KR101004141B1 - Structural Statement Checking Method through CSS Conversion and Schema Matching of Text File - Google Patents

Abstract

The present invention is to prepare a structural statement standard information model reflecting the standardized design review items, and to define the hierarchical structure of each item of the structural statement prepared by the user, in order to review the design document missing items of the structural statement prepared by the user, Through the process of converting to an XML SCHEMA DEFINITION (XSD) file, the XML schema matching technique with the structural statement standard information model is performed to determine whether there is a missing design item of the structural statement prepared in practice.

The method proposed in the present invention is expected to be used to improve the efficiency and accuracy of structural bill design review.

Schema Matching, Design Items, Civil Engineering, Structural Statement, Tree, Transformation, XSD

Description

METHODS FOR ANALYZING AND SCHEMA MATCHING STRUCTURAL CALCULATION DOCUMENT CONVERTED INTO XML}

The present invention relates to the field of construction technology information system, and in detail, converts a structural statement text information written by a user into an XML (Extensible Markup Language) or XSD (XML SCHEMA DEFINITION) format document and standardizes the XML of a structural statement. The present invention relates to a method of checking whether there is a missing item in a structured statement prepared by a user by performing schema matching with data.

The structural calculation of the bridge is an important design information document that contains the design conditions and the results of the structural review occurring in the design process, so it is a basic document for evaluating the safety of the structure. This document is also very useful for doing so. Therefore, Korea Facility Safety Technology Corporation has kept design book information under the 'Special Act on Safety Management of Facilities' (Ministry of Construction and Transportation, 1995), and recently, it has been using STEP and XML technologies to improve its utilization. Korea Institute of Facility Safety and Technology, 2004).

Recently, various programs related to bridge design automation such as AASHTOWare, ANSYS CivilFEM Bridge, ARoad, MIDAS, LUSAS Bridge have been developed and used in many areas. In this program, various design criteria referred to in the bridge design are embedded in the form of a database, which is used as a recommendation in determining the values of design variables or providing information to users when reviewing design criteria during the bridge design process. Sometimes.

However, such a review process is performed according to a pre-programmed procedure, and the review of a design previously prepared without a program is still performed by a manual method.

Since construction informatization generally aims to increase the efficiency of work by ensuring a smooth flow of information between each work process of a construction project, it is important to organize each item according to the standardized structural statement preparation guidelines.

However, at present, each company is constructing and using a checklist for structural calculation according to the characteristics of each company, and its items, formats, and criteria are different. Therefore, it is very difficult to find the missing items by comparing each item in the structural statement.

 Accordingly, in the present invention, by converting the structure calculation prepared by the user to XSD (XML Schema Definition), and performing a schema matching process (SCHEMA MATCHING PROCESS) with a predefined structure calculation standard information model, It aims to automatically find missing items.

To this end, first of all, the present invention proposes a conversion module for automatically converting a textual structure calculation written by a user into a document in XML or XSD format.

In addition, the present invention proposes a comparative analysis module for converting a structural calculation prepared by a user by the conversion module to an XML schema definition and extracting missing items of the structural calculation by using a schema matching technique based on the structural calculation standard data. .

In order to achieve the object as described above, the present invention is a finite string set having an order and is divided into finite lines, and the string set S ⁱ of the i th line is a text file defined by Equation (1). About structural statement of form,

Sequentially storing the string information for each row from the string set S ⁱ into heading symbols, titles, contents, and references in a temporary table; A hierarchical information assigning step of giving hierarchical information in which each heading is located in a tree structure of a document by using information on the header of the stored temporary table; An XML file generation step of generating an XML file in a depth-first order of a tree by using the hierarchical information and information stored in the temporary table; converting the document into an XML Schema Definition (XSD) by Wow;

By performing a schema matching process that compares the XSD data and the XML data of the standardized structure statement with each other, the present invention proposes a structure statement checking method including a missing item determination step of determining whether an item of the structure statement is missing.

The hierarchical information assigning step in the document converting step may include determining whether a head symbol exists for the title; And determining whether a string for a pure title among the titles matches the title.

The title string h ⁱ is preferably defined by Equations 2, 3a and 3b.

A set of predefined character string group IDs used for representing the title is referred to as an HS _ID , and when ∀ID and HS _ID ⊂ Σ +, the condition hs ⁱ ≠ ø is preferably defined by Equation 4.

A set of kinsoku characters that should not appear more than f times in the string S ⁱ is X _f , and when x _f ∈ X _f , the condition of hc ⁱ = ø is preferably defined by Equation 5.

B ₀ is the set of characters representing the left parenthesis, B ₀ ⊂HS _ID , B _c is the set of characters representing the right parenthesis, and <a, b> is the set of parentheses of the same type paired together. Defined as B ₀ , _b 정의 B _c , and the set of delimiters indicating the end of the title D _e = B _c , C _e , C _e is a set of characters used to distinguish the title from the content B _c ∩C _e = When defined as ø, l is preferably defined by equation (6).

The reference string r ⁱ is preferably defined by Equation 7,8.

A set that contains a string representing a predefined bibliographic name as an element is called RN and any set of strings trn ⁱ _ab = s _a s _{a + 1} ... s _b , trn ⁱ _ab ⊂ (S ⁱ ) *, 2≤ When a ≦ b ≦ n, rn ⁱ is preferably defined by equation (9).

일때 m+1은 수학식 10과 같이 정의되는 것이 바람직하다.A set of predefined reference start delimiters is RS, and when rn ⁱ ≠ ø, trs ⁱ _β = s _α-β , β = min (δ),

When m + 1 is preferably defined as in Equation 10.

When rn ⁱ ≠ ø, rp ⁱ = s _{b + 1} s _{b + 2} ... s _c , and when a set of delimiters representing the end of a predefined reference is RE, c is expressed in Equation 11 It is desirable to be defined by.

BS with the set of reference headers BS ^d = bs ^d ₁ , bs ^d ₂ , ... bs ^d _n , ..., bs ^d _n appears only once in the document, and appears on any line i in the document When the head symbol of the table of contents corresponding to bs _n ⁱ is n always increases as i increases, one BS must match a predetermined depth, and each depth is defined when several BSs are defined. are sequentially increased, and the number of groups defined by the head symbols BS ^d d, and wherein d _c is to indicate the layer a BS of the element occupies in the document from the document, layer d _i is the i-th table of contents share of the tree mathematics It is preferable that it is defined by Formula 12,13.

The present invention provides a method for converting unstructured text document information of a structural statement, which is a representative engineering document in the civil engineering field, into a semi-structured XML document in a tree form.

Structural calculation in the civil engineering field is a document that records the process of designing any facility, and in the case of bridges with long public periods, it is an important document that is continuously referred to during maintenance. Rather, it can be significant at the maintenance stage.

The structure of the document can be defined in various aspects, which can be largely divided into a structure revealed by the document itself and a structure according to the meaning of a text paragraph.

1 shows a conceptual diagram of the division of such a document structure.

Again, the structure of the document itself can be divided into a physical structure and a logical structure (Summers, 1998; Worring and Smeulders, 1999).

Physical structure refers to geometrical features such as the layout of text, drawings, tables, etc., the composition of pages, and the thickness of text in a document.The logical structure refers to the author, document title, summary, text, and references of a document. It refers to the structure of the detailed headings and contents within the text which the same structure and document author have distinguished.

Wang et al. (2005) defined only the structure of the document as a logical structure and semantic structure separately, and this was again defined as an explicit semantic structure and a latent semantic structure. structure).

Explicit semantic structure refers to the structure of detailed headings and contents in the above-mentioned text, and internal semantic structure is defined as the division of meaning contained in the paragraphs of the document.

The present invention focuses on the conversion to quasi-structured documents with explicit semantic structures.

The process of extracting an explicit semantic structure from an imaged document blocks each text paragraph in the image, stores the physical properties, and uses the stored physical properties and a predefined knowledge-base. To classify the hierarchical structure of the text (Summers, 1998; Altamura et al., 2001; Klink and Kieninger, 2001; Anjewierden, 2001).

Most researches on constructing semantic structures focus on extracting internal semantic structures, and two approaches are used.

One is a method of manually decomposing a document through the GUI and performing markups containing specific knowledge meanings by the user, as in the case of Adelberg (1998). The other is Salton et al. As in the case of (1997), it is a method of vectorizing the term frequency.

Wang et al. (2005) used a method of classifying hierarchies by using a head symbol used in a title and a predefined knowledge base in extracting an explicit semantic structure.

However, the methods used for the extraction of explicit semantic structures have some problems that cannot be directly applied to the structural statement targeted by the present invention. These problems are described below along with the characteristics of the document of the structural statement.

Like general engineering documents, structural statements include text and drawings as well as tables.

Extracting information from drawings and tables in a document, as shown by studies of Worring and Smeulders (1999) and drawings of documents in Embley et al. (2006) and Kawanaka et al. (2007). Research in this context requires additional methods from a different perspective than the study of document structure.

Therefore, in the present invention, only text content is used as an object for analyzing the structure of the document.

In addition, the present invention proposes a complementary method for overcoming the problems that may occur when the existing techniques target the structural statement, rather than a method of completely replacing the document structure extraction method using the characteristics of the aforementioned physical structure. Puts meaning on things.

In the present invention, since the preferred embodiment of the present invention was applied to the structural girder bridge, particularly among the various structural statements for each technical field, the following description will be given in detail.

In order to analyze the characteristics of the structural statement, the structural statement file for the superstructure of the steel girder bridge was collected in cooperation with a domestic engineering company. We analyzed 22 documents that were designed from six representative engineering firms, considering that different programs or forms may be used.

Based on the analysis results of the structural statement, which is actually produced and used by domestic engineering companies, the structural characteristics of the structural statement are summarized as follows.

The body of the structural statement can be divided into titles and contents as in general documents.

In the document, the title is used to convey to the reader what category the content appears under that title.

The logical structure of the headings defined in the document can be represented in a tree.

A tree is a kind of graph that is a directed graph with one vertex called root, which does not have an entry path, and the level in the tree refers to the number of paths from root to the node.

Generally, in long documents, such as structural statements, the author uses some symbol in the title to convey to the reader the hierarchy in which the title is located in the document's tree structure.

Therefore, in order to extract the logical structure, the symbols used to identify the contents corresponding to the title of the table of contents may be used as the most important clues.

However, the existing methods described above have three major problems to be applied to the structural statement of the present invention.

The first problem is that identical headings within a document are considered to represent the same hierarchy in explicit semantic structures (Wang et al., 2005).

This assumption can be used for most general documents, but in the case of structural statements, the symbols used within the same document often mean different hierarchies. Cannot be seen as meaning a particular hierarchy.

The second problem is that most of the knowledge base and machine learning methods used to process imaged documents are focused on the characteristics of the document format.

However, in most engineering practices, there is no standard format for formal document format for writing structural statements, so there is a limit to using the format of document produced by various companies to extract explicit semantic structures.

Finally, in the structural statement, many items that can be regarded as one of the explicit logical structures are often displayed on one line with the contents without discernment of the physical format.

In this case, analyzing the document based on the rules that make up the string rather than processing the document's format may be more appropriate for applying the structural statement to the target.

It has the feature of supporting the markup by user definition and the independence of the platform. Therefore, it is still widely used for not only document information but also heterogeneous information exchange in various fields.

The elements used for markup in an XML document consist of the element's name, the attribute assigned to the element, the attribute value, and the element's content. It may consist of text (Bray et al., 2006).

Accordingly, the elements appearing below are divided into child elements of the above, and thus the contents of the document can be expressed in a tree form.

In Korea, as part of the construction CALS / EC project (Ministry of Construction and Transportation, 2003) to facilitate the sharing and exchange of information in the construction industry, the Construction CALS / EC Electronic Document Pool (Korea Institute of Construction Technology, 2004) was established as a group standard.

Here is an XML schema for 220 document structures, including structure statements.

In the structural statement of the present invention, there are markups for document management such as the name of the ordering organization, the name of the document authoring agency, and the date of submission. There is a markup for.

However, in the given chapter and demarcation markup method, there is a drawback that the depth of the document tree can be expressed only limitedly.

Therefore, the present invention focused on expressing all hierarchies in which the table of contents can appear.

When the same design method is applied in the same bridge type, the structural statement shows the pattern almost identical.

In the present invention, a preferred embodiment of the present invention will be described in detail using a steel girder bridge structural calculation example.

Figure 2 shows a part of the title of the structure of the girder bridge structure as an example.

As shown in FIG. 2, the titles described in the Gurder bridge structural statement viewed through the collected documents can be broadly classified into three types: "action name", "part name", and "variable name".

"Action Name" refers to the detailed actions involved in performing structural calculations, including "slab design", "mould design", "joint design", and "section review", mainly with "site name" Used or appear repeatedly under "Site Name".

The term "part name" refers to elements or spaces or types that physically constitute the structure, and refers to elements such as "cantilever part", "first interstitial center part" and "splice-1" which are repeated at least once. appear.

Finally, "variable name" is often used to describe a given condition in performing an activity, or to describe the final result after the execution of the activity, and "bridge specifications", "used materials", "fixed load", "live load" And the like, which appear repeatedly below the “site name” when it corresponds to a site of the same type.

Such a phenomenon was common in all cases of steel bridges targeted in the present invention.

Therefore, it is judged that the table of contents used in the structure of the girder bridge can be used to give meaning to detailed text information.

That is, even if the length of the same word is used to distinguish the length of the bridge or the length of the girder, the table of contents information itself in the structural statement is used to determine what actions or specific parts belong to the specific text contents. Can provide clues to the search.

3 illustrates a process of generating a structured XML document for a text document.

In the present invention, a document stored as a text file is used as an input file in a document preparation program used in practice to exclude errors occurring in recognizing characters from an imaged document.

As shown in FIG. 3, the input text file is converted into a structured XML document in three steps.

Firstly, according to the document model extracted from the collected girder bridge structural statement, the string information is stored in the temporary table in the temporary table, divided into header symbols, titles, contents, and references.

Using the information about the header of the stored temporary table, information about the hierarchies of each title in the document tree structure is given. Finally, an XML file is generated using the hierarchical information and the information stored in the temporary table.

The definition of the model and hierarchy rules of the document used in such a process will be described later.

Hereinafter, a document model defined for parsing text document information of a steel bridge structure statement will be described.

First, some expressions are defined as follows (Lopresti, 2000; Linz, 2005).

1) A set of strings S = s ₁ s ₂ ... s _n means a set of characters with a finite order, where s _k ∈∑ 1≤k≤n and ∑ is all including spaces covered in a given document Is a set of characters.

2) A superscripted * after a character set name means a powerset of the character set (e.g., ∑ ^* is the powerset of ∑)

3) The superscripted + after a character set name means a set excluding the empty character λ from the set of characters.

4) n _sk (S) means the number of characters s _k included in the string set S.

5) The symbol ":: =" in A :: = B means that the former (A) is represented by the latter (B).

6) The symbol "|" means "or".

7) The symbol "| S |" means the number of characters included in the string S.

The characteristics of the text document information of the structural statement through the analysis of the collected steel bridge structural statements are as follows.

The components of the textual information in the structure of the girder bridge can be defined as in definition 1 below.

Definition 1. (components of text information) structure, the statement text document is set as a finite set of strings with string sequence are separated by a finite one row (line), a set of strings S ⁱ i-th row is composed as follows: .

Where h ⁱ is the set of strings for the title h ⁱ = s ₁ s ₂ ... s _l , c ⁱ is the set of strings for the content c ⁱ = s _{l + 1} s _{l + 2} ... s _m , r ⁱ is a set of strings for the reference r ⁱ = s _{m + 1} s _{m + 2} ... s _n , where 0≤l≤m≤n.

As shown in definition 1, the textual information consists largely of title, content, and reference, and there are five cases in which they appear, if each line consists only of title or content, and title and content and reference appear simultaneously, It can be divided into a case consisting of and a reference, and finally a case consisting of a content and a reference.

There are many forms of arrays for the content of a document, and finding the rules for those arrays is a very difficult task.

Therefore, a rule that can identify a title and a reference in the input text line to identify these components can be defined and classified as a content if the string does not satisfy the rule.

The rules for identifying titles and references are described in Definitions 2 and 3, respectively, as follows.

The string for the title defined in definition 1 is again constructed as in definition 2 below.

Definition 2. (components of headings) The title string h ⁱ defined in Definition 1 consists of:

Where hs ⁱ is the set of strings used to represent the title hs ⁱ = s ₁ s ₂ ... s _o , hs ⁱ ⊂ ∑ +, hc ⁱ is the set of strings for the pure title hc ⁱ = s _{o +1} s _{o + 2} ... s _p , hd ⁱ are delimiter symbols that indicate the end of hc ⁱ , meaning hd ⁱ = s _l , and the relationship between o, p, and l is as follows:

As defined in the definition 2, the necessary and sufficient condition for being subject string h ⁱ ≠ ø was set to ^{hs i ≠ ø ∧ hc i ≠} ø. Therefore, if hs ⁱ = ø or hc ⁱ = ø in the string given by Demorgan's law, it can be regarded as h ⁱ = ø, which can be used to determine the existence of the title string.

The condition hs ⁱ ≠ ø and the condition hc ⁱ = ø were defined as the following definitions 2a and 2b, respectively.

Definition 2a. (heading symbols) A set of predefined string group IDs used for representing titles is referred to as HS _ID , and when ∀ID and HS _ID ⊂ ∑ ⁺ , hs ⁱ ≠ ø is as follows.

Definition 2b. (heading contents) A set of kinsoku characters that must not appear more than f times in a string is X _f , and when x _f ∈ X _f , the condition hc ⁱ = ø is as follows.

Definitions 2a and 2b are respectively defined to determine whether a heading for a title exists and whether a string for a pure title matches the title.

If a character string similar to the pattern of the head symbol is placed at the beginning such as "1.2" or "2.3", the existence of the head symbol alone cannot determine the title.

Definition 2b is a condition for this case. If a decimal point is placed at the beginning of a string in the structure statement, the line is followed by arithmetic expressions.

Tables 1 and 2 show the set of predefined head symbols and the set of lexical characters for the five applications through the analysis of the collected documents, respectively.

division ID HS _ID Number Group 1 One 1 2 3 4 5 6 7 8 9 10, … Number Group 2 2 1.1, 1.2,... , 2.1, 2.2,... , 3.1, 3.2,... Number Group 3 3 1.1.1, 1.1.2,... , 1.2.1, 1.2.2,... , 2.1.1, 2.1.2,... Number Group 4 4 1.1.1.1, 1.1.1.2,... , 1.1.2.1, 1.1.2.2,... , 1.2.1.1, 1.2.1.2,... Number Group 5 5 1-1, 1-2,... , 2-1, 2-2,... , 3-1, 3-2,... Number Group 6 6 1-1-1, 1-1-2,... , 1-2-1, 1-2-2,... , 2-1-1, 2-1-2,... Number Group 7 7 1-1-1-1, 1-1-1-2,... , 1-1-2-1, 1-1-2-2,... , 2-1-1-1,... Number Group 8 8 I, II, III, IV, V, VI, V, V, V, V,. Number Group 9 9 Ⅰ, ii, ⅲ, ⅳ, ⅴ, ⅵ, ⅶ, ⅷ, ⅸ, ⅹ,… English Symbol Group 1 10 A, B, C, D,... , Z English Symbol Group 2 11 a, b, c, d,... , z Hangul Symbol Group 1 12 Go, me, da, la,… , Ha Hangul Symbol Group 2 13 A, b, c, d,. , ㅎ Original Character 1 14 ①, ②, ③, ④, ⑤, ⑥, ⑦, ⑧, ⑨, ⑩, ⑪, ⑫, ⑬, ⑭, ⑮,… Original Character 2 15 Ⓐ, ⓑ, ⓒ, ⓓ,… , Ⓩ Original Character 3 16 Ⓐ, ⓑ, ⓒ, ⓓ,… , Ⓩ Original Character 4 17 ㉮, ㉯, ㉰, ㉱,… , ㉻ Original Character 5 18 ㉠, ㉡, ㉢, ㉣,… , ㉭ Parentheses Group 1 19-31 Groups with “)” at the end of strings of numeric groups, English and Korean symbol groups
Example: HS ₁₉ = (1), 2), 3),... )}, HS ₂₉ = {i), ii), iii),... )} Parentheses Group 2 32-44 Groups with "(", ")" appended to both strings of numeric groups, English and Korean symbol groups
Example: HS ₁₉ = {(1), (2), (3),... }, HS ₂₉ = {(a), (b), (c),... , (z)} Parentheses Group 3 45 ~ 47 Open parenthesis symbols {[}, {(}, {<} Other groups 48-59 {-}, {*}, {◁}, {◀}, {▷}, {▶}, {⊙}, {◈}, {◎}, {·}, {○}, {"},

f X _f One {+, *, ×, ÷, =, ≤, ≥, ∑,!, ±,%,?,.} 2 {[,],,, <,>, /} 3 {(,),-}

According to definition 2a and definition 2b, the string to be analyzed should be extracted to determine the existence of the title string.

As defined in definition 1, when the title string and the string related to the content or reference coexist, the area of the temporary string that can be estimated as the title string can be found using the definition 2c below.

Definition 2c. (location of heading end) B0 is the set of characters representing the left parenthesis, B ₀ ⊂HS _ID , B _c is the set of characters representing the right parenthesis, and <a, b> are in the same kind of parentheses and for the set defined as a∈B _0, b∈B _c, a set of separators, which indicates the end of the title _{D e = {B c, c} e}, c e is the set of characters used to distinguish the subject and content If hs ⁱ ≠ ø hc ⁱ ≠ ø, defined as B _c ∩C _e = ø, in definition 2, l is as follows.

In the girder bridge structural statement, the references represent the basis for the process of designing the structure. When references are entered in the collected structural statement, they are located at the end of each line, with the specific page number or section number of the reference.

The composition of the string for the reference defined in definition 1 and the definition for extracting each component are as follows.

Definition 3. (component of reference) The reference string ri defined in Definition 1 is composed as follows.

Where rs ⁱ is the delimiter used to distinguish the bibliography of the bibliography, rn ⁱ is the set of strings representing the name of the bibliography, re ⁱ is the set of end point symbols of the bibliographic information, and rp ⁱ is the bibliography of the bibliography It is a set of character strings indicating additional descriptions like page numbers, and the condition for r ⁱ ≠ ø is as follows.

Definition 3a. (reference name) A set of elements with a string representing a predefined reference name is called RN, and a random set of strings trn ⁱ _ab = s _a s _{a + 1} ... s _b and trn ⁱ _ab ⊂ (S ⁱ ) Rn ⁱ is as follows when ^* , 2≤a≤b≤n.

Depending on the author of the document, the name of the bibliography may be abbreviated.

Therefore, the set RN of the string with the bibliographic name predefined in definition 3a as an element must contain the string for this abbreviation as an element.

In definition 3a, trn ⁱ is defined as a string containing spaces (λ) in the given string S ⁱ to prevent errors in bibliographic identification due to errors in spacing by the document author. Such a problem may occur in a document intended for Korean.

Therefore, in the process of reviewing trn ⁱ ∈RN, it is more consistent to replace trn ⁱ with a string without spaces and compare it with each element of RN. Should be

In the case of English where spaces are written for each word, such a processing method is meaningless.

이다. 만약 rnⁱ≠ø이고 참고문헌정보의 시작위치가 m+1이라고 하면 m+1은 다음과 같다.Definition 3b. (start location of reference) When RS sets a predefined reference start delimiter, trs ⁱ _β = s _α-β , β = min (δ),

to be. If rn ⁱ ≠ ø and the starting position of bibliographic information is m + 1, m + 1 is

Definition 3b is defined to identify the delimiters used before the bibliography to prevent unnecessary characters from being incorporated into the text content.

In the collected documents, the delimiter for the beginning of the bibliography is mainly used ?? [??, ?? (??, ??-??, so RS = {[, (,-} is also referred to in the application described later). It was used as a set for recognizing the starting delimiter in the literature.

Definition 3c. (reference page) If rn ⁱ ≠ ø, then rp ⁱ = s _{b + 1} s _{b + 2} ... s _c , and when a set of delimiters representing the end of a predefined reference is RE, c is Is the same as

Definition 3c is a definition for extracting a string meaning a specific page or section of a reference.

When r ⁱ ≠ ø in the string S ⁱ given by definition 1, when the last character sn is a delimiter indicating the end of the bibliographic string, the string corresponding to the reference page is as defined in definition 3c, and in the present invention, RE = {],) ,.} was used.

4 illustrates a process of parsing an input text file using the above-described definitions.

As shown in FIG. 4, the text file is read line by line, and the title and the reference are identified using the definition 2 and the definition 3 in the string of each line.

The identified results are used to estimate the body content by definition 1, and the finally identified results are stored in a temporary table.

In a real text file, there may be lines that consist only of empty strings.

Therefore, in the case of the actual program implementation, in the case of a line consisting of empty strings only, the processing below is skipped and the next line is read again and processed only when valid strings exist so that the information stored in the temporary table is stored in a valid string. It can be used efficiently for labeling layers.

As mentioned earlier, the document uses a head symbol before the table of contents to estimate the hierarchy of each table of contents.

However, different authors or companies have different usage patterns for headings that can be used to estimate the hierarchical location of a table of contents, so defining a group of headings as a single hierarchy estimates the hierarchy of tables of contents. It depends on the pattern.

Therefore, in order to estimate the hierarchy of the table of contents in a more generalized method, the present invention proposes a method of assigning the hierarchy of the table of contents with the relative difference between the tables.

In this way, the rules for assigning levels to the rows recognized as the table of contents are defined as follows.

Definition 4. (order of subtitles) The order in which the table of contents appears in a text document corresponds to the depth-first order of the tree.

When sorting the order of each node in the tree in a linear order, there are two methods, as shown in FIG. 5, a depth first method and a breadth-first method.

5 is a conceptual diagram illustrating a sorting order in a depth first method and a width first method.

In general, the order in which each heading appears in the document is the same as the depth-first method.

Therefore, if the contents of the document are described in the correct order according to definition 4, the problem of converting the structure of the table of contents into a tree form while reading text information sequentially is connected to the problem of estimating the depth of the table of contents in the tree structure. .

To identify the difference in relative depth between headings, we first need a reference to the headings that are the reference and the depth they occupy in the tree.

In the simplest case, you can refer to a group of headers first recognized in the document, or you can define your own specific headers and their depths that do not change the hierarchy in the document.

However, the head symbol used in the table of contents should have the following characteristics.

Definition 5. If (base-symbol group) d = 1,2, ... the base set of symbols BS ^d = bs ^d ₁ , bs ^d ₂ , ... bs ^d _n , ... Bs ^d _n appears only once in the document, and _n always increases as i increases, given that bs _n ⁱ is the heading of the table of contents corresponding to the BS in any line i in the document.

Definition 5a. (depth of base-symbol group) One BS must match a defined depth, and if multiple BSs are defined, each depth must be increased sequentially.

If you look for a group of headings in Table 1 that can be used as a set of base headings according to definition 5 above, the IDs range from 1 to 44.

However, in ordinary documents, including structural statements, the condition that IDs 1 through 8 appear only once in the document.

Once the defined group of reference headings is defined, definition 4 can be used to define the hierarchy for titles that appear sequentially in the document, relative to the previously defined hierarchy.

The rules for this are as follows:

Definition 6. (depth of headings) The group defined by the set of reference headings is BS ^d , where D _c represents the hierarchy D _i occupies in the tree when the element of BS in the document represents the hierarchy occupied by the document. Is as follows.

이고 만약 K=ø이면 k=0이며, 가상노드의 필요성을 검토하는 E(i,j,k)는 수학식 13과 같다.Where g (hs ⁱ ) is a function to convert the group ID according to Table 1 when entering the head symbol hs ⁱ , where j = i-1 means the previous title, and k = max (K)

If k = ø k = 0, E (i, j, k) to examine the necessity of a virtual node is represented by the equation (13).

는 수학식 14와 같다.Where md means the group ID of the highest order level primary group at the predefined order level,

Is the same as Equation 14.

, BE는 순서를 가지는 머리기호 그룹들에서 제일 처음에 나타나는 머리기호의 집합을 의미한다.Where e = max (L) ,

, BE means the first set of headings in the ordered group of headings.

6 illustrates an example in which each title is converted into a tree structure of a document by using Equation 12.

As shown in FIG. 6, when D _{j + 2} is represented in Equation 12, another header symbol appears between layers as reference.

In this case, in this case, the hierarchy of the given line is increased by increasing the hierarchy of the corresponding line by +2 of the previous hierarchy and newly creating a virtual node with 'parent name_add' when converting the document to the actual tree structure. It could be maintained.

According to the method described above, the process of classifying the titles of the temporary table for storing the parsed text information as to which layer of the tree structure is allocated is shown in FIG. 7.

8 to 10 is a screen capture of an embodiment of the present invention, Figure 8 is an example of an input structure statement text document (state stored as a text document in the Excel program), Figure 9 is an example of the document conversion module operation, 10 illustrates an example of an XML document converted through a module.

So far, the document conversion module has been described in detail. Hereinafter, the missing item determination step of determining whether the XML data converted by the document conversion module is a design missing item will be described in detail.

Before performing this step, it is natural that XML data of the standardized structural statement should be defined in order to determine whether the structural statement item prepared by the user is corrected.

The schema matching technique used to extract the design review missing items is a two-step process. The first process is Tversky and Shafir (2004), and Yi et al. (2005) proposed by Lee et al. The similarity measure method modified by (2006) was used. Equation 15 shows the similarity in the set of element names, and Equation 16 is the final similarity in which the element items of the parent, sibling, and child are added together.

In the equation, i, j represents an element to be compared and matched, N represents a set of element definitions, and S represents a value representing matching similarity.

here,

는 각각 그 요소, 부모 요소, 형제 요소, 자식 요소의 가중치이며,

이다.

의 의미는 각각의 집합에 공통으로 포함되어 있는 최소화 단어의 교집합을 의미하며, N, P, B, C는 각각 자신의 항목, 부모의 항목, 형제의 항목, 자식의 항목을 의미한다.Equation (16) is the final similarity of the sum of the element items of its own element and parent, sibling, and child. here,

Are the weights of the element, parent element, sibling element, and child element, respectively,

to be.

Means the intersection of the minimum words commonly included in each set, and N, P, B, and C mean their own items, their parent items, their siblings, and their children.

And as a second process, Yi et al. (2005) and Relaxation labeling method modified by Bong-Keun Kim et al. (2006) were used.

는 계산단계 t에서 임시 표준의 i번째 항목을 수집한 문서의 k번째 항목에 매칭시켰을 때의 신뢰도이며,

는 지지함수이다.Equation 17 shows a relaxation labeling method, wherein

Is the reliability of matching the i th item of the temporary standard to the k th item of the document collected in the calculation step t,

Is the support function.

Equation 17 may be expressed as Equation 18 below.

은 아래 수학식 19과 같다.Where m is the number of elements in the temporary standard document and n is the number of elements in the collected document. And,

Is expressed by Equation 19 below.

Here, e is defined as a pair of two elements such as e (i, j) as an edge, and if element i and j are directly connected to each other, e (i, j) = 1, otherwise E (i, j) = 0. In addition, e ~ (i, j) means a path. If elements between any element i and j are consecutively defined, e ~ (i, j) = 1, otherwise e ~ (i , j) = 0, and d _ij means the number of edges existing between elements i and j when e ~ (i, j) = 1.

In this manner, the reliability matrix P ^(t) generated after schema matching is expressed by Equation 20. These values will be used to extract the missing items from the predefined structural statement standard information model and the actual structural statement.

FIG. 11 is a flowchart conceptually illustrating technical features of the present invention, in which a portion indicated by a dotted box is a document conversion for converting a structural calculation written by a user into an XSD file (Source) having a hierarchy defined for XML schema matching. It shows the module, and shows the comparison analysis module that identifies the missing items by extracting the missing items of the structural statement by the schema matching module between the XSD file (Source) and the standard schema data (the best model defined by the standard).

12 is a flow chart of a comparative analysis module for determining missing items for design review of a structural statement, which will be briefly described as follows.

First of all, the structural statement standard information model is defined and converted into XSD, the target value is set (defined as standard value), and the structural statement is made into the XSD file using the document conversion module described above and used as the source file. After measuring the similarity between the elements by comparing each other, the total similarity values are arranged by adding values related to structural relationships through relaxation labeling. Thus, if the result value is lower than the reference value, the value is determined not to match the target and is treated as a missing item.

FIG. 13 shows a matching diagram showing matching with a standard schema arbitrarily set for an item of “reinforcement design” portion of a source structural statement, and Table 3 shows similarity measurement of two items of FIG. 13. The final result is shown after the measure process and relaxation labeling process. In Table 3, the vertical items are the XML schema items of the predefined standard structure statement, and the horizontal items are the items of the actual structure statement.

Source

Target Design of Stiffeners Web reinforcement Vertical Reinforcement Design Horizontal Reinforcement Design Branch reinforcement Flange reinforcement Longitudinal rib design Design of Transverse Ribs Jack Up Reinforcement Design of Stiffeners One 0 0 0 0 0 0 0 0 Flange reinforcement 0 0 0 0 0 One 0 0 0 Longitudinal Rib Design (Point Section) 0 0 0 0 0 0 One 0 0 Transverse Rib Design (Point Section) 0 0 0 0 0 0 0 One 0 Web reinforcement 0 One 0 0 0 0 0 0 0 Horizontal reinforcement design (branch) 0 0 0 0 0 0 0 0 0 Longitudinal rib design (general part) 0 0 0 0 0 0 0 0 0 Lateral rib design (general part) 0 0 0 0 0 0 0 0 0 Design of Horizontal Reinforcement (General) 0 0 0 0 0 0 0 0 0 Vertical Reinforcement Design 0 0 One 0 0 0 0 0 0 Horizontal Reinforcement Design 0 0 0 One 0 0 0 0 0 Branch reinforcement 0 0 0 0 One 0 0 0 0 Just point 0 0 0 0 0 0 0 0 0 Middle point 0 0 0 0 0 0 0 0 0 Jack Up Reinforcement 0 0 0 0 0 0 0 0 One Just point 0 0 0 0 0 0 0 0 0 Middle point 0 0 0 0 0 0 0 0 0

The order in which each item of the XML schema is listed in a table is listed descending from the top element to its child elements as shown in FIG. That is, after listing all the items for a child element for an element, it goes to the sibling element, and is listed in the table in the order of the child elements of the sibling element.

In Figure 13, items such as left and right may be expected to match at the same position because the letters of the items are the same, but in the actual structural statement, the design of the longitudinal ribs and the design of the horizontal ribs are temporary standard schemas. Is included in the 'flange reinforcement' part and the 'web reinforcement' part, so that the matching of the letters is difficult. However, as can be seen in Table 3, the letters are matched exactly in one place, because the structural position is calculated through the relaxation labeling described above.

Therefore, the design of the longitudinal ribs and the design of the longitudinal ribs, which are sub-items of the flange reinforcement in the actual structural statement, are the design general sections of the longitudinal ribs and the yellow, which are sub-items of the web reinforcement in the standard schema. It is not matched with the directional rib design general part, but is matched with the design point part of the longitudinal rib and the lateral rib design point part, which are sub-items of the flange reinforcement. In this way, in Table 3, the design of the horizontal reinforcement (branch), the design of the longitudinal ribs (general part), the design of the horizontal ribs (general part), and the design of the horizontal reinforcement of the temporary standard schema are shown in Table 3. (Normal part) ',' End point part ',' Middle point part ',' End point part ',' Middle point part 'have no matching values.

From the matching result, the items without the matching value may be regarded as items that do not exist in the source structure statement, and thus are determined as missing items.

FIG. 15 is a diagram illustrating an embodiment of a screen showing hierarchical visualization and schema matching results of an XML file converted through an upload process, an XML conversion process, and the like of a structure statement by a program to which the method according to the present invention is applied.

Since the above has been described only with respect to some of the preferred embodiments that can be implemented by the present invention, the scope of the present invention, as is well known, should not be construed as limited to the above embodiments, the present invention described above It will be said that both the technical idea and the technical idea which together with the base are included in the scope of the present invention.

1 to 10 are views for explaining an embodiment of a document conversion step in the present invention,

1 is a conceptual diagram for dividing a document structure.

2 is a diagram showing a part of the title of structural structure of girder bridge,

3 is a block diagram illustrating a process of generating a structured XML document for a text document.

4 is a block diagram illustrating a process of parsing an input text file using definitions.

5 is a conceptual diagram for the order of sorting in the depth first and width first methods,

6 is a conceptual diagram illustrating an example in which each title is converted into a tree structure of a document by using Equation 12.

7 is a block diagram of a process of classifying titles of a temporary table storing parsed text information as to which layer of a tree structure is allocated.

8 to 10 are captured screens according to an embodiment of the present invention, Figure 8 is a structural statement text document screen (state stored as a text document in the Excel program), Figure 9 is a document conversion module operation screen, 10 is an XML document screen converted through a module.

11 to 14 are explanatory diagrams illustrating a schema matching process for comparing source XML schema data and target schema data in accordance with the present invention.

11 is a flowchart conceptually illustrating the overall flow of the technical features of the present invention;

12 is a simplified view showing a comparison analysis module for explaining the step of identifying missing items in the present invention,

FIG. 13 is an explanatory diagram exemplarily illustrating a schema matching diagram for reinforcement design items of a temporary standard schema (left) and an actual structural statement (right) as an embodiment of the present invention;

14 illustrates a display order of elements of an XML schema.

15 is a view showing an embodiment of a screen showing a structure statement upload, XML transformation and hierarchical visualization, and schema matching result by a program to which the method according to the present invention is applied.

Claims (6)

A finite set of strings that are ordered and separated by a finite line, and the string set S ⁱ of the i th line is a structured statement in a text file format defined by Equation 1. Sequentially storing, from the string set S ⁱ , string information for each row into a temporary table by dividing the string information for each row into a head symbol, a title, a content, and a reference; A hierarchical information assigning step of giving hierarchical information in which each heading is located in a tree structure of a document by using information on the header of the stored temporary table; An XML file generation step of generating an XML file in a depth-first order of a tree by using the hierarchical information and information stored in the temporary table; converting the document into an XML Schema Definition (XSD) by Wow; And a missing item determination step of determining whether an item of the structural statement is missing by performing a schema matching process for comparing the XSD data and the XML data of the standardized structural statement with each other. [Equation 1]

Where h ⁱ is the set of strings for the title h ⁱ = s ₁ s ₂ ... s _l , c ⁱ is the set of strings for the content c ⁱ = s _{l + 1} s _{l + 2} ... s _m , r ⁱ is a set of strings for the reference r ⁱ = s _{m + 1} s _{m + 2} ... s _n , where 0≤l≤m≤n.) Missing item determination step in claim 1, A similarity calculation step according to Equations 16 and 15; The total similarity of arranging the total similarity values by adding structural relationship values calculated by a relaxation labeling process defined by Equation 17 to the similarity values between the schema elements and the elements calculated by the similarity calculating step. An array of values; And determining that the target value of the item is not matched when the result value for the specific item is lower than the reference value. [Equation 16]

(here,

Are the weights of the element, parent element, sibling element, and child element, respectively,

to be.

Means the intersection of the minimum words commonly included in each set, and N, P, B, and C represent their own items, their parent items, their siblings, and their children.) [Equation 15]

(here,

) [Equation 17]

(here,

Is the reliability of matching the i th item of the temporary standard to the k th item of the document collected in the calculation step t,

Is the support function.) In the document conversion step of claim 1, The title string h ⁱ is a structural statement check method, characterized in that defined by the equation (2, 3a, 3b). [Equation 2]

Where hs ⁱ is the set of strings used to denote the title hs ⁱ = s ₁ s ₂ ... s _o , hs ⁱ ⊂ ∑ ⁺ , hc ⁱ is the set of strings for the pure title hc ⁱ = s _{o + 1} s _{o + 2} ... s _p , hd ⁱ are delimiter symbols that indicate the end of hc ⁱ , meaning hd ⁱ = s _l , and the relationship between o, p, and l is represented by equations 3a, 3b Is the same.) Equation 3a

[Equation 3b]

In the document conversion step of claim 3, The reference set of symbols BS ^d is in sequence bs ^d ₁ , bs ^d ₂ , ... bs ^d _n , ..., where bs ^d _n appears only once in the document and appears in the BS appearing on any line i in the document. When the heading of the corresponding table of contents is bs _n ⁱ , n always increases as i increases, and one BS must match a predetermined depth, and if multiple BSs are defined, each depth is sequentially increased, and the number of groups defined by the head symbols BS ^d d, and wherein d _c is to indicate the layer a BS of the element occupies in the document from the document, layer d _i is the i-th table of contents share of the tree formula Method of inspection of structural statement, as defined by 12,13,14. [Equation 12]

(Where g (hs ⁱ ) is a function that converts the group ID according to Table 1 when entering the head symbol hs ⁱ , where j = i-1 means the previous title, and k = max (K)

If K = ø k = 0, E (i, j, k) to examine the necessity of a virtual node is represented by Equation 13. [Equation 13]

(Where md means the group ID of the highest order level primary group at the predefined order level,

Is the same as Equation 14.) [Equation 14]

(Where e = max (L) ,

, BE refers to the first set of headings in an ordered group of headings.) In the document conversion step of claim 4, A set of predefined string group IDs used to represent the title is referred to as HSID, and when ∀ID and HSID ⊂ ∑ +, the condition of hsi = ø is defined by Equation 4. Way. &Quot; (4) &quot;

In the document conversion step of claim 5, A set of kinsoku characters that must not appear more than f times in the string set is X _f , and when x _{f f} X _f , the condition of hc ⁱ = ø is defined by Equation 5, B ₀ is the set of characters representing the left parenthesis, B ₀ ⊂HS _ID , B _c is the set of characters representing the right parenthesis, and <a, b> is the set of parentheses of the same type paired together. Defined as B ₀ , _b 정의 B _c , the set of delimiters that denote the end of the title D _e = {B _c , C _e} , C _e is a set of characters used to distinguish the title from the content B _c ∩C When defined as _e = ø, l is defined by Equation 6. The reference string r ⁱ in Equation 6 is defined by Equation 7. [Equation 5]

&Quot; (6) &quot;

[Equation 7]

Where rs ⁱ is the delimiter used to distinguish the bibliography of the bibliography, rn ⁱ is the set of strings representing the name of the bibliography, re ⁱ is the set of endpoints for bibliographic information, and rp ⁱ is the bibliography A set of strings that represent additional descriptions, such as the page number of, and the condition for r ⁱ ≠ ø is

Is the same.)

Images