JP2001265762A - Document structure extracting device and document structure information extracting method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a document structure extracting device capable of extracting a document structure from an electronic document without using a dictionary. SOLUTION: Concerning the document structure extracting device for extracting document structure information from the electronic document, this device is provided with a character information generating part 103 for generating character information containing information on the position, character size and character type of each character by analyzing the document, a line information generating part 105 for generating line information containing information on the character string of each line, the main character size and main character type of each line and the score of each line by analyzing the character information and a document information generating part 107 for generating the document structure information by analyzing this line information. The document structure information generating part 107 generates the document structure information by grouping the lines on the basis of the score of the line information and the continuity of lines. Thus, the document structure information can be extracted from the electronic document without using the dictionary.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document structure extracting apparatus and method for extracting document structure information such as chapters and sections from an electronic document, and more particularly, to a method for extracting document structure information without using a dictionary. It was done.

[0002]

2. Description of the Related Art In February 1998, W3C recommended X
With the release of ML1.0, XML as a page description language has recently attracted attention. In the future, XML
It is anticipated that document management that makes use of the structure that is a characteristic of XML will become widespread for electronic documents described in.

However, existing documents have a structure on a print image, but many documents do not have document structure information such as a chapter or section start position as data. A document structure extraction device that extracts document structure information from document data of a document is being developed.

As shown in FIG. 22, a conventional document structure extracting apparatus includes an input unit 2601 for inputting an original sentence, an original sentence storage unit 2602 for storing an original sentence, and a word used for separating a heading from a character other than the heading. Headword Separator Dictionary 2608
And a heading delimiter extraction unit 2607 for searching the document data for words included in the heading delimiter dictionary 2608 to extract heading delimiter candidates, and a heading range determination for determining a heading range from the heading delimiter candidates based on the heading delimiter rule dictionary 2610 Unit 2609, a headword dictionary 2613 containing headwords, a heading determination unit 2611 that determines whether a character string in the heading range is a heading based on the headword dictionary 2613 and the heading rule dictionary 2612, and a heading determination A document structure rule dictionary is used for a character string determined by the
A document structure determining unit 2614 that determines a logical structure of a chapter heading, a section heading, a paragraph, and the like using 2615, and a document structure determining unit 2614
Logical structure storage unit 2606 that stores the logical structure determined in
And a display unit 2604 for displaying the determined logical structure and the like,
A display control unit 2603 that controls display on the display unit 2604 and a document management unit 2605 that controls each unit are provided.

In this document structure extraction device, the original sentence input from the input unit 2601 or, if the original sentence has already been input, the original sentence called from the original sentence storage unit 2602 is sent to the document management unit 2605. The document management unit 2605 sends the document data to the headline break candidate extraction unit 2607.

The heading delimiter extraction unit 2607 uses the heading delimiter dictionary 2608 to search the document data for a character that has a high possibility of separating a heading from a heading other than a heading. This is a heading delimiter candidate indicating the end position. The heading range determination unit 2609
Document management unit 2605 based on heading separation rule dictionary 2610
The headline range is determined by judging the headline delimiter candidates sent from.

The heading judging unit 2611 includes a heading range judging unit 26
The entry word candidates in the range obtained in 09 are checked against the entry word dictionary 2613 to determine whether or not the entry is a heading.

[0008] The document structure determination unit 2614 includes a heading determination unit 2611.
Thus, the logical structure such as a chapter heading, a section heading, or a paragraph is determined for a heading or a non-heading. The result is written into the logical structure storage unit 2606 by the document management unit 2605, and is displayed by the display control unit 2603 on the display unit.
Displayed in 2604.

[0009]

However, since the conventional document structure extraction device extracts a document structure using a plurality of dictionaries such as a headword dictionary, it is necessary to first create these dictionaries. However, there is a problem that the cost of creating the dictionary is high. Further, it is necessary to change the dictionary to be used depending on the type of document such as a patent document or a novel, and there is a problem that various types of dictionaries must be prepared.

An object of the present invention is to solve such a conventional problem, and an object of the present invention is to provide a document structure extracting apparatus and a method for extracting a document structure from a document without using a dictionary. And

[0011]

Therefore, according to the present invention, in a document structure extracting device for extracting document structure information from an electronic document, the document is analyzed and a character including information on the position, character size and character type of each character is analyzed. A character information generating unit that generates information, and a line information generating unit that analyzes the character information to generate a character string of each line, and line information including information of a character size and a character type most frequently used in each line. And a document structure information generating section for analyzing the line information to generate document structure information.

In a document structure information extracting method for extracting document structure information from an electronic document, a document is analyzed to generate character information including information on the position, character size and character type of each character, and from this character information Generate line information including the character string of each line, the character size and character type most frequently used in each line, and the score of each line, and group the lines based on the score of this line information and the continuity of the line, It generates document structure information including page, start line, end line, score, and character string information.

Further, the document is analyzed to generate character information including information on the position of each character, and from this character information, line information including the character string information of each line is generated. Analyze, generate table of contents information including title information indicating the character string of the title, page information indicating the number of pages, and a score obtained from the title information, search the character string of the title of the table of contents information from the line information, and Is obtained, and based on this, document structure information including information on a page, a start line, an end line, a score, and a character string is generated.

Therefore, the document structure information can be extracted from the electronic document without using a dictionary.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS (First Embodiment) As shown in FIG. 1, a document structure extraction apparatus according to a first embodiment stores a document input unit 101 for inputting document data and the input document data. Document storage unit 102 to analyze the document and character information (information such as the page where each character in the document is located, the number of characters from the first character of the page, character code, character type, character size, and position on the page) Character information generation unit 10 that generates
3 and a character information storage unit 104 for storing the generated character information
And character information, and analyzes the line information (the page where each line in the document is located, the position of the line on the paper, the main character size, main character type, character string, score (evaluation score)
Line information generating unit 105 that generates the line information, a line information storage unit 106 that stores the generated line information, and analyzes the line information to obtain document structure information (score of each structure, start page of each structure) , A start line, an end page, an end line, a character string, etc.), and a document structure information storage unit 108 for storing the generated document structure information.

The operation of the document structure extracting apparatus configured as described above will be described.

The document input from the document input unit 101 is stored in the document storage unit 102. The character information generation unit 103 analyzes the document in the document storage unit 102, generates character information, and stores it in the character information storage unit 104. The line information generation unit 105 analyzes the character information in the character information storage unit 104 to generate line information,
The information is stored in the row information storage unit 106. Document structure information generation unit 107
Analyzes the line information in the line information storage unit 106 to generate document structure information, and stores it in the document structure information storage unit 108.

Next, a method of generating line information and document structure information will be described with reference to FIGS. 2, 3, 4, 5, and 6. FIG.

FIG. 2 shows an example of a document stored in the document storage unit 102, and FIG. 3 shows an example of character information generated by the character information generation unit 103 from the document of FIG. As character information, for each character, information on the page in the document, the number of characters from the first character of the page, the character, the character size, the character type, and the position on the page are generated.
Is stored in

FIG. 4 is a flowchart showing a method of generating line information in the line information generation unit 105.

Step 402: The line information generation unit 105 sets LINE = 1 as a line for which line information is to be generated. Step 403: Reads one character from the character information storage unit 104. Step 404: Character information storage unit It is determined whether characters of the same page remain in 104, and if so, step 405: read the next character from the character information storage 104, step 406: the Y coordinate of the character obtained in step 405,
The Y-coordinate of the previous character is compared, and if they are the same, the process returns to step 404 to read the next character from the character information storage unit 104.

If the Y coordinates are different in step 406, step 407: the number of pages, the character string, the most frequently used character size, and the character type are obtained from the character information obtained in steps 403 and 405. Step 408: LINE = LINE + 1 is set, and the procedure from step 404 is repeated.

If no character of the same page remains in the character information storage unit 104 in step 404, step 409: from the character information acquired in step 403 and step 405, the number of pages, the character string, The obtained character size and character type are obtained and stored in the record of the number of lines = LINE. Step 410: A score is calculated from the line information excluding the score, and stored in the line information in the line information storage unit 106. The score is, for example, (Character size + character type (Gothic +
1. Calculate according to +0)) for Mincho.

FIG. 5 is an example of line information generated through the process of FIG.

The document structure information generation unit 107 groups continuous lines having the same or similar scores, generates start page, start line, end page, end line, and character string data, and stores the document structure information. It is stored in the unit 108.

FIG. 6 shows an example of the generated document structure information. In the row information shown in FIG. 5, rows 5 and 6 have the same score and are continuous, and are therefore grouped into one.

As described above, in this document structure extraction device,
Character information of each character is obtained from the electronic document, line information of each line is obtained based on the character information, and document structure information can be extracted from the line information.

As the character information, the page of each character in the document, the number of characters from the first character of the page, the character code, the character type, the character size, and the position on the page are shown, but the present invention is not limited thereto. For example, it is also possible to add the color of the character to the character information.

As the line information, the page of each line in the document, the position of the line on the paper, the main character size, the main character type,
Although the character string and the score are shown, the present invention is not limited to this.

Although (character size + character type (+1 for Gothic, +0 for Mincho)) is used as the formula for calculating the score, the present invention is not limited to this.

Further, an XML document can be created by using the document structure information stored in the document structure information storage unit 108.

(Second Embodiment) The document structure extracting apparatus according to the second embodiment acquires document structure information by using line information of a document registered in the past as a learning material.

As shown in FIG. 7, the apparatus includes a past row information storage unit 708 for storing past row information.
The past line information is line information of a document analyzed in the past,
As illustrated in FIG. 8, in addition to the information of the score, the main character size, and the main character type, information of the structure name (chapter title, section title, text, etc.) is included. The document structure information generation unit 707 generates necessary items of the document structure information (here, a structure name) using the past line information stored in the past line information storage unit 708. Other configurations are the same as those of the first embodiment (FIG. 1).

In this document structure extraction device, the document input unit 70
The document input from 1 is stored in the document storage unit 702. The character information generation unit 703 analyzes the document stored in the document storage unit 702, and obtains the character information (page of each character in the document,
The number of characters, the character code, the character type, the character size, and the position on the page of the page from the top character of the page are generated and stored in the character information storage unit 704.

The line information generation unit 705 analyzes the character information in the character information storage unit 704, and obtains the line information (page of each line in the document, line on paper, main character size, main character type, character string) in FIG. , Score) and stores them in the row information storage unit 706.

The document structure information generation unit 708 groups continuous lines having the same or similar scores from the line information, and forms a document structure including data of a start page, a start line, an end page, an end line, and a character string. Generate information, and
The past line information having the same main character type, main character size, and score set as the line information is searched from the past line information of FIG. 8 stored in the past line information storage unit 708, and the structure name of the corresponding past line information is searched. Is acquired and added to the document structure information, and the document structure information including the (score of each structure, structure name, start page, start line, end page, end line, character string) is stored in the document structure information storage unit 708. .

FIG. 9 shows an example of the document structure stored in the document structure information storage unit 708.

As described above, the document structure information extracting apparatus can obtain necessary item information in the document structure information by using the past line information.

Further, line information can be generated based on the document structure information and used as past line information. By accumulating the past line information and enriching the learning materials, it is possible to improve the accuracy of obtaining the document structure information using the past line information.

(Third Embodiment) The document structure extracting apparatus according to the third embodiment acquires necessary document structure information by using the table of contents information.

As shown in FIG. 10, this apparatus acquires a page in which a table of contents is described from a document storage unit 1102 and generates table of contents information, and stores the generated table of contents information. And a table of contents information storage unit 1108,
The document structure information generation unit 1109 analyzes the table of contents information and the line information to generate document structure information. Other configurations are first
There is no difference from the embodiment (FIG. 1).

In this document management device, the document input from the document input unit 1101 is stored in the document storage unit 1102, and the character information generation unit 1103 performs the same operation as in the first embodiment. Analyzing the document, as shown in FIG. 3, (the page where each character in the document exists, the number of characters from the top of the page,
Character information including a character code, a character size, a character type, and a position on a page of paper is generated, and stored in the character information storage unit 1104.

From the character information, the line information generation unit 1105
In the same manner as in the first embodiment, line information including (a page of each line in a document, a line on paper, a main character size, a main character type, and a character string) is generated. However, the score is not included in the line information.

The table of contents information generation unit 1107 obtains a page in which a table of contents is described from the document storage unit 1102. FIG. 11 is an example of a table of contents page. At this time, the table of contents information generation unit 1107
Indicates a table of contents such as "table of contents" or "table of contents" from the character information storage unit 1104, and searches for a character string that is larger than the main character size of the document and close to the top of the page, and obtains the corresponding page as the table of contents I do. It should be noted that the table of contents page may be the one specified by the user.

The table of contents information generating unit 1107 analyzes each line described in this table of contents page,
It is divided into a number part indicating the number of start pages, and a score is set according to the character start position (X coordinate) of the character part on the paper and the number at the beginning of the character part, and the table of contents information illustrated in FIG. 12 is created. Is stored in the table of contents information storage unit 1108.

The document structure information generation unit 1109 includes a line information storage unit
The line information in 1106 and the table of contents information in the table of contents information storage unit 1108 are analyzed to generate document structure information, and the document structure information storage unit
Stored in 1108.

At this time, the document structure information generation unit 1109 searches the line information for a character string of the character part of the table of contents information from a page corresponding to the numeric part of the table of contents information, and deletes the line corresponding to the character string. get. Further, a line of the body between the lines corresponding to the character string of the character part is calculated, and the character string of the body is obtained from the line information. In addition, the score of the table of contents information is used as the score of the document structure information, and the score of the text not included in the table of contents information is lowered by one rank. In this way, document structure information including the start page, start line, end page, end line, character string, and score of each structure is generated. FIG. 13 shows the generated document structure information.

As described above, in this document structure extraction device,
Document structure information can be generated using the table of contents information. At this time, since the line information is used together with the table of contents information, it is possible to acquire detailed structural information (paragraphs, bullet points, etc.) not described in the table of contents.

(Fourth Embodiment) The document structure extracting apparatus according to the fourth embodiment calculates the score of the document structure information in consideration of the space between the previous line and the position of the first character of the line.

The block configuration of this device is the same as that of the first embodiment (FIG. 1).

In this device, the document input unit 101, the document storage unit 102, the character information generation unit 103, and the character information storage unit 104
The same operation as in the first embodiment is performed. When the document shown in FIG. 14 is input from the document input unit 101, the character information generation unit 10
3 generates document information including (a page of each character in the document, the number of characters from the first character of the page, a character code, a character size, a character type, and a position on the page) as shown in FIG. It is stored in the storage unit 104.

The line information generation unit 106 analyzes the character information in the character information storage unit 104, and as shown in FIG. 17, (the page of each line in the document, the line on the paper, the main character size, the main character type, Line information including a line between the previous line, a line between the next line, the first character X coordinate, a character string, and a score), and a line information storage unit.
Store in 106.

FIG. 16 is a flowchart showing a method of generating line information in the line information generation unit 105. In this procedure, steps 1901 to 1907 correspond to steps 401 to 4 of the flowchart (FIG. 4) of the first embodiment.
Same as 07. That is, one character is read from the character information storage unit 102 (step 1903), and it is determined whether characters of the same page remain in the character information storage unit 102 (step 1904). Is obtained (step 1905). The Y coordinate of the character obtained in step 1905 is compared with the Y coordinate of the previous character (step 1906). If they are the same, the next character is obtained from the character information storage unit 104. If the comparison in step 1906 is different,
From the character information acquired in steps 1903 and 1905, the number of pages, the character string, the most frequently used character size, and the character type are acquired and stored in the record of line number = LINE (step 1907).

Step 1908: Further, the line spacing from the previous line and the X coordinate of the first character position are acquired from the character information, and stored in the record of the line number = LINE, and the line spacing with the previous line = LINE is obtained. -1 is recorded in "between lines after the next line".

Step 1909: Next, LINE = LIN
Set to E + 1 and repeat the procedure from step 1904.

In step 1904, the character information storage unit 10
If there are no characters on the same page left in step 2, go to step 1
From the character information acquired in step 903 and step 1905, the number of pages, the character string, the most frequently used character size, and the character type are acquired and stored in the record of line number = LINE (19)
Ten).

Step 1911: Further, the line spacing from the preceding line and the X coordinate of the first character position are acquired from the character information and stored in the record of the line number = LINE, and the line spacing from the preceding line = LINE is obtained. -1 is recorded in "between lines after the next line".

Step 1912: From the line information excluding the score, (character size + character type (+1 for Gothic, +
0) + (line spacing from the previous line (in the case of “-”, the same as the line spacing between the following lines) −line spacing from the following line) ÷ 10 + first character X coordinate ÷ 5) Stored in 106 line information.

FIG. 17 shows an example of line information generated through such a procedure.

The document structure information generation unit 107 includes a line information storage unit 1
The line information in 06 is analyzed, and document structure information (score, start page, start line, end page, end line, character string) of each structure is generated and stored in the structure information storage unit 108.

FIG. 18 shows the generated document structure information.

In this apparatus, by adding the line position information to the score, the document structure of the text can be more finely extracted.

In this case, the score is calculated by (character size + character type (+1 for Gothic, +0 for Mincho) +
(Line spacing with the previous line (in the case of "-", this is the same as line spacing with the following line)
Although the calculation is performed according to -line spacing between the succeeding line) ÷ 10 + the first character X coordinate 、 5), it is also possible to use an expression or the like obtained by multivariate analysis.

(Fifth Embodiment) In the fifth embodiment,
A case will be described in which the document structure extraction device (FIG. 10) of the third embodiment uses the line information of a document registered in the past as a learning material to acquire information on necessary items of the document structure information.

As shown in FIG. 19, the document structure extracting apparatus includes a past line information storage unit 2210 for storing past line information, and a document structure information generating unit 2209 uses the past line information to store a document. Acquires information on required items of structure information. Other configurations are the same as those of the third embodiment.

The past line information is line information of a document registered in the past, and includes, for example, a score, a main character size, a main character type, and a structure name. Document structure information generation unit 2209
When the document structure information is generated according to the procedure described in the third embodiment, the past line information stored in the past line information storage unit 2210 is compared with the line information of the target document, and The structure name is obtained and fetched, and as shown in FIG. 20, (score of each structure, structure name, start page, start line,
It generates document structure information including an end page, an end line, and a character string.

FIG. 20 shows the document structure information when the score of the line information is set based on the table of contents information. FIG. 21 shows the score of the line information set by the method shown in the fourth embodiment. It shows the document structure information in the case where it is set.

As described above, the document structure information of a new document is extracted by using the line information of the document registered in the past as a learning teaching material, and the line information obtained from the extracted document structure information is added to the past line information. By enriching past line information,
The accuracy of extraction is improved.

As described in the first embodiment, the character information used in each embodiment includes the page of each character in the document, the number of characters from the first character of the page, the character code, the character type, the character size, and the page space. The present invention is not limited to the upper position, and it is also possible to use, for example, the color of characters.

Also, by using the structure information stored in the document structure information storage unit of each embodiment, the XML
It is possible to create documents.

[0071]

As is clear from the above description, the document structure extracting apparatus and document structure information extracting method of the present invention can extract document structure information from an electronic document without using a dictionary.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a document structure extraction device according to a first embodiment;

FIG. 2 shows an example of a registered document according to the first embodiment;

FIG. 3 shows an example of character information according to the first embodiment;

FIG. 4 is a flowchart of line information generation according to the first embodiment;

FIG. 5 is an example of line information according to the first embodiment;

FIG. 6 shows an example of document structure information according to the first embodiment;

FIG. 7 is a block diagram illustrating a configuration of a document structure extraction device according to a second embodiment;

FIG. 8 shows an example of past line information according to the second embodiment;

FIG. 9 shows an example of structure information according to the second embodiment;

FIG. 10 is a block diagram illustrating a configuration of a document structure extraction device according to a third embodiment;

FIG. 11 shows an example of a table of contents in a registered document according to the third embodiment;

FIG. 12 shows an example of table of contents information according to the third embodiment;

FIG. 13 shows an example of document structure information according to the third embodiment;

FIG. 14 shows an example of a registered document according to the fourth embodiment;

FIG. 15 shows an example of character information according to the fourth embodiment;

FIG. 16 is a flowchart of line information creation according to the fourth embodiment;

FIG. 17 shows an example of line information according to the fourth embodiment;

FIG. 18 shows an example of structure information according to the fourth embodiment;

FIG. 19 is a block diagram showing a configuration of a document structure extraction device according to a fifth embodiment;

FIG. 20 shows an example of structure information according to the fifth embodiment;

FIG. 21 shows another example of the structure information according to the fifth embodiment;

FIG. 22 is a block diagram showing a configuration of a conventional document structure extraction device.

[Explanation of symbols]

 101, 701, 1101, 2201 Document management device 102, 702, 1102, 2202 Document storage unit 103, 703, 1103, 2203 Character information generation unit 104, 704, 1104, 2204 Character information storage unit 105, 705, 1105, 2205 lines Information generation unit 106, 706, 1106, 2206 Line information storage unit 107, 707, 1109, 2209 Document structure information generation unit 108, 709, 1110, 2211 Document structure information storage unit 708, 2210 Past line information storage unit 1107 Table of contents information generation Unit 1108 Table of contents information storage unit 2601 Input unit 2602 Original text storage unit 2603 Display control unit 2604 Display unit 2605 Document management unit 2606 Logical structure storage unit 2607 Heading separator candidate extraction unit 2608 Heading separator dictionary 2609 Heading range determination unit 2610 Heading separator rule dictionary 2611 Heading judgment unit 2612 Heading rule dictionary 2613 Headword dictionary 2614 Document structure judgment unit 2615 Document structure rule dictionary

 ──────────────────────────────────────────────────の Continuing on the front page (72) Inventor Masao Ito 1006 Kazuma Kadoma, Kadoma City, Osaka Prefecture Matsushita Electric Industrial Co., Ltd. (72) Inventor ▲ Tsuru ▼ Ken Hayashi 1006 Kadoma Kadoma Kadoma City, Osaka Matsushita Electric Industrial Co. In-company (72) Inventor Takashi Shimojima 1006 Kadoma Kadoma, Kadoma-shi, Osaka Matsushita Electric Industrial Co., Ltd. F-term (reference) 5B009 QA01 QA06 QA07 VA01 VA02

Claims (14)

[Claims] 1. A document structure extracting apparatus for extracting document structure information from an electronic document, comprising: a character information generating unit that analyzes the document and generates character information including information on a position, a character size, and a character type of each character; A line information generating unit that analyzes the character information to generate line information including information of a character string of each line, and a character size and a character type most frequently used in each line; and a document structure by analyzing the line information. A document structure extraction device, comprising: a document structure information generation unit that generates information. 2. The method according to claim 1, wherein the line information generating unit calculates a score of each line, and the document structure information generating unit generates the document structure information by grouping the lines based on the score and the continuity of the lines. The document structure extraction device according to claim 1, wherein 3. A document structure, comprising: a table of contents information generating unit that analyzes a table of contents table of the document to generate table of contents information including title information indicating a character string of a title and page information indicating a page number. 2. The document structure according to claim 1, wherein the information generation unit searches the character string of the title from the line information to obtain a line corresponding to the title, and generates the document structure information based on the line. Extraction device. 4. The apparatus according to claim 1, wherein the line information generation unit calculates a score of each line based on information on a character size and a character type most frequently used in each line.
The document structure extraction device according to any one of the above. 5. The line information generating section calculates a score of each line based on information on a character size and a character type most frequently used in each line, a space between lines before a line, and a position of a leading character of each line. 4. The document structure extracting apparatus according to claim 1, wherein the document structure is extracted. 6. The table of contents information calculating section calculates a score corresponding to the title information, and the document structure information generating section calculates a score of the document structure information based on the score calculated by the table of contents information generating section. 4. The document structure extracting apparatus according to claim 3, wherein: 7. A past line information storage unit for storing line information of a document analyzed in the past, wherein the document structure information generation unit uses past line information stored in the past line information storage unit. 7. The document structure information extracting device according to claim 1, wherein necessary information in the document structure information is acquired by using the document structure information. 8. The apparatus according to claim 7, wherein the document structure information generation unit acquires a structure name of the document structure information using past line information stored in the past line information storage unit. Document structure information extraction device. 9. A document structure information extraction method for extracting document structure information from an electronic document, wherein the document is analyzed to generate character information including information on the position, character size, and character type of each character, and from the character information Generate the line information including the character string of each line, the character size and character type most frequently used in each line, and the score of each line, and group the lines based on the score of the line information and the continuity of the line. A document structure information extraction method characterized by generating document structure information including information of a page, a start line, an end line, a score, and a character string. 10. The document structure information extracting method according to claim 9, wherein a score of each line of the line information is calculated based on information on a character size and a character type most frequently used in each line. 11. Calculating a score of each line of the line information based on information of a character size and a character type most frequently used in each line, a space between lines before each line, and a position of a leading character of each line. 10. The document structure information extracting method according to claim 9, wherein: 12. A document structure information extracting method for extracting document structure information from an electronic document, wherein the document is analyzed to generate character information including information on the position of each character, and the character string information of each line is generated from the character information. On the other hand, a page of the table of contents of the document is analyzed, and the table of contents including the title information representing the character string of the title, the page information representing the number of pages, and the score obtained from the title information is generated. The document structure information including the information of the page, the start line, the end line, the score, and the character string based on the generated line information and searching the character string of the title of the table of contents information from the line information to obtain the line corresponding to the title. Generating document information. 13. The document structure information extraction according to claim 9, wherein line information of a document analyzed in the past is used as a learning material to acquire information to be included in the document structure information. Method. 14. The document structure information extracting method according to claim 13, wherein a structure name of document structure information is obtained using line information of the document analyzed in the past.