KR20070090321A - System for extracting information from document, method thereof and recording medium thereof - Google Patents

Abstract

A system and a method for extracting information from a document, and a recording medium storing the same are provided quickly to offer data to a user by constructing a database with the information obtained from recognition of patent documents, and generate a new drawing by combining a drawing and drawing information extracted from the patent document. A recognizer(71) recognizes a code from a drawing section comprising an explanation part, which is a text part, and a drawing part, which is an image part. An extractor(72) extracts the information related to the code from the explanation part. The database(73) stores the patent documents and the information extracted by the extractor. The information includes at least one of the number and order of drawings forming the drawing part, the codes, and code names. The extractor extracts the information by using location information of text blocks, which are detail explanation of each drawing, of each drawing.

Description

System for Extracting Information from Document, Method Thereof And Recording Medium Thereof}

1 is a view briefly showing the structure of a patent document according to an embodiment of the present invention,

2 is a block diagram showing a document information extraction system according to an embodiment of the present invention;

3 is a flowchart illustrating a document information extraction method according to an embodiment of the present invention.

Explanation of symbols on the main parts of the drawings

10, 30, 50: description 20, 40, 60: drawing

70: information extraction system 71: recognition unit

72: extractor 73: database

151, 357, 560: Drawing explanation part 156, 355, 570: Configuration description part

The present invention relates to a document information extraction system, a method of extracting the same, and a recording medium recording the same. In particular, a document information extraction system for extracting drawing information from a patent document and generating a new drawing combining drawing and drawing information, and a method of extracting the same It relates to a recording medium recording this.

In general, document recognition is a technology that reads and analyzes printed matter divided into a picture area and a text area in the form of an image, and then converts the characters of the text area into an editable text form in a text editor. It appeared in order to automatically input the user's desired part in the pre-printed printed matter through a computer, and it is a detailed field of Character Recognition.

Character recognition is a field of pattern recognition that recognizes a character through visual information and understands the meaning of the character, and has been put to practical use in the fields of optical character recognition (OCR) and drawing recognition. Recently, it has been integrated with Neural Network, Fuzzy Theory, Genetic Algorithm, Natural Language Processing, Psychology and Cognitive Science in the field of Artificial Intelligence. It is developing anew. In addition, OCR, which is a representative example of text recognition, has been put into practical use in various fields such as automatic mail sorting, industrial product inspection or classification, drawing recognition, text recognition of faxes, and automatic entry of checks.

One example of such a technique for editing documents and images is disclosed in US Pat. No. 56,3681 (a method and apparatus for synchronizing, displaying and manipulating registration, text and image documents on April 22, 1997).

The technique disclosed in US Pat. No. 56,3681 is an apparatus for generating an equivalent text file of a document from a source text file and a source image file, the source text file and the source from at least one storage medium. Extraction means for extracting an image file, page dividing means for page dividing a source text file by using the source image file of the document to generate an equivalent text file, a display device for displaying text and images, and a CPU (Central) Processing unit), storage means for storing at least one text document coupled to the CPU and composed of an equivalent text file and at least one image document composed of the stored image file, for displaying on the display device It consists of a user interface (Interface), etc. generated by the CPU, Source text file and source through a step of extracting the source text file and the source image file from a storage medium and by dividing the source text file using the source image file of the document to generate an equivalent text file. It is described that an equivalent text file of a document is generated from an image file.

In addition, an example of a technology for providing information consisting of text and pictures is disclosed in Korean Patent No. 0495034 (registered on June 02, 2005, an information providing system and method using an info box).

The technology disclosed in Korean Patent No. 0495034 includes a specialized DB system that periodically collects data for specialized DBs and collects predetermined contents in a contents master DB, and includes at least one client. (Client) An information providing method using an information box that is coupled through a terminal and a network (Network) network and an info box (Infobox) performed by a specialized DB system, the information box processing module of the information providing server ( The module generates an extraction variable file for each specialized DB in a rule-based manner, extracts a keyword using the extraction variable file, info box information of the content master DB created through the keyword extraction step (here, info Box information includes at least a search keyword and detailed information). It is described that users can easily check the text or image related information by simply hovering the mouse over a predetermined text or picture activated while the user reads content such as a news article.

However, in the prior art, including the technologies disclosed in the above publications, there is a problem in that the service production cost increases due to the purchase of the OCR product.

In addition, in the case of a service providing document information combining images and texts in real time, each time a user requests the document to extract the information requested by the user, the service is required to extract and provide the information. There was a problem that can not provide a quick service to the user in need.

In addition, when a large number of users request the service at the same time when providing the real-time service as described above, there was a problem that the service is delayed or stopped due to an overload of the system.

SUMMARY OF THE INVENTION An object of the present invention is to solve the problems described above, and provides a document information extraction system, a method for extracting the same, and a recording medium recording the information obtained through document recognition as a database. will be.

Another object of the present invention is to provide a document information extraction system, a method of extracting the same, and a recording medium recording the same, which can provide a faster service to a user through a built database.

In order to achieve the above object, a document information extraction system according to the present invention comprises a recognition unit for recognizing a code from a drawing part of a document comprising a description part which is a text area and a drawing part which is a picture area, and extracting information related to the code from the description part. And an extraction unit, a database in which a plurality of documents are stored and the information extracted by the extraction unit is stored.

In the document information extracting system according to the present invention, the information includes at least one of the number and order of the drawings constituting the drawing unit, and the code and code name.

In the document information extracting system according to the present invention, the extracting unit extracts information by using location information of text blocks for each drawing, which is detailed description of each drawing.

,

또는

의 형식으로 기재된 내용을 상기 인식부에서 인식한 부호에 대응한다고 간주하여 판독하는 것을 특징으로 한다.In the document information extracting system according to the present invention, the extracting section is provided in the description section.

,

or

The content described in the form of? Is considered to correspond to a code recognized by the recognition unit, and is read.

In the document information extracting system according to the present invention, the extracting unit reads the noun type existing in the preceding or rearward of the code in the description section as a code name.

In the document information extraction system according to the present invention, one or more positions of the text blocks may be set for each drawing.

In order to achieve the above object, the document information extraction method according to the present invention comprises the steps of: (a) selecting a document from a database in which a plurality of documents including a description part which is a text area and a drawing part which is a picture area are stored; (B) extracting number information and order information of the drawings from the drawing description section described in the description section; and (d) counting information and order information of the drawing as drawing information to the database. Storing (e) setting the initial position of the component description described in the description as an initial search position; (G) reading a code and a code name corresponding to the code recognized in step (b) in the text block for each drawing, (h) And adding the code and the code name to the drawing information of the step (d).

In addition, in the document information extraction method according to the present invention, the drawing description part is characterized in that the part described in the 'short description of the drawings' when the document is a Korean patent document.

Further, in the document information extracting method according to the present invention, the drawing description section is characterized in that the part described in the 'simple description of the drawings' when the document is a Japanese patent document.

In addition, in the document information extraction method according to the present invention, the drawing description portion is characterized in that the part described in the BRIEF DESCRIPTION OF THE DRAWINS when the document is a US patent document.

In addition, in the document information extraction method according to the present invention, when the document is a Korean patent document, the document is characterized in that the part 'the composition and operation of the invention' is described.

Further, in the document information extraction method according to the present invention, the configuration description portion is a part in which "embodiment of the invention" is described when the document is a Japanese patent document.

In addition, in the method for extracting document information according to the present invention, if the document is a US patent document, it is characterized in that the 'detailed description of the invention (DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS)' is described.

,

또는

의 형식으로 기재된 내용을 부호로 간주하여 판독하는 단계를 포함하는 것을 특징으로 한다.In the document information extracting method according to the present invention, the step (g)

,

or

It is characterized in that it comprises the step of reading the content described in the form of a sign.

The document information extracting method according to the present invention is characterized in that the step (g) further includes a step of reading the noun type existing at the front or the rear of the code as a code name.

In the method for extracting document information according to the present invention, in the step (f), at least one position of the text block for each drawing may be set per drawing.

In addition, to achieve the above object, a computer-readable recording medium according to the present invention comprises the steps of: (a) selecting a document from a database in which a plurality of documents including a description part which is a text area and a drawing part which is a picture area are stored; (b) recognizing a code from the drawing part, (c) extracting number information and order information of the drawing from the drawing description part described in the description part, (d) number information and order of the drawing which is drawing information Storing information in the database, (e) setting the initial position of the component description described in the description as the initial search position, and (f) drawing-specific text that is a detailed description of each drawing in the configuration description. Specifying the location information of the blocks; (g) selling the code and the code name corresponding to the code recognized in the step (b) in the text block for each drawing; Step, (h) that is characterized by recording a program for executing the step of adding the code as the code name of the drawing information of the step (d).

DETAILED DESCRIPTION Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the present invention. In addition, in describing this invention, the same code | symbol is attached | subjected and the repeated description is abbreviate | omitted.

1 is a view briefly showing the structure of a patent document according to an embodiment of the present invention.

1A is a structure of a Korean patent document, FIG. 1B is a Japanese patent document, and FIG. 1C is a US patent document.

As shown in Figs. 1A to 1C, Korean, Japanese, and US patent documents have a specific structure. That is, the patent document has a distinguishable form and has a state in which it can be stored and processed. The patent document is largely composed of a description section (10, 30, 50) that is a text area and a drawing section (20, 40, 60) that is a picture area.

First, the description portion 10 of the Korean patent document consists of bibliographic information 110, summary 120, representation diagram 130, index word 140, specification 150, claims 160. Among them, the bibliographic item 110 includes an application number 110a, an application date 110b, an applicant 110c, a name of the invention 110d, and the like. In addition, the specification 150 consists of a brief description 151 of the drawings and a detailed description 152 of the invention. The detailed description 152 of the invention consists of the object 153 of the invention, the constitution and operation 156 of the invention, and the effect 157 of the invention. At this time, the object 153 of the invention consists of the technology to which the invention belongs, and the prior art 154 in the field and the technical problem 155 to achieve the invention. The drawing unit 20 consists of a drawing 210.

Next, the explanation section 30 of the Japanese patent document consists of a bibliographic item 310, a summary 320, a representation diagram 330, a claim 340, and a detailed description 350 of the invention. Among these, the bibliographic information 310 includes an application number 310a, an application date 310b, an applicant 310c, a name 310d of the invention, and the like. In addition, the detailed description 350 of the invention includes technical field 351 to which the invention belongs, conventional technology 352, problem 353 to be solved by the invention, means 354 to solve the problem, and embodiments of the invention. 355, the effect 356 of the invention, and the brief description 357 of the drawings. The drawing unit 40 is made up of a drawing 410.

Finally, the description section 50 of the US patent document includes bibliography 510, summary 520, representation 530, invention background 540, invention summary 550, and a brief description of the drawings (BRIEF). DESCRIPTION OF THE DRAWINS 560, DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS 570, and CLAIM 580. Among these, the bibliographic information 510 includes an application number 510a, an application date 510b, an applicant 510c, a name of the invention 510d, and the like. In addition, the background 540 of the invention consists of the field of invention 541 and the background art 542. The drawing unit 60 is made up of the drawing 610.

Although the structure of the patent documents of Korea, Japan, and the United States has been described in the above description, not only the structure mentioned in the above description but also a more detailed structure may be used. Of course.

In the structure of such a patent document, there are sections describing the drawings and the description of the composition of the invention, which are referred to as a drawing description section and a configuration description section, respectively. The drawing description section is a brief description of the drawings (151, 357) for a Korean or Japanese patent document, and BRIEF DESCRIPTION OF THE DRAWINS (560) for a US patent document. In addition, the configuration description is the configuration and operation (156) of the invention in the case of the Korean patent document, the embodiment of the invention (355) in the case of the Japanese patent document, DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS (570) to be. Hereinafter, the drawing description section will be referred to as 'drawing description sections 151, 357 and 560', and the configuration description section will be referred to as 'configuration description sections 156, 355 and 570'.

Next, a document information extraction system according to an embodiment of the present invention will be described with reference to FIG.

2 is a block diagram illustrating a document information extraction system according to an embodiment of the present invention.

As shown in FIG. 2, the document information extraction system 70 is a drawing section 20 of a patent document consisting of a description section 10, 30, 50, which is a text area, and drawing sections 20, 40, 60, which are picture areas. Recognition unit 71 for recognizing a code from 40, 60, extraction unit 72 for extracting drawing information corresponding to code recognized by recognition unit 71 from description unit 10, 30, 50, many And a database 73 for storing the document information and storing the drawing information extracted by the extraction unit 72. Drawings 20, 40, and 60 are drawing areas 1001, 211, 411 and 611 and letters indicating specific portions of the drawing (for example, 100 shown in the drawing of the drawing in Figs. 1A, 1B and 1C). , 200, 300), which are the symbols of the drawings. That is, the code | symbol recognized by the recognition part 71 means the member numbers 211a, 411a, 611a described in the drawing part 20, 40, 60. As shown in FIG. In addition, the drawing information extracted by the description units 10, 30, and 50 refers to the number of drawings, the order of drawings, signs, code names, and the like.

The extraction unit 72 shown in FIG. 2 extracts the number and order information of the drawings from the drawing description units 151, 357, and 560 of the description units 10, 30, and 50. That is, in the drawing description units 151, 357, and 560 of the patent document, a brief description of each drawing is described, such as 'Fig. 1 is' or 'Fig. 2 is'. Therefore, it is possible to obtain information on how many drawings are inserted in the patent document and in which order the drawings are arranged.

,

또는

(이하,

는 공란(Space)을 의미한다)의 형식으로 기재된 내용을 인식부(71)가 인식한 부호에 대응한다고 간주하여 판독한다. 예를 들어, 인식부(71)가 도면부(20, 40, 60)에서 부호 '100'을 인식할 경우, 추출부(72)는 설명부에서

,

또는

의 형식으로 기재된 내용을 부호 '100'에 대응한다고 판독한다. 한국 특허문서의 경우는

, 일본 특허문서의 경우는

, 그리고 미국 특허문서의 경우는

을 주로 사용한다. In addition, the extraction unit 72 extracts a code and a code name from the text described in the configuration description units 156, 355, and 570 of the description units 10, 30, and 50. In this case, the code retrieves and reads the code corresponding to the code recognized by the recognition unit 71, and extracts the drawing information using the location information of the text blocks of each drawing, which is a detailed description of each drawing. Here, the location information is obtained by using an identification number in case of a Korean or Japanese patent document and a line number in case of a US patent. That is, descriptions of the drawings described in the configuration description units 156, 355, and 570 as 'FIG. 1' and 'FIG. 2' are divided into text blocks for each drawing, and the drawing information of FIG. Searching in the text block of FIG. 1 extracts the drawing information more effectively. The extraction unit 72 extracts drawing information corresponding to the code recognized by the recognition unit 71. In the description units 10, 30, and 50,

,

or

(Below,

The contents described in the form of "space" are regarded as corresponding to the code recognized by the recognition unit 71, and are read. For example, when the recognition unit 71 recognizes the symbol '100' in the drawing units 20, 40, and 60, the extraction unit 72 is used in the description unit.

,

or

The content described in the following format is read as corresponding to the symbol '100'. In the case of Korean patent documents

For Japanese patent documents

, And in the case of US patent documents,

Mainly used.

The extraction unit 72 reads the code name after the code is read by the description units 10, 30, and 50. The code name is regarded as a noun form existing at or after the code read from the description. For example, when the read code is '(100)', the code name searches for the potential or the back of '(100)'. If the sign '(100)' belongs to the word 'computer 100', the code name becomes 'computer' which is a noun type located at the potential of '(100)'. As another example, when the read code is '200', the code name is 'monitor' if the code '200' belongs to the word 'monitor 200'.

One or more positions of the text block for each drawing used when extracting the drawing information may be set for each drawing. For example, after the text block of FIG. 1 begins with 'FIG. 1 begins with' and the FIG. 2 text block begins with 'FIG. 2 begins with', the text block of FIG. 1 begins with 'FIG. This can happen. In addition, when only one sign is not found in the first text block of FIG. 1, a case may be required to search for a sign in the second text block. Therefore, one or more positions of text blocks may be set for each drawing.

After the drawing information extraction is completed, the database 73 stores drawing information such as the number of drawings, drawing order, codes, and code names extracted from the drawing description units 151, 357, and 560 and the configuration description units 156, 355, and 570. Store in

Next, a document information extraction method according to an embodiment of the present invention will be described with reference to FIG.

3 is a flowchart illustrating a document information extraction method according to an embodiment of the present invention.

,

또는

의 형식으로 기재된 내용을 부호로 간주하여 판독한 후, 판독한 부호의 전위 또는 후위에 존재하는 명사형을 부호 명칭으로 간주하여 판독한다. 판독이 완료되면, 부호와 부호 명칭을 데이터베이스(73)에 저장한 도면 정보에 추가한다(ST 3080). 이로써, 도면 정보가 완성된다.As shown in FIG. 3, one document to be worked on is selected from among a plurality of documents stored in the database 73 (ST 3010). Next, the recognition unit 71 recognizes a code from the drawing units 20, 40, and 60 of the document (ST 3020). The extractor 72 extracts the characters corresponding to the codes recognized by the drawing units 20, 40, and 60 from the description units 10, 30, and 50. The extraction process is as follows. BRIEF DESCRIPTION OF THE DRAWINS (560) or the brief description of the drawings (151, 357, 560), wherein the extraction section (72) is the drawing description (151, 357, 560) of the description (10, 30, 50). ) And extract the number and order information of the drawing (ST 3030). That is, information about the number and order of the entire drawings is obtained through the contents described as 'FIG. 1 is' and 'FIG. 2 is'. The extraction unit 72 stores the number and order information of the extracted drawings in the database 73 as drawing information (ST 3040). Next, the structure and action 156 of the invention, the embodiment 355 of the invention, or the invention of which the extraction section 102 is the component description parts 156, 355, 570 of the description parts 10, 30, 50. The position where the DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS 570 starts is set as the initial search position (ST 3050). This is to reduce search time and server load. Now, the extraction unit 72 embodies the position information of the text block for each drawing from the initial search position of the configuration description units 156, 355, and 570 (ST 3060). For example, each of the locations described in FIGS. 1 to 2, 3 and 3 is searched, and the searched location is stored. In this case, the contents described between FIGS. 1 to 2 and 2 to 2 are text blocks of FIG. Extraction of the drawing information is made in the component description sections 156, 355, and 570. FIG. Of course, one or more positions of the text block for each drawing may be set. Now, the sign and the sign name are read from the text block of each drawing by using the acquired position information (ST 3070). In this case, the code only needs to be searched for the form corresponding to the code recognized in step (b). Therefore, the extraction unit 72

,

or

The content described in the form of " a " is regarded as a code, and then read. When the reading is completed, the code and code name are added to the drawing information stored in the database 73 (ST 3080). Thus, the drawing information is completed.

Next, a more specific embodiment of the present invention will be described.

The document information extraction method is largely performed by five steps.

1. Step 1

First, after any one patent document stored in the database 73 is selected (ST 3010), the recognition unit 71 recognizes a code in the drawing units 20, 40, and 60 (ST 3020). Next, the extraction unit 72 obtains the information of the order of the drawings for each document and the number of data fields per document in the drawing description units 151, 357, and 560 (ST 3030). For example, assume that patent documents A and B are as follows.

Patent Document A

"[A Brief Description of Drawings]

1 shows a frame of a video display screen in which the present invention is used;

2 is a block diagram of a memory circuit constructed in accordance with the principles of the invention;

3 is a block diagram of another first embodiment of an address generator portion of a memory circuit constructed in accordance with the principles of the invention;

4 is a block diagram of another second embodiment of an address generator portion of a memory circuit constructed in accordance with the principles of the invention;

5 is a block diagram of an address sequencer used by an address generator portion of a memory circuit constructed in accordance with the principles of the present invention.

The patent document A has five drawings, which are configured in the order of FIGS. 1, 2, 3, 4, and 5.

Patent Document B

"[A Brief Description of Drawings]

1 shows a frame of a video display screen in which the present invention is used;

2 is a block diagram of a memory circuit constructed in accordance with the principles of the invention;

 (a) Embodiment 1 of the block diagram

 (b) Embodiment 2 of the block diagram

3 is a block diagram of another first embodiment of an address generator portion of a memory circuit constructed in accordance with the principles of the invention;

4 is a block diagram of an address sequencer used by an address generator portion of a memory circuit constructed in accordance with the principles of the present invention.

The patent document B has five drawings, and is composed in the order of FIGS. 1 and 2 (a) and 2 (b), 3, 4 and 5.

The extraction unit 72 automatically stores the drawing information generated for each document in the database 104 as shown in Table 1 (ST 3040).

Document number drawing sign Code Name sign Code Name sign Code Name   Document A Figure 1 100 computer 200 monitor 300 keyboard Figure 2 ... ... ... ... ... ... Figure 3 ... ... ... ... ... ... Figure 4 ... ... ... ... ... ... Figure 5 ... ... ... ... ... ...   Document B Figure 1 ... ... ... ... ... ... Figure 2 (a) ... ... ... ... ... ... Figure 2 (b) ... ... ... ... ... ... Figure 3 ... ... ... ... ... ... Figure 4 ... ... ... ... ... ... Figure 5 ... ... ... ... ... ...

2. Step 2

The extraction unit 72 sets the position where the configuration description units 156, 355, and 570 start as the initial search position (ST 3050). The initial search position is set in order to reduce the server load and the search time for searching for characters in the description sections 10, 30, and 50 of the patent document.

3. Step 3

The extraction unit 72 starts at the rear of the set initial search position and starts with the positions of 'Fig. 1', 'Fig. 2', 3 'or' Fig 1 ',' Fig 2 'and' Fig 3 'that appear first. The searched position is stored (ST 3060). That is, in order to store each drawing information, a text block in which information for each drawing is located is distinguished. This is to distinguish between text blocks between the position where FIG. 1 is searched and the position where FIG. 2 is searched, so that the order information between drawings embodies the position information between blocks based on the drawing order information obtained in step 1. .

4. Step 4

,

또는

의 형식을 가진 문자가 존재하면 각 문자들을 구분하여 인식한다. 인식한 문자들은 부호로 간주하여 판독한다. 이때, 단계 (ST 3030)에서 획득한 정보에 의거하여 작업하는 도면이 가장 마지막에 위치하는 도면일 경우, 텍스트 블록은 구성 설명부(156, 355, 570)이내로 제한다. 즉, 마지막 도면인 '도 5'의 경우에는 디폴트 텍스트 블록(Default Text Block) 값을 지정하거나, 각 설명부의 발명의 구성 및 작용(156), 발명의 실시 형태(355) 또는 발명의 상세한 설명(DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS)(570)이내의 내용만을 도면 정보 획득 블록으로 지정한다.The extraction unit 72 blocks the text blocks of each of FIGS. 1, 2, and 3, respectively, and displays the contents between the position where FIG. 1 is searched and the position where FIG. 2 is searched. 1 'information is recognized as a block, and then, corresponding to a code recognized by the recognition unit 72 in the recognized text block,

,

or

If characters with the form of are present, they are recognized separately. Recognized characters are read as if they are signs. In this case, when the drawing working on the basis of the information obtained in step ST 3030 is the last drawing, the text block is removed within the configuration description units 156, 355, and 570. That is, in the case of FIG. 5, which is the last drawing, a default text block value is designated, or the configuration and operation 156 of the invention of each description section, the embodiment 355 of the invention, or the detailed description of the invention ( Only contents within the DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS) 570 are designated as drawing information acquisition blocks.

5. Step 5

The extraction unit 72 reads the noun type existing at the front or the rear of the code extracted in step 4 as a code name capable of matching the code (ST 3070). At this time, one or more positions of blocks in each drawing can be set in the patent document. This means that even one code format may not be searched in the text block that the information of FIG. 1 is considered to be located.

Finally, the extraction unit 72 adds the read code and the code name to the drawing information stored in the database 73 (ST 3080). Thus, the drawing information is completed.

As described above, the present invention is capable of acquiring and processing information corresponding to the position and size between the respective drawing elements through text morpheme analysis located in the drawing information. That is, whether or not the drawing component 100, 110 is located at the top, the component 200, whether the component 300 of the same size on the upper portion intersecting, the component 400 of the component, 100, 200, 300, 400 Data such as whether it is located in the form of connecting 100 and 300 can be obtained by directly interpreting a patent document.

As mentioned above, although the invention made by this inventor was demonstrated concretely according to the said Example, this invention is not limited to the said Example and can be variously changed in the range which does not deviate from the summary.

As described above, according to the document information extraction system, the extraction method thereof, and the recording medium recording the same, the information obtained through the document recognition can be constructed into a database and provided to the user.

In addition, according to the document information extraction system, the extraction method and the recording medium recording the same according to the present invention, it is also possible to provide a faster service to the user through the built database.

In addition, according to the document information extraction system, the extraction method and the recording medium recording the same according to the present invention, it is also possible to reduce the purchase cost of OCR products due to the database construction.

Claims (17)

A recognition unit for recognizing a code from a drawing unit of a document including a description unit which is a text area and a drawing unit which is a picture area; An extraction unit for extracting information associated with the code from the description unit; And a database in which a plurality of the documents are stored and the information extracted by the extracting unit is stored. The method of claim 1, And the information includes at least one of the number and order of drawings constituting the drawing unit and the code and code name. The method of claim 1, And the extracting unit extracts information using location information of the text blocks corresponding to each drawing, which is a detailed description of each drawing. The method according to claim 1 or 3, The extraction section is in the description section

,

or

The document information extraction system characterized in that it reads the content described in the form of "corresponding to the code recognized by the said recognition part." The method of claim 4, wherein And the extracting unit reads the noun form existing at the front or the rear of the code as the code name in the description unit and reads it. The method of claim 3, wherein And at least one position of the text block per drawing can be set for each drawing. (a) selecting any one document from a database in which a plurality of documents including a description part which is a text area and a drawing part which is a picture area are stored; (b) recognizing a sign from the drawing part, (c) extracting number information and order information of the drawings from the drawing description section described in the description section; (d) storing the number information and the order information of the drawing as drawing information in the database; (e) setting the initial position of the component description described in the description as the initial search position, (f) embodying location information of the text blocks for each drawing which is a detailed description of each drawing in the configuration description section; (g) reading the code and code name corresponding to the code recognized in step (b) in the text block for each drawing; (h) adding the code and the code name to the drawing information of the step (d). The method of claim 7, wherein The drawing description unit is a document information extraction method, characterized in that the part is described 'simple description of the drawings' when the document is a Korean patent document. The method of claim 7, wherein The document description section is a document information extraction method, characterized in that the part is described in the "short description of the drawings" when the document is a Japanese patent document. The method of claim 7, wherein The drawing description unit is a document information extraction method, characterized in that the 'BRIEF DESCRIPTION OF THE DRAWINS' is described when the document is a US patent document. The method of claim 7, wherein The configuration description unit is a document information extraction method, characterized in that the part is described 'constitution and operation of the invention' when the document is a Korean patent document. The method of claim 7, wherein And the configuration description portion is a portion in which 'embodiment of the invention' is described when the document is a Japanese patent document. The method of claim 7, wherein The configuration description unit is a document information extraction method, characterized in that the 'DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS' is described when the document is a US patent document. The method of claim 7, wherein Step (g) is

,

or

And reading out the content described in the form of a code as a code. The method of claim 14, The step (g) further includes the step of reading the noun form existing at the front or the rear of the code as a code name and reading the document information. The method of claim 7, wherein And in the step (f), at least one position of the text block for each drawing can be set for each drawing. (a) selecting any one document from a database in which a plurality of documents including a description part which is a text area and a drawing part which is a picture area are stored; (b) recognizing a sign from the drawing part, (c) extracting number information and order information of the drawings from the drawing description section described in the description section; (d) storing the number information and the order information of the drawing as drawing information in the database; (e) setting the initial position of the component description described in the description as the initial search position, (f) embodying location information of the text blocks for each drawing which is a detailed description of each drawing in the configuration description section; (g) reading the code and code name corresponding to the code recognized in step (b) in the text block for each drawing; (h) a computer-readable recording medium having recorded thereon a program for executing the step of adding the code and the code name to the drawing information of the step (d).

Images