefore, there is a semantically related edge between the node pair, and the edge structure is expressed as e(v i-1 , v i ). The two nodes should form a directed link, and the corresponding basic article unit contents have a semantic connection relationship.

[0108] If F = 0, then ψ(v i , v i-1 )=0, indicating that the current node v i and the previous adjacent node v i-1 There is no semantic connection relationship between nodes v i With v i-1 The two nodes have the same attribute value (the current row and the previous row belong to the same level of headings, which means that the semantic information of the regulatory provisions expressed in the previous row is complete). i Should be attached to the list later than the current node v i A node at a lower heading level, represented by: (v f , l i -1), where v f Represents node v i Before, and level is l i -1 node. That is, the node (v f , l i -1) should be the same as the node (v i , l i ) form a directed link, and the corresponding basic clause units have a semantic connection relationship. The edge structure with semantic connection and hierarchical affiliation relationship should exist at node v f With node v i Between, expressed as e(v f , v i ).

[0109] If F<0, then ψ(v i , v i-1 )=0, the edge structure with semantic connection and hierarchical affiliation is represented by e(v f , v i ). Node v i With v i-1 Both nodes have the same heading level, and v i Yes i-1 The upper node of v (up to the previous line, a piece of legal text with a complete semantic structure has been successfully converted). Therefore, the current node v i Should be attached to the list later than the current node v i A node at a lower heading level, represented by: (v f , l i -1). That is, node (v f , l i -1) should be the same as the node (v i , l i ) form a directed link, and the corresponding basic clause units have a semantic connection relationship. The edge structure with semantic connection and hierarchical affiliation relationship should exist at node vf With node v i between.

[0110] (6) Return to step (3).

[0111] (7) Construct the corresponding maximum link with semantic connection relationship for the last node and write it out.

[0112] (8) After all nodes have been executed, the newly generated text is post-processed, including removing duplicate lines and correcting punctuation.

[0113] According to the above steps, we can get Figure 5 (c) shows the result of the text structure conversion of the airworthiness regulations. The text structure conversion algorithm of the present invention is executed on the text of the 119 currently effective airworthiness regulations. The text structure of the airworthiness regulations of type I structure is as follows: Figure 7 (a) As shown in the figure, the text structure of the airworthiness regulations for Category II structures is as follows: Figure 8 (a) shows the results of text structure conversion and semantic connection processing of the two Chinese airworthiness regulations. Figure 7 (b) with Figure 8 As shown in (b), the required CPU time is Fig. 9 (a) with Fig. 9 (b) Fig. 9 The CPU time shown in the figure is the average CPU time of executing 50 experiments, which is given by Fig. 9 It can be seen that the airworthiness regulations text clause structure processing algorithm of the present invention can complete the structure conversion task in a very short time.

[0114] The present invention also provides an airworthiness regulations structure semantic association system, comprising:

[0115] The format cleaning module is used to clean the format of airworthiness regulations text.

[0116] The template construction module is used to construct a clause level feature rule template based on the hierarchical structure features of the airworthiness regulations text after format cleaning.

[0117] The identification and extraction module is used to identify the current line unit article level according to the article level feature rule template, extract the basic article unit content as a node, and assign a corresponding attribute value to each basic article unit; the attribute value is the title level.

[0118] The judgment module is used to judge whether there is a semantic connection relationship between adjacent nodes according to the edge structure judgment mechanism.

[0119] The conversion module is used to perform structural conversion on the airworthiness regulations text according to the judgment results.

[0120] Also includes:

[0121] The post-processing module is used to perform post-processing on the converted airworthiness regulations text, including removing duplicate lines and correcting punctuation.

[0122] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0123] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for semantic association of airworthiness regulations structure, characterized in that: include: Clean up the format of airworthiness regulations text; According to the hierarchical structure characteristics of the airworthiness regulations text after format cleaning, a rule template for the hierarchical characteristics of the clauses is constructed; According to the article level feature rule template, the article level of the current row unit is identified, the basic article unit content is extracted as a node, and a corresponding attribute value is assigned to each basic article unit; the attribute value is the title level; According to the edge structure determination mechanism, determine whether there is a semantic connection relationship between the adjacent nodes; specifically, if the attribute value of the current node is greater than the attribute value of the previous adjacent node, then determine that there is a semantic connection relationship between the current node and the previous adjacent node, the two nodes form a directed link, and the corresponding basic article unit content has a semantic connection relationship; if the attribute value of the current node is equal to the attribute value of the previous adjacent node, then determine that there is no semantic connection relationship between the current node and the previous adjacent node, and then re-determine the semantic connection node; if the attribute value of the current node is less than the attribute value of the previous adjacent node, then determine that there is no semantic connection relationship between the current node and the previous adjacent node, and then re-determine the semantic connection node; The airworthiness regulations text is structurally transformed based on the judgment results.

2. The method for semantic association of airworthiness regulations structure according to claim 1, characterized in that: After the structural conversion of the airworthiness regulations text according to the judgment result, it also includes: The converted airworthiness regulations text is post-processed, including removing duplicate lines and correcting punctuation.

3. The method for semantic association of airworthiness regulations structure according to claim 1, characterized in that: The clause level feature rule template includes a first rule template and a second rule template; The first-level title in the first rule template is the CCAR department number, the second-level title is the article, the third-level title is represented by brackets and lowercase English letters, the fourth-level title is represented by Arabic numerals in brackets, the fifth-level title is represented by Roman numerals in brackets, and the sixth-level title is represented by uppercase English letters in brackets; The first-level title in the second rule template is the CCAR department number, the second-level title is the article, the third-level title is represented by Chinese numbers in brackets, the fourth-level title is represented by Arabic numerals and the symbol '.', and the fifth-level title is represented by Arabic numerals in brackets.

4. An airworthiness regulations structure semantic association system, characterized in that: Includes |: Format cleaning module, used to clean the format of airworthiness regulations text; The template construction module is used to construct a rule template for clause level characteristics according to the hierarchical structure characteristics of the airworthiness regulations text after format cleaning; The identification and extraction module is used to identify the level of the current line unit article according to the article level feature rule template, extract the basic article unit content as a node, and assign a corresponding attribute value to each basic article unit; the attribute value is the title level; A judgment module is used to judge whether there is a semantic connection relationship between adjacent nodes according to an edge structure judgment mechanism; specifically comprising: a first result determination unit, used to determine that there is a semantic connection relationship between the current node and the previous adjacent node when the attribute value of the current node is greater than the attribute value of the previous adjacent node, the two nodes form a directed link, and the corresponding basic article unit contents have a semantic connection relationship; a second result determination unit, used to determine that there is no semantic connection relationship between the current node and the previous adjacent node when the attribute value of the current node is equal to the attribute value of the previous adjacent node, and then redetermine the semantic connection node; a third result determination unit, used to determine that there is no semantic connection relationship between the current node and the previous adjacent node when the attribute value of the current node is less than the attribute value of the previous adjacent node, and then redetermine the semantic connection node; The conversion module is used to perform structural conversion on the airworthiness regulations text according to the judgment results.

5. The airworthiness regulations structure semantic association system according to claim 4, characterized in that: Also includes: The post-processing module is used to perform post-processing on the converted airworthiness regulations text, including removing duplicate lines and correcting punctuation.

6. The airworthiness regulations structure semantic association system according to claim 4, characterized in that: The clause level feature rule template includes a first rule template and a second rule template; The first-level title in the first rule template is the CCAR department number, the second-level title is the article, the third-level title is represented by brackets and lowercase English letters, the fourth-level title is represented by Arabic numerals in brackets, the fifth-level title is represented by Roman numerals in brackets, and the sixth-level title is represented by uppercase English letters in brackets; The first-level title in the second rule template is the CCAR department number, the second-level title is the article, the third-level title is represented by Chinese numbers in brackets, the fourth-level title is represented by Arabic numerals and the symbol '.', and the fifth-level title is represented by Arabic numerals in brackets.

Images